maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[mlir][NFC] Update textual references of `func` to `func.func` in Tensor/Tosa/Vector tests 

The special case parsing of `func` operations is being removed.
This commit is contained in:
River Riddle 2022-04-20 16:19:48 -07:00
parent 2c7836ef15
commit c48e3a13f3
45 changed files with 1013 additions and 1013 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
mlir/test/Dialect/Tensor/bufferize.mlir
View File

@ -13,7 +13,7 @@
// CHECK: %[[MEMREF:.*]] = bufferization.to_memref %[[TENSOR]] : memref<f32>
// CHECK: %[[EXTENT:.*]] = memref.dim %[[MEMREF]], %[[INDEX]] : memref<f32>
// CHECK: return %[[EXTENT]] : index
func @dim(%arg0: tensor<f32>, %arg1: index) -> index {
func.func @dim(%arg0: tensor<f32>, %arg1: index) -> index {
%0 = tensor.dim %arg0, %arg1 : tensor<f32>
return %0 : index
}
@ -22,7 +22,7 @@ func @dim(%arg0: tensor<f32>, %arg1: index) -> index {
// CHECK-SAME: %[[TENSOR:.*]]: tensor<*xf32>) -> index {
// CHECK: %[[MEMREF:.*]] = bufferization.to_memref %[[TENSOR]]
// CHECK: %[[EXTENT:.*]] = memref.rank %[[MEMREF]] : memref<*xf32>
func @rank(%arg0: tensor<*xf32>) -> index {
func.func @rank(%arg0: tensor<*xf32>) -> index {
%0 = tensor.rank %arg0 : tensor<*xf32>
return %0 : index
}
@ -33,7 +33,7 @@ func @rank(%arg0: tensor<*xf32>) -> index {
// CHECK: %[[CASTED:.*]] = memref.cast %[[MEMREF]] : memref<?xindex> to memref<2xindex>
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[CASTED]]
// CHECK: return %[[RET]] : tensor<2xindex>
func @tensor.cast(%arg0: tensor<?xindex>) -> tensor<2xindex> {
func.func @tensor.cast(%arg0: tensor<?xindex>) -> tensor<2xindex> {
%0 = tensor.cast %arg0 : tensor<?xindex> to tensor<2xindex>
return %0 : tensor<2xindex>
}
@ -44,7 +44,7 @@ func @tensor.cast(%arg0: tensor<?xindex>) -> tensor<2xindex> {
// CHECK: %[[CASTED_MEMREF:.*]] = memref.cast %[[MEMREF]] : memref<*xf32> to memref<2xf32>
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[CASTED_MEMREF]] : memref<2xf32>
// CHECK: return %[[RET]] : tensor<2xf32>
func @tensor.cast_from_unranked(%arg0: tensor<*xf32>) -> tensor<2xf32> {
func.func @tensor.cast_from_unranked(%arg0: tensor<*xf32>) -> tensor<2xf32> {
%0 = tensor.cast %arg0 : tensor<*xf32> to tensor<2xf32>
return %0 : tensor<2xf32>
}
@ -55,7 +55,7 @@ func @tensor.cast_from_unranked(%arg0: tensor<*xf32>) -> tensor<2xf32> {
// CHECK: %[[CASTED_MEMREF:.*]] = memref.cast %[[MEMREF]] : memref<2xf32> to memref<*xf32>
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[CASTED_MEMREF]] : memref<*xf32>
// CHECK: return %[[RET]] : tensor<*xf32>
func @tensor.cast_to_unranked(%arg0: tensor<2xf32>) -> tensor<*xf32> {
func.func @tensor.cast_to_unranked(%arg0: tensor<2xf32>) -> tensor<*xf32> {
%0 = tensor.cast %arg0 : tensor<2xf32> to tensor<*xf32>
return %0 : tensor<*xf32>
}
@ -67,7 +67,7 @@ func @tensor.cast_to_unranked(%arg0: tensor<2xf32>) -> tensor<*xf32> {
// CHECK: %[[RET:.*]] = memref.load %[[MEMREF]][%[[IDX]]] : memref<?xf32>
// CHECK: return %[[RET]] : f32
// CHECK: }
func @tensor.extract(%arg0: tensor<?xf32>, %arg1: index) -> f32 {
func.func @tensor.extract(%arg0: tensor<?xf32>, %arg1: index) -> f32 {
%0 = tensor.extract %arg0[%arg1] : tensor<?xf32>
return %0 : f32
}
@ -75,7 +75,7 @@ func @tensor.extract(%arg0: tensor<?xf32>, %arg1: index) -> f32 {
// CHECK-LABEL: func @tensor.from_elements_no_elements() -> tensor<0xindex> {
// CHECK: %[[RET:.*]] = arith.constant dense<> : tensor<0xindex>
// CHECK: return %[[RET]] : tensor<0xindex>
func @tensor.from_elements_no_elements() -> tensor<0xindex> {
func.func @tensor.from_elements_no_elements() -> tensor<0xindex> {
%0 = tensor.from_elements : tensor<0xindex>
return %0 : tensor<0xindex>
}
@ -86,7 +86,7 @@ func @tensor.from_elements_no_elements() -> tensor<0xindex> {
// CHECK: store %[[ELEM0]], %[[MEMREF]]
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[MEMREF]]
// CHECK: return %[[RET]] : tensor<index>
func @tensor.from_elements_0d(%arg0: index) -> tensor<index> {
func.func @tensor.from_elements_0d(%arg0: index) -> tensor<index> {
%0 = tensor.from_elements %arg0 : tensor<index>
return %0 : tensor<index>
}
@ -101,7 +101,7 @@ func @tensor.from_elements_0d(%arg0: index) -> tensor<index> {
// CHECK: store %[[ELEM1]], %[[MEMREF]][%[[C1]]]
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[MEMREF]]
// CHECK: return %[[RET]] : tensor<2xindex>
func @tensor.from_elements_1d(%arg0: index, %arg1: index) -> tensor<2xindex> {
func.func @tensor.from_elements_1d(%arg0: index, %arg1: index) -> tensor<2xindex> {
%0 = tensor.from_elements %arg0, %arg1 : tensor<2xindex>
return %0 : tensor<2xindex>
}
@ -121,7 +121,7 @@ func @tensor.from_elements_1d(%arg0: index, %arg1: index) -> tensor<2xindex> {
// CHECK: store %[[ELEM1]], %[[MEMREF]][%[[C2]], %[[C1]]]
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[MEMREF]]
// CHECK: return %[[RET]] : tensor<3x2xindex>
func @tensor.from_elements_2d(%arg0: index, %arg1: index) -> tensor<3x2xindex> {
func.func @tensor.from_elements_2d(%arg0: index, %arg1: index) -> tensor<3x2xindex> {
%0 = tensor.from_elements %arg0, %arg1, %arg0, %arg1, %arg0, %arg1
: tensor<3x2xindex>
return %0 : tensor<3x2xindex>
@ -163,7 +163,7 @@ func @tensor.from_elements_2d(%arg0: index, %arg1: index) -> tensor<3x2xindex> {
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[MEMREF]]
// CHECK: return %[[RET]] : tensor<3x2x2xf32>
func @tensor.from_elements_3d(%f0 : f32) -> tensor<3x2x2xf32> {
func.func @tensor.from_elements_3d(%f0 : f32) -> tensor<3x2x2xf32> {
%f1 = arith.constant 1.0 : f32
%f2 = arith.constant 2.0 : f32
%f3 = arith.constant 3.0 : f32
@ -195,7 +195,7 @@ func @tensor.from_elements_3d(%f0 : f32) -> tensor<3x2x2xf32> {
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[MEMREF]] : memref<?xindex>
// CHECK: return %[[RET]] : tensor<?xindex>
// CHECK: }
func @tensor.generate(%arg: tensor<*xf32>, %dynamic_extent: index) -> tensor<?xindex> {
func.func @tensor.generate(%arg: tensor<*xf32>, %dynamic_extent: index) -> tensor<?xindex> {
%result = tensor.generate %dynamic_extent {
^bb0(%i : index):
%elem = tensor.dim %arg, %i : tensor<*xf32>
@ -221,7 +221,7 @@ func @tensor.generate(%arg: tensor<*xf32>, %dynamic_extent: index) -> tensor<?xi
// CHECK: %[[RET:.*]] = bufferization.to_tensor %[[MEMREF]] : memref<16x?xindex>
// CHECK: return %[[RET]] : tensor<16x?xindex>
// CHECK: }
func @tensor.generate_static_and_dynamic(%arg0: index) -> tensor<16x?xindex> {
func.func @tensor.generate_static_and_dynamic(%arg0: index) -> tensor<16x?xindex> {
%result = tensor.generate %arg0 {
^bb0(%i: index, %j: index):
%sum = arith.addi %i, %j : index
@ -231,7 +231,7 @@ func @tensor.generate_static_and_dynamic(%arg0: index) -> tensor<16x?xindex> {
}
// CHECK-LABEL: func @tensor.generate_unknown_ops_in_body
func @tensor.generate_unknown_ops_in_body(%arg0: index) -> tensor<?xindex> {
func.func @tensor.generate_unknown_ops_in_body(%arg0: index) -> tensor<?xindex> {
// CHECK-NOT: tensor.generate
%tensor = tensor.generate %arg0 {
^bb0(%iv: index):
@ -244,7 +244,7 @@ func @tensor.generate_unknown_ops_in_body(%arg0: index) -> tensor<?xindex> {
// CHECK-LABEL: func @tensor.extract_slice(
// CHECK-SAME: %[[t1:.*]]: tensor<?x?xf32>, %[[idx1:.*]]: index, %[[idx2:.*]]: index
func @tensor.extract_slice(
func.func @tensor.extract_slice(
%t1: tensor<?x?xf32>, %idx1: index, %idx2: index) -> tensor<?x10xf32> {
// CHECK: %[[m:.*]] = bufferization.to_memref %[[t1]] : memref<?x?xf32>
// CHECK: %[[r:.*]] = memref.subview %[[m]][5, %[[idx2]]] [%[[idx1]], 10] [1, 1] : memref<?x?xf32> to memref<?x10xf32, #[[$MAP0]]>
@ -258,7 +258,7 @@ func @tensor.extract_slice(
// CHECK-LABEL: func @tensor.extract_slice_rank_reducing(
// CHECK-SAME: %[[t1:.*]]: tensor<?x10x?xf32>, %[[idx1:.*]]: index,
// CHECK-SAME: %[[idx2:.*]]: index
func @tensor.extract_slice_rank_reducing(
func.func @tensor.extract_slice_rank_reducing(
%t1: tensor<?x10x?xf32>, %idx1: index, %idx2: index) -> tensor<?x15xf32> {
// CHECK: %[[m1:.*]] = bufferization.to_memref %[[t1]] : memref<?x10x?xf32>
// CHECK: %[[r:.*]] = memref.subview %[[m1]][5, %[[idx1]], 10] [%[[idx2]], 1, 15] [1, 1, 1] : memref<?x10x?xf32> to memref<?x15xf32, #[[$MAP0]]>
@ -272,7 +272,7 @@ func @tensor.extract_slice_rank_reducing(
// CHECK-LABEL: func @tensor.insert_slice(
// CHECK-SAME: %[[t1:.*]]: tensor<?x?xf32>, %[[t2:.*]]: tensor<?x10xf32>,
// CHECK-SAME: %[[idx1:.*]]: index, %[[idx2:.*]]: index
func @tensor.insert_slice(%t1: tensor<?x?xf32>, %t2: tensor<?x10xf32>,
func.func @tensor.insert_slice(%t1: tensor<?x?xf32>, %t2: tensor<?x10xf32>,
%idx1: index, %idx2: index) -> tensor<?x?xf32> {
// CHECK-DAG: %[[c0:.*]] = arith.constant 0 : index
// CHECK-DAG: %[[c1:.*]] = arith.constant 1 : index
@ -295,7 +295,7 @@ func @tensor.insert_slice(%t1: tensor<?x?xf32>, %t2: tensor<?x10xf32>,
// CHECK-LABEL: func @tensor.insert(
// CHECK-SAME: %[[t1:.*]]: tensor<5xf32>, %[[idx1:.*]]: index,
// CHECK-SAME: %[[f:.*]]: f32
func @tensor.insert(%t1: tensor<5xf32>, %idx1: index, %f: f32) -> tensor<5xf32> {
func.func @tensor.insert(%t1: tensor<5xf32>, %idx1: index, %f: f32) -> tensor<5xf32> {
// CHECK-DAG: %[[alloc:.*]] = memref.alloc() {{.*}} : memref<5xf32>
// CHECK-DAG: %[[m1:.*]] = bufferization.to_memref %[[t1]] : memref<5xf32>
// CHECK: memref.copy %[[m1]], %[[alloc]]
@ -309,7 +309,7 @@ func @tensor.insert(%t1: tensor<5xf32>, %idx1: index, %f: f32) -> tensor<5xf32>
// CHECK-LABEL: func @tensor.expand_shape(
// CHECK-SAME: %[[t1:.*]]: tensor<?x10xf32>
func @tensor.expand_shape(%t1: tensor<?x10xf32>) -> tensor<2x?x10xf32> {
func.func @tensor.expand_shape(%t1: tensor<?x10xf32>) -> tensor<2x?x10xf32> {
// CHECK: %[[m1:.*]] = bufferization.to_memref %[[t1]] : memref<?x10xf32>
// CHECK: %[[expanded:.*]] = memref.expand_shape %[[m1]] [
// CHECK-SAME: [0, 1], [2]] : memref<?x10xf32> into memref<2x?x10xf32>
@ -323,7 +323,7 @@ func @tensor.expand_shape(%t1: tensor<?x10xf32>) -> tensor<2x?x10xf32> {
// CHECK-LABEL: func @tensor.expand_shape_of_slice(
// CHECK-SAME: %[[t1:.*]]: tensor<?x20xf32>
func @tensor.expand_shape_of_slice(
func.func @tensor.expand_shape_of_slice(
%t1: tensor<?x20xf32>, %o1: index, %s1: index) -> tensor<?x7x2x5xf32> {
// CHECK: %[[m1:.*]] = bufferization.to_memref %[[t1]] : memref<?x20xf32>
// CHECK: %[[subview:.*]] = memref.subview %[[m1]][%{{.*}}, 5] [%{{.*}}, 10] [1, 1] : memref<?x20xf32> to memref<?x10xf32, #[[$MAP1]]>
@ -340,7 +340,7 @@ func @tensor.expand_shape_of_slice(
// CHECK-LABEL: func @tensor.expand_shape_of_slice2(
// CHECK-SAME: %[[t1:.*]]: tensor<1x2xf32>
func @tensor.expand_shape_of_slice2(%t1: tensor<1x2xf32>) -> tensor<1xf32> {
func.func @tensor.expand_shape_of_slice2(%t1: tensor<1x2xf32>) -> tensor<1xf32> {
// CHECK: memref.subview {{.*}} : memref<1x2xf32> to memref<1x1xf32, #[[$MAP5]]>
%0 = tensor.extract_slice %t1[0, 0][1, 1][1, 1] : tensor<1x2xf32> to tensor<1x1xf32>
// CHECK: memref.collapse_shape %{{.*}} [
@ -351,7 +351,7 @@ func @tensor.expand_shape_of_slice2(%t1: tensor<1x2xf32>) -> tensor<1xf32> {
// CHECK-LABEL: func @tensor.collapse_shape(
// CHECK-SAME: %[[t1:.*]]: tensor<2x?x?xf32>
func @tensor.collapse_shape(%t1: tensor<2x?x?xf32>) -> tensor<?x?xf32> {
func.func @tensor.collapse_shape(%t1: tensor<2x?x?xf32>) -> tensor<?x?xf32> {
// CHECK: %[[m1:.*]] = bufferization.to_memref %[[t1]] : memref<2x?x?xf32>
// CHECK: %[[collapsed:.*]] = memref.collapse_shape %[[m1]] [
// CHECK-SAME: [0, 1], [2]] : memref<2x?x?xf32> into memref<?x?xf32>
@ -365,7 +365,7 @@ func @tensor.collapse_shape(%t1: tensor<2x?x?xf32>) -> tensor<?x?xf32> {
// CHECK-LABEL: func @tensor.collapse_shape_to_scalar(
// CHECK-SAME: %[[t1:.*]]: tensor<1x1x1xf32>
func @tensor.collapse_shape_to_scalar(%t1: tensor<1x1x1xf32>) -> tensor<f32> {
func.func @tensor.collapse_shape_to_scalar(%t1: tensor<1x1x1xf32>) -> tensor<f32> {
// CHECK: %[[m1:.*]] = bufferization.to_memref %[[t1]] : memref<1x1x1xf32>
// CHECK: %[[collapsed:.*]] = memref.collapse_shape %[[m1]] [] : memref<1x1x1xf32> into memref<f32>
%0 = tensor.collapse_shape %t1 []
@ -377,7 +377,7 @@ func @tensor.collapse_shape_to_scalar(%t1: tensor<1x1x1xf32>) -> tensor<f32> {
}
// CHECK-LABEL: func @tensor.collapse_shape_of_slice(
func @tensor.collapse_shape_of_slice(%arg0: tensor<2xi32>) -> tensor<i32> {
func.func @tensor.collapse_shape_of_slice(%arg0: tensor<2xi32>) -> tensor<i32> {
// CHECK: memref.subview %{{.*}}[1] [1] [1] : memref<2xi32> to memref<1xi32, #[[$MAP3]]>
%0 = tensor.extract_slice %arg0[1] [1] [1] : tensor<2xi32> to tensor<1xi32>
// CHECK: memref.collapse_shape %{{.*}} [] : memref<1xi32, #[[$MAP3]]> into memref<i32, #[[$MAP4]]>
@ -386,7 +386,7 @@ func @tensor.collapse_shape_of_slice(%arg0: tensor<2xi32>) -> tensor<i32> {
}
// CHECK-LABEL: func @tensor.collapse_shape_of_slice2(
func @tensor.collapse_shape_of_slice2(
func.func @tensor.collapse_shape_of_slice2(
%arg0: tensor<?x?x?x?xi64>, %o1: index, %o2: index, %o3: index, %o4: index)
-> tensor<87x63648xi64> {
// CHECK: %[[subview:.*]] = memref.subview %{{.*}} : memref<?x?x?x?xi64> to memref<87x78x68x12xi64, #{{.*}}>

164
mlir/test/Dialect/Tensor/canonicalize.mlir
View File

@ -2,7 +2,7 @@
// Checks that NOP casts are removed.
// CHECK-LABEL: cast_values
func @cast_values(%arg0: tensor<*xi32>) -> tensor<2xi32> {
func.func @cast_values(%arg0: tensor<*xi32>) -> tensor<2xi32> {
// NOP cast
%0 = tensor.cast %arg0 : tensor<*xi32> to tensor<*xi32>
// CHECK-NEXT: %[[RET:.*]] = tensor.cast %arg0 : tensor<*xi32> to tensor<2xi32>
@ -17,7 +17,7 @@ func @cast_values(%arg0: tensor<*xi32>) -> tensor<2xi32> {
// CHECK-LABEL: @tensor.cast_chain_ok
// CHECK-SAME: %[[IN:.*]]: tensor<*xi32>
func @tensor.cast_chain_ok(%input: tensor<*xi32>) -> tensor<4x8xi32> {
func.func @tensor.cast_chain_ok(%input: tensor<*xi32>) -> tensor<4x8xi32> {
// CHECK-NEXT: %[[RES:.*]] = tensor.cast %[[IN]] : tensor<*xi32> to tensor<4x8xi32>
%0 = tensor.cast %input : tensor<*xi32> to tensor<4x?xi32>
%1 = tensor.cast %0 : tensor<4x?xi32> to tensor<4x8xi32>
@ -29,7 +29,7 @@ func @tensor.cast_chain_ok(%input: tensor<*xi32>) -> tensor<4x8xi32> {
// CHECK-LABEL: @tensor.cast_chain_regain
// CHECK-SAME: %[[IN:.*]]: tensor<4xi32>
func @tensor.cast_chain_regain(%input: tensor<4xi32>) -> tensor<4xi32> {
func.func @tensor.cast_chain_regain(%input: tensor<4xi32>) -> tensor<4xi32> {
%0 = tensor.cast %input : tensor<4xi32> to tensor<?xi32>
%1 = tensor.cast %0 : tensor<?xi32> to tensor<4xi32>
// CHECK-NEXT: return %[[IN]]
@ -40,7 +40,7 @@ func @tensor.cast_chain_regain(%input: tensor<4xi32>) -> tensor<4xi32> {
// CHECK-LABEL: @tensor.cast_chain_keep
// CHECK-SAME: %[[IN:.*]]: tensor<?x?xi32>
func @tensor.cast_chain_keep(%input: tensor<?x?xi32>) -> tensor<?x8xi32> {
func.func @tensor.cast_chain_keep(%input: tensor<?x?xi32>) -> tensor<?x8xi32> {
// CHECK-NEXT: %[[C1:.*]] = tensor.cast %[[IN]]
%0 = tensor.cast %input : tensor<?x?xi32> to tensor<4x?xi32>
// CHECK-NEXT: %[[C2:.*]] = tensor.cast %[[C1]]
@ -53,7 +53,7 @@ func @tensor.cast_chain_keep(%input: tensor<?x?xi32>) -> tensor<?x8xi32> {
// CHECK-LABEL: @tensor.cast_chain_invalid
// CHECK-SAME: %[[IN:.*]]: tensor<4x8xi32>
func @tensor.cast_chain_invalid(%input: tensor<4x8xi32>) -> tensor<8x4xi32> {
func.func @tensor.cast_chain_invalid(%input: tensor<4x8xi32>) -> tensor<8x4xi32> {
// CHECK-NEXT: %[[C1:.*]] = tensor.cast %[[IN]]
%0 = tensor.cast %input : tensor<4x8xi32> to tensor<?x?xi32>
// CHECK-NEXT: %[[C2:.*]] = tensor.cast %[[C1]]
@ -65,7 +65,7 @@ func @tensor.cast_chain_invalid(%input: tensor<4x8xi32>) -> tensor<8x4xi32> {
// -----
// CHECK-LABEL: func @fold_extract
func @fold_extract(%arg0 : index) -> (f32, f16, f16, i32, complex<f32>) {
func.func @fold_extract(%arg0 : index) -> (f32, f16, f16, i32, complex<f32>) {
%const_0 = arith.constant 0 : index
%const_1 = arith.constant 1 : index
%const_3 = arith.constant 3 : index
@ -102,7 +102,7 @@ func @fold_extract(%arg0 : index) -> (f32, f16, f16, i32, complex<f32>) {
// -----
// CHECK-LABEL: func @fold_insert
func @fold_insert(%arg0 : index) -> (tensor<4xf32>) {
func.func @fold_insert(%arg0 : index) -> (tensor<4xf32>) {
// Fold an insert into a splat.
// CHECK-DAG: %[[C4:.+]] = arith.constant dense<4.{{0*}}e+00> : tensor<4xf32>
%0 = arith.constant dense<4.0> : tensor<4xf32>
@ -116,7 +116,7 @@ func @fold_insert(%arg0 : index) -> (tensor<4xf32>) {
// CHECK-LABEL: func @extract_from_tensor.cast
// CHECK-SAME: %[[TENSOR:.*]]: tensor<*xf32>
func @extract_from_tensor.cast(%tensor: tensor<*xf32>) -> f32 {
func.func @extract_from_tensor.cast(%tensor: tensor<*xf32>) -> f32 {
// CHECK-NEXT: %[[C0:.*]] = arith.constant 0 : index
%c0 = arith.constant 0 : index
// CHECK-NOT: tensor.cast
@ -129,7 +129,7 @@ func @extract_from_tensor.cast(%tensor: tensor<*xf32>) -> f32 {
// -----
// CHECK-LABEL: func @extract_from_tensor.from_elements
func @extract_from_tensor.from_elements(%element : index) -> index {
func.func @extract_from_tensor.from_elements(%element : index) -> index {
// CHECK-SAME: ([[ARG:%.*]]: index)
%c0 = arith.constant 0 : index
%tensor = tensor.from_elements %element : tensor<1xindex>
@ -141,7 +141,7 @@ func @extract_from_tensor.from_elements(%element : index) -> index {
// -----
// CHECK-LABEL: func @extract_from_tensor.from_elements_0d
func @extract_from_tensor.from_elements_0d(%element : index) -> index {
func.func @extract_from_tensor.from_elements_0d(%element : index) -> index {
// CHECK-SAME: ([[ARG:%.*]]: index)
%c0 = arith.constant 0 : index
%tensor = tensor.from_elements %element : tensor<index>
@ -153,7 +153,7 @@ func @extract_from_tensor.from_elements_0d(%element : index) -> index {
// -----
// CHECK-LABEL: func @extract_from_tensor.from_elements_3d
func @extract_from_tensor.from_elements_3d()
func.func @extract_from_tensor.from_elements_3d()
-> (f32, f32, f32, f32, f32, f32, f32, f32, f32, f32, f32, f32) {
%f0 = arith.constant 0.0 : f32
%f1 = arith.constant 1.0 : f32
@ -209,7 +209,7 @@ func @extract_from_tensor.from_elements_3d()
// Ensure the optimization doesn't segfault from bad constants
// CHECK-LABEL: func @extract_negative_from_tensor.from_elements
func @extract_negative_from_tensor.from_elements(%element : index) -> index {
func.func @extract_negative_from_tensor.from_elements(%element : index) -> index {
// CHECK-SAME: ([[ARG:%.*]]: index)
%c-1 = arith.constant -1 : index
%tensor = tensor.from_elements %element : tensor<1xindex>
@ -224,7 +224,7 @@ func @extract_negative_from_tensor.from_elements(%element : index) -> index {
// Ensure the optimization doesn't segfault from bad constants
// CHECK-LABEL: func @extract_oob_from_tensor.from_elements
func @extract_oob_from_tensor.from_elements(%element : index) -> index {
func.func @extract_oob_from_tensor.from_elements(%element : index) -> index {
// CHECK-SAME: ([[ARG:%.*]]: index)
%c1 = arith.constant 1 : index
%tensor = tensor.from_elements %element : tensor<1xindex>
@ -239,7 +239,7 @@ func @extract_oob_from_tensor.from_elements(%element : index) -> index {
// Ensure the optimization doesn't segfault from bad constants
// CHECK-LABEL: func @extract_oob_from_tensor.from_elements
func @extract_oob_from_tensor.from_elements(%element : index) -> index {
func.func @extract_oob_from_tensor.from_elements(%element : index) -> index {
// CHECK-SAME: ([[ARG:%.*]]: index)
%c2 = arith.constant 2 : index
%tensor = tensor.from_elements %element : tensor<1xindex>
@ -254,7 +254,7 @@ func @extract_oob_from_tensor.from_elements(%element : index) -> index {
// CHECK-LABEL: func @extract_from_tensor.generate
// CHECK-SAME: %[[IDX:.*]]: index, %[[TENSOR:.*]]: tensor<*xf32>
func @extract_from_tensor.generate(%idx: index, %tensor: tensor<*xf32>) -> index {
func.func @extract_from_tensor.generate(%idx: index, %tensor: tensor<*xf32>) -> index {
%size = tensor.rank %tensor : tensor<*xf32>
// CHECK-NEXT: %[[RES:.*]] = tensor.dim %[[TENSOR]], %[[IDX]]
%0 = tensor.generate %size {
@ -271,7 +271,7 @@ func @extract_from_tensor.generate(%idx: index, %tensor: tensor<*xf32>) -> index
// CHECK-LABEL: func @extract_from_tensor.generate_2d
// CHECK-SAME: %[[IDX0:.*]]: index, %[[IDX1:.*]]: index, %[[TENSOR:.*]]: tensor<*xf32>
func @extract_from_tensor.generate_2d(%idx0: index, %idx1: index, %tensor: tensor<*xf32>) -> index {
func.func @extract_from_tensor.generate_2d(%idx0: index, %idx1: index, %tensor: tensor<*xf32>) -> index {
%size = tensor.rank %tensor : tensor<*xf32>
// CHECK-NEXT: %[[DIM0:.*]] = tensor.dim %[[TENSOR]], %[[IDX0]]
// CHECK-NEXT: %[[DIM1:.*]] = tensor.dim %[[TENSOR]], %[[IDX1]]
@ -292,7 +292,7 @@ func @extract_from_tensor.generate_2d(%idx0: index, %idx1: index, %tensor: tenso
// CHECK-LABEL: func @extract_from_tensor.generate_sideeffects
// CHECK-SAME: %[[IDX:.*]]: index
func @extract_from_tensor.generate_sideeffects(%idx: index, %tensor: tensor<*xf32>, %mem: memref<?xindex>) -> index {
func.func @extract_from_tensor.generate_sideeffects(%idx: index, %tensor: tensor<*xf32>, %mem: memref<?xindex>) -> index {
%size = tensor.rank %tensor : tensor<*xf32>
// CHECK: %[[DTENSOR:.*]] = tensor.generate
%0 = tensor.generate %size {
@ -311,7 +311,7 @@ func @extract_from_tensor.generate_sideeffects(%idx: index, %tensor: tensor<*xf3
// CHECK-LABEL: @static_tensor.generate
// CHECK-SAME: %[[SIZE1:.*]]: index, %[[SIZE4:.*]]: index)
func @static_tensor.generate(%size1: index, %size4: index) -> tensor<3x?x?x7x?xindex> {
func.func @static_tensor.generate(%size1: index, %size4: index) -> tensor<3x?x?x7x?xindex> {
%c5 = arith.constant 5 : index
// CHECK: tensor.generate %[[SIZE1]], %[[SIZE4]]
%0 = tensor.generate %size1, %c5, %size4 {
@ -327,7 +327,7 @@ func @static_tensor.generate(%size1: index, %size4: index) -> tensor<3x?x?x7x?xi
// -----
// CHECK-LABEL: @from_elements.constant
func @from_elements.constant() -> tensor<3xindex> {
func.func @from_elements.constant() -> tensor<3xindex> {
// CHECK: %[[CST:.*]] = arith.constant dense<[1, 2, 1]> : tensor<3xindex>
// CHECK: return %[[CST]]
%c1 = arith.constant 1 : index
@ -338,7 +338,7 @@ func @from_elements.constant() -> tensor<3xindex> {
// -----
func @slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index) -> tensor<?x?x?xf32>
{
%c0 = arith.constant 0 : index
@ -357,7 +357,7 @@ func @slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
// -----
func @rank_reducing_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @rank_reducing_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index) -> tensor<?x?xf32>
{
%c0 = arith.constant 0 : index
@ -380,7 +380,7 @@ func @rank_reducing_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
// CHECK-SAME: %[[ARG0:.[a-z0-9A-Z_]+]]: tensor<4x6x16x32xi8>
// CHECK-NOT: tensor.extract_slice
// CHECK: return %[[ARG0]] : tensor<4x6x16x32xi8>
func @trivial_slice(%arg0 : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
func.func @trivial_slice(%arg0 : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
%0 = tensor.extract_slice %arg0[0, 0, 0, 0] [4, 6, 16, 32] [1, 1, 1, 1] : tensor<4x6x16x32xi8> to tensor<4x6x16x32xi8>
return %0 : tensor<4x6x16x32xi8>
}
@ -391,7 +391,7 @@ func @trivial_slice(%arg0 : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
// CHECK-SAME: %[[ARG0:.[a-z0-9A-Z_]+]]: tensor<4x6x16x32xi8>
// CHECK-NOT: tensor.extract_slice
// CHECK: return %[[ARG0]] : tensor<4x6x16x32xi8>
func @trivial_insert_slice(%arg0 : tensor<4x6x16x32xi8>, %arg1 : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
func.func @trivial_insert_slice(%arg0 : tensor<4x6x16x32xi8>, %arg1 : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
%0 = tensor.insert_slice %arg0 into %arg1[0, 0, 0, 0] [4, 6, 16, 32] [1, 1, 1, 1] : tensor<4x6x16x32xi8> into tensor<4x6x16x32xi8>
return %0 : tensor<4x6x16x32xi8>
}
@ -404,7 +404,7 @@ func @trivial_insert_slice(%arg0 : tensor<4x6x16x32xi8>, %arg1 : tensor<4x6x16x3
// Tensor cast is moved after slice and then gets canonicalized away.
// CHECK-NOT: tensor.cast
// CHECK: return %[[S]] : tensor<16x32xi8>
func @rank_reducing_tensor_of_cast(%arg : tensor<4x6x16x32xi8>) -> tensor<16x32xi8> {
func.func @rank_reducing_tensor_of_cast(%arg : tensor<4x6x16x32xi8>) -> tensor<16x32xi8> {
%0 = tensor.cast %arg : tensor<4x6x16x32xi8> to tensor<?x?x16x32xi8>
%1 = tensor.extract_slice %0[0, 1, 0, 0] [1, 1, 16, 32] [1, 1, 1, 1] : tensor<?x?x16x32xi8> to tensor<16x32xi8>
return %1 : tensor<16x32xi8>
@ -419,7 +419,7 @@ func @rank_reducing_tensor_of_cast(%arg : tensor<4x6x16x32xi8>) -> tensor<16x32x
// Tensor cast is folded away.
// CHECK-NOT: tensor.cast
// CHECK: return %[[S]] : tensor<4x6x16x32xi8>
func @rank_reducing_insert_slice_of_cast(%a : tensor<16x32xi8>, %b : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
func.func @rank_reducing_insert_slice_of_cast(%a : tensor<16x32xi8>, %b : tensor<4x6x16x32xi8>) -> tensor<4x6x16x32xi8> {
%c0 = arith.constant 0: index
%cast = tensor.cast %a : tensor<16x32xi8> to tensor<?x32xi8>
%sz = tensor.dim %cast, %c0: tensor<?x32xi8>
@ -429,7 +429,7 @@ func @rank_reducing_insert_slice_of_cast(%a : tensor<16x32xi8>, %b : tensor<4x6x
// -----
func @insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index, %arg3 : tensor<?x?x?xf32>) -> tensor<?x?x?xf32>
{
%c0 = arith.constant 0 : index
@ -448,7 +448,7 @@ func @insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
// -----
func @slice_to_insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @slice_to_insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index, %arg3 : tensor<?x?x?xf32>) -> tensor<?x?x?xf32>
{
%c0 = arith.constant 0 : index
@ -471,7 +471,7 @@ func @slice_to_insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : inde
// -----
func @rank_reducing_insert_slice_canonicalize(%arg0 : tensor<?x?xf32>, %arg1 : index,
func.func @rank_reducing_insert_slice_canonicalize(%arg0 : tensor<?x?xf32>, %arg1 : index,
%arg2 : index, %arg3 : tensor<?x?x?xf32>) -> tensor<?x?x?xf32>
{
%c0 = arith.constant 0 : index
@ -490,7 +490,7 @@ func @rank_reducing_insert_slice_canonicalize(%arg0 : tensor<?x?xf32>, %arg1 : i
// -----
func @rank_reducing_slice_to_insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @rank_reducing_slice_to_insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index, %arg3 : tensor<?x?x?xf32>) -> tensor<?x?x?xf32>
{
%c0 = arith.constant 0 : index
@ -513,7 +513,7 @@ func @rank_reducing_slice_to_insert_slice_canonicalize(%arg0 : tensor<?x?x?xf32>
// -----
func @insert_slice_propagate_dest_cast(%arg0 : tensor<2x?xi32>, %arg1 : tensor<i32>,
func.func @insert_slice_propagate_dest_cast(%arg0 : tensor<2x?xi32>, %arg1 : tensor<i32>,
%arg2 : index, %arg3 : index) -> tensor<?x?xi32> {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -536,7 +536,7 @@ func @insert_slice_propagate_dest_cast(%arg0 : tensor<2x?xi32>, %arg1 : tensor<i
// -----
func @insert_slice_output_dest_canonicalize(%arg0 : tensor<2x3xi32>, %arg1 : tensor<i32>) -> tensor<3x9xi32> {
func.func @insert_slice_output_dest_canonicalize(%arg0 : tensor<2x3xi32>, %arg1 : tensor<i32>) -> tensor<3x9xi32> {
%c9 = arith.constant 9 : index
%c3 = arith.constant 3 : index
%2 = tensor.extract %arg1[] : tensor<i32>
@ -563,7 +563,7 @@ func @insert_slice_output_dest_canonicalize(%arg0 : tensor<2x3xi32>, %arg1 : ten
// CHECK-SAME: %[[IDX0:[0-9a-z]+]]: index, %[[IDX1:[0-9a-z]+]]: index
// CHECK-NOT: tensor.dim
// CHECK: return %[[IDX1]] : index
func @dim_of_tensor.generate(%arg0: index, %arg1: index) -> index {
func.func @dim_of_tensor.generate(%arg0: index, %arg1: index) -> index {
%c3 = arith.constant 3 : index
%0 = tensor.generate %arg0, %arg1 {
^bb0(%arg2: index, %arg3: index, %arg4: index, %arg5: index, %arg6: index):
@ -582,7 +582,7 @@ func @dim_of_tensor.generate(%arg0: index, %arg1: index) -> index {
// CHECK-DAG: %[[C4:.+]] = arith.constant 4 : index
// CHECK: %[[T0:.+]] = tensor.dim %[[ARG0]], %[[C1]]
// CHECK-NEXT: return %[[C4]], %[[T0]]
func @fold_dim_of_tensor.cast(%arg0 : tensor<4x?xf32>) -> (index, index) {
func.func @fold_dim_of_tensor.cast(%arg0 : tensor<4x?xf32>) -> (index, index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
%0 = tensor.cast %arg0 : tensor<4x?xf32> to tensor<?x?xf32>
@ -598,7 +598,7 @@ func @fold_dim_of_tensor.cast(%arg0 : tensor<4x?xf32>) -> (index, index) {
// CHECK: %[[cast:.*]] = tensor.cast %[[arg0]] : tensor<?x5x?xf32> to tensor<64x5x64xf32>
// CHECK: %[[r:.*]] = tensor.insert_slice %[[cast]] into %[[arg1]][0, 1, 2] [64, 5, 64] [1, 1, 1] : tensor<64x5x64xf32> into tensor<?x?x?xf32>
// CHECK: return %[[r]]
func @insert_tensor_cast_on_insert_slice_src(
func.func @insert_tensor_cast_on_insert_slice_src(
%arg0 : tensor<?x5x?xf32>, %arg1 : tensor<?x?x?xf32>, %sz0: index, %sz2: index) -> tensor<?x?x?xf32> {
%c64 = arith.constant 64: index
%r = tensor.insert_slice %arg0 into %arg1[0, 1, 2] [%c64, 5, %c64] [1, 1, 1]
@ -610,7 +610,7 @@ func @insert_tensor_cast_on_insert_slice_src(
// CHECK-LABEL: func @fold_extract_insert
// CHECK-SAME: %{{.+}}: tensor<?x?x?xf32>, %[[SLICE:.+]]: tensor<4x?x8xf32>
func @fold_extract_insert(%input : tensor<?x?x?xf32>, %slice: tensor<4x?x8xf32>, %i: index, %size: index) -> (tensor<4x?x8xf32>) {
func.func @fold_extract_insert(%input : tensor<?x?x?xf32>, %slice: tensor<4x?x8xf32>, %i: index, %size: index) -> (tensor<4x?x8xf32>) {
%c0 = arith.constant 0: index
%c1 = arith.constant 1: index
%0 = tensor.insert_slice %slice into %input[%c0, %i, 0] [4, %size, 8] [1, 1, %c1] : tensor<4x?x8xf32> into tensor<?x?x?xf32>
@ -624,7 +624,7 @@ func @fold_extract_insert(%input : tensor<?x?x?xf32>, %slice: tensor<4x?x8xf32>,
// CHECK-LABEL: func @fold_extract_constant_splat
// CHECK-NOT: tensor.extract_slice
// CHECK: arith.constant dense<42> : tensor<4x4xi32>
func @fold_extract_constant_splat() -> (tensor<4x4xi32>) {
func.func @fold_extract_constant_splat() -> (tensor<4x4xi32>) {
%cst = arith.constant dense<42> : tensor<1024x1024xi32>
%1 = tensor.extract_slice %cst[0,0] [4,4] [1, 1] : tensor<1024x1024xi32> to tensor<4x4xi32>
return %1 : tensor<4x4xi32>
@ -634,7 +634,7 @@ func @fold_extract_constant_splat() -> (tensor<4x4xi32>) {
// CHECK-LABEL: func @fold_overlapping_insert
// CHECK-SAME: %[[INPUT:.+]]: tensor<?x?x?xf32>, %{{.+}}: tensor<4x?x8xf32>, %[[SLICE2:.+]]: tensor<4x?x8xf32>
func @fold_overlapping_insert(%input : tensor<?x?x?xf32>, %slice1: tensor<4x?x8xf32>, %slice2: tensor<4x?x8xf32>, %i: index, %size: index) -> (tensor<?x?x?xf32>) {
func.func @fold_overlapping_insert(%input : tensor<?x?x?xf32>, %slice1: tensor<4x?x8xf32>, %slice2: tensor<4x?x8xf32>, %i: index, %size: index) -> (tensor<?x?x?xf32>) {
%c0 = arith.constant 0: index
%c1 = arith.constant 1: index
%0 = tensor.insert_slice %slice1 into %input[%c0, %i, 0] [4, %size, 8] [1, 1, %c1] : tensor<4x?x8xf32> into tensor<?x?x?xf32>
@ -646,7 +646,7 @@ func @fold_overlapping_insert(%input : tensor<?x?x?xf32>, %slice1: tensor<4x?x8x
// -----
func @compose_expand_of_expand(%arg0 : tensor<?x?xf32>)
func.func @compose_expand_of_expand(%arg0 : tensor<?x?xf32>)
-> tensor<?x6x4x?x5xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1], [2]]
: tensor<?x?xf32> into tensor<?x4x?xf32>
@ -660,7 +660,7 @@ func @compose_expand_of_expand(%arg0 : tensor<?x?xf32>)
// -----
func @compose_expand_of_expand_of_zero_dim(%arg0 : tensor<f32>)
func.func @compose_expand_of_expand_of_zero_dim(%arg0 : tensor<f32>)
-> tensor<1x1x1xf32> {
%0 = tensor.expand_shape %arg0 [] : tensor<f32> into tensor<1xf32>
%1 = tensor.expand_shape %0 [[0, 1, 2]]
@ -673,7 +673,7 @@ func @compose_expand_of_expand_of_zero_dim(%arg0 : tensor<f32>)
// -----
func @fold_collapse_of_expand(%arg0 : tensor<12x4xf32>) -> tensor<12x4xf32> {
func.func @fold_collapse_of_expand(%arg0 : tensor<12x4xf32>) -> tensor<12x4xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1], [2]]
: tensor<12x4xf32> into tensor<3x4x4xf32>
%1 = tensor.collapse_shape %0 [[0, 1], [2]]
@ -685,7 +685,7 @@ func @fold_collapse_of_expand(%arg0 : tensor<12x4xf32>) -> tensor<12x4xf32> {
// -----
func @fold_collapse_of_expand_dynamic(%arg0 : tensor<?x?xf32>)
func.func @fold_collapse_of_expand_dynamic(%arg0 : tensor<?x?xf32>)
-> tensor<?x?xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1], [2]]
: tensor<?x?xf32> into tensor<?x4x?xf32>
@ -698,7 +698,7 @@ func @fold_collapse_of_expand_dynamic(%arg0 : tensor<?x?xf32>)
// -----
func @compose_expand_of_collapse(%arg0 : tensor<2x3x4x5x6x7x8xf32>)
func.func @compose_expand_of_collapse(%arg0 : tensor<2x3x4x5x6x7x8xf32>)
-> tensor<24x5x42x8xf32> {
%0 = tensor.collapse_shape %arg0 [[0, 1, 2, 3, 4, 5, 6]]
: tensor<2x3x4x5x6x7x8xf32> into tensor<40320xf32>
@ -714,7 +714,7 @@ func @compose_expand_of_collapse(%arg0 : tensor<2x3x4x5x6x7x8xf32>)
// -----
func @compose_expand_of_collapse_7D(%arg0 : tensor<24x5x42x8xf32>)
func.func @compose_expand_of_collapse_7D(%arg0 : tensor<24x5x42x8xf32>)
-> tensor<2x3x4x5x6x7x8xf32> {
%0 = tensor.collapse_shape %arg0 [[0, 1, 2, 3]]
: tensor<24x5x42x8xf32> into tensor<40320xf32>
@ -730,7 +730,7 @@ func @compose_expand_of_collapse_7D(%arg0 : tensor<24x5x42x8xf32>)
// -----
func @compose_collapse_of_expand(%arg : tensor<?x?x?xi64>)
func.func @compose_collapse_of_expand(%arg : tensor<?x?x?xi64>)
-> tensor<?x?xi64> {
%0 = tensor.expand_shape %arg [[0], [1], [2, 3]]
: tensor<?x?x?xi64> into tensor<?x?x?x1xi64>
@ -746,7 +746,7 @@ func @compose_collapse_of_expand(%arg : tensor<?x?x?xi64>)
// -----
func @compose_collapse_of_expand_1D(%arg0 : tensor<2048xf32>)
func.func @compose_collapse_of_expand_1D(%arg0 : tensor<2048xf32>)
-> tensor<4x512xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1, 2, 3]]
: tensor<2048xf32> into tensor<1x4x1x512xf32>
@ -760,7 +760,7 @@ func @compose_collapse_of_expand_1D(%arg0 : tensor<2048xf32>)
// -----
func @compose_expand_of_collapse_0_rank_to_expand(%arg0 : tensor<1x1x1xf32>)
func.func @compose_expand_of_collapse_0_rank_to_expand(%arg0 : tensor<1x1x1xf32>)
-> tensor<1x1x1x1xf32> {
%0 = tensor.collapse_shape %arg0 []
: tensor<1x1x1xf32> into tensor<f32>
@ -776,7 +776,7 @@ func @compose_expand_of_collapse_0_rank_to_expand(%arg0 : tensor<1x1x1xf32>)
// -----
func @compose_expand_of_collapse_0_rank_to_collapse(%arg0 : tensor<1x1x1x1xf32>)
func.func @compose_expand_of_collapse_0_rank_to_collapse(%arg0 : tensor<1x1x1x1xf32>)
-> tensor<1x1x1xf32> {
%0 = tensor.collapse_shape %arg0 []
: tensor<1x1x1x1xf32> into tensor<f32>
@ -793,7 +793,7 @@ func @compose_expand_of_collapse_0_rank_to_collapse(%arg0 : tensor<1x1x1x1xf32>)
// -----
// CHECK-LABEL: func @zero_rank_reshape_multi
func @zero_rank_reshape_multi(%arg0: tensor<f32>) -> tensor<f32> {
func.func @zero_rank_reshape_multi(%arg0: tensor<f32>) -> tensor<f32> {
// CHECK: return %arg0
%0 = tensor.expand_shape %arg0 [] : tensor<f32> into tensor<1xf32>
%1 = tensor.expand_shape %0 [[0, 1]] : tensor<1xf32> into tensor<1x1xf32>
@ -803,7 +803,7 @@ func @zero_rank_reshape_multi(%arg0: tensor<f32>) -> tensor<f32> {
// -----
func @compose_collapse_of_collapse(%arg0 : tensor<?x?x?x?x?xf32>)
func.func @compose_collapse_of_collapse(%arg0 : tensor<?x?x?x?x?xf32>)
-> tensor<?x?xf32> {
%0 = tensor.collapse_shape %arg0 [[0, 1], [2], [3, 4]]
: tensor<?x?x?x?x?xf32> into tensor<?x?x?xf32>
@ -817,7 +817,7 @@ func @compose_collapse_of_collapse(%arg0 : tensor<?x?x?x?x?xf32>)
// -----
func @compose_collapse_of_collapse_zero_dim(%arg0 : tensor<1x1x1xf32>)
func.func @compose_collapse_of_collapse_zero_dim(%arg0 : tensor<1x1x1xf32>)
-> tensor<f32> {
%0 = tensor.collapse_shape %arg0 [[0, 1, 2]]
: tensor<1x1x1xf32> into tensor<1xf32>
@ -830,7 +830,7 @@ func @compose_collapse_of_collapse_zero_dim(%arg0 : tensor<1x1x1xf32>)
// -----
func @fold_collapse_of_expand_1D(%arg0 : tensor<4x512xf32>) -> tensor<2048xf32> {
func.func @fold_collapse_of_expand_1D(%arg0 : tensor<4x512xf32>) -> tensor<2048xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1, 2], [3]]
: tensor<4x512xf32> into tensor<1x4x1x512xf32>
%1 = tensor.collapse_shape %0 [[0, 1, 2, 3]]
@ -843,7 +843,7 @@ func @fold_collapse_of_expand_1D(%arg0 : tensor<4x512xf32>) -> tensor<2048xf32>
// -----
func @fold_collapse_of_expand_unit_dims(%arg0 : tensor<2048x1x1xf32>)
func.func @fold_collapse_of_expand_unit_dims(%arg0 : tensor<2048x1x1xf32>)
-> tensor<4x512x1x1xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1, 2, 3], [4], [5]]
: tensor<2048x1x1xf32> into tensor<1x4x1x512x1x1xf32>
@ -857,7 +857,7 @@ func @fold_collapse_of_expand_unit_dims(%arg0 : tensor<2048x1x1xf32>)
// -----
func @compose_collapse_of_expand_unit_dims(%arg0 : tensor<2048x1x2048xf32>)
func.func @compose_collapse_of_expand_unit_dims(%arg0 : tensor<2048x1x2048xf32>)
-> tensor<4x512x1x512x4xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1, 2, 3, 4], [5], [6, 7, 8]]
: tensor<2048x1x2048xf32> into tensor<1x4x1x512x1x1x512x1x4xf32>
@ -871,7 +871,7 @@ func @compose_collapse_of_expand_unit_dims(%arg0 : tensor<2048x1x2048xf32>)
// -----
func @compose_collapse_of_expand_trailing_unit_dims(%arg0: tensor<2xf32>)
func.func @compose_collapse_of_expand_trailing_unit_dims(%arg0: tensor<2xf32>)
-> tensor<2x1xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1, 2]]
: tensor<2xf32> into tensor<2x1x1xf32>
@ -885,7 +885,7 @@ func @compose_collapse_of_expand_trailing_unit_dims(%arg0: tensor<2xf32>)
// -----
func @compose_collapse_of_collapse_unit_dims_dynamic(
func.func @compose_collapse_of_collapse_unit_dims_dynamic(
%arg0 : tensor<?x1x?x1x1x?x?x1x1xf32>) -> tensor<?x?x?x?xf32> {
%0 = tensor.collapse_shape %arg0 [[0], [1, 2], [3], [4], [5], [6, 7, 8]]
: tensor<?x1x?x1x1x?x?x1x1xf32> into tensor<?x?x1x1x?x?xf32>
@ -900,7 +900,7 @@ func @compose_collapse_of_collapse_unit_dims_dynamic(
// -----
func @fold_collapse_of_expand_trailing_unit_dims(%arg0: tensor<2xf32>)
func.func @fold_collapse_of_expand_trailing_unit_dims(%arg0: tensor<2xf32>)
-> tensor<2x1xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1, 2]]
: tensor<2xf32> into tensor<2x1x1xf32>
@ -914,7 +914,7 @@ func @fold_collapse_of_expand_trailing_unit_dims(%arg0: tensor<2xf32>)
// -----
func @fold_collapse_of_collapse_trailing_unit_dims_dynamic(
func.func @fold_collapse_of_collapse_trailing_unit_dims_dynamic(
%arg0: tensor<1x1x?x1x1x1xf32>) -> tensor<?xf32> {
%0 = tensor.collapse_shape %arg0 [[0, 1, 2], [3], [4], [5]]
: tensor<1x1x?x1x1x1xf32> into tensor<?x1x1x1xf32>
@ -928,7 +928,7 @@ func @fold_collapse_of_collapse_trailing_unit_dims_dynamic(
// -----
func @fold_collapse_of_expand_trailing_unit_dims(%arg0: tensor<12x42x1x1xf32>)
func.func @fold_collapse_of_expand_trailing_unit_dims(%arg0: tensor<12x42x1x1xf32>)
-> tensor<12x42xf32> {
%0 = tensor.expand_shape %arg0 [[0], [1], [2], [3, 4]]
: tensor<12x42x1x1xf32> into tensor<12x42x1x1x1xf32>
@ -942,7 +942,7 @@ func @fold_collapse_of_expand_trailing_unit_dims(%arg0: tensor<12x42x1x1xf32>)
// -----
func @fold_collapse_of_expand_unit_dims_in_middle(%arg0 : tensor<?x?x?xf32>)
func.func @fold_collapse_of_expand_unit_dims_in_middle(%arg0 : tensor<?x?x?xf32>)
-> tensor<?x?xf32> {
%0 = tensor.expand_shape %arg0 [[0], [1], [2, 3]]
: tensor<?x?x?xf32> into tensor<?x?x1x?xf32>
@ -957,7 +957,7 @@ func @fold_collapse_of_expand_unit_dims_in_middle(%arg0 : tensor<?x?x?xf32>)
// -----
func @no_fold_collapse_of_expand_incompatible(%arg0 : tensor<4x6x8xf32>)
func.func @no_fold_collapse_of_expand_incompatible(%arg0 : tensor<4x6x8xf32>)
-> tensor<2x6x16xf32> {
%0 = tensor.expand_shape %arg0 [[0, 1], [2, 3], [4]]
: tensor<4x6x8xf32> into tensor<2x2x3x2x8xf32>
@ -971,7 +971,7 @@ func @no_fold_collapse_of_expand_incompatible(%arg0 : tensor<4x6x8xf32>)
// -----
func @no_fold_collapse_of_expand_empty_expr(%arg0: tensor<3x2x2xf32>)
func.func @no_fold_collapse_of_expand_empty_expr(%arg0: tensor<3x2x2xf32>)
-> tensor<12x1xf32> {
%0 = tensor.expand_shape %arg0 [[0], [1], [2, 3]]
: tensor<3x2x2xf32> into tensor<3x2x2x1xf32>
@ -989,7 +989,7 @@ func @no_fold_collapse_of_expand_empty_expr(%arg0: tensor<3x2x2xf32>)
// -----
func @reshape_splat_constant_int32() -> tensor<2x4x2xi32> {
func.func @reshape_splat_constant_int32() -> tensor<2x4x2xi32> {
%c0 = arith.constant dense<42> : tensor<2x8xi32>
%0 = tensor.expand_shape %c0 [[0], [1, 2]]
: tensor<2x8xi32> into tensor<2x4x2xi32>
@ -1002,7 +1002,7 @@ func @reshape_splat_constant_int32() -> tensor<2x4x2xi32> {
// -----
func @reshape_splat_constant_int16() -> tensor<2x4x2xi16> {
func.func @reshape_splat_constant_int16() -> tensor<2x4x2xi16> {
%c0 = arith.constant dense<42> : tensor<2x8xi16>
%0 = tensor.expand_shape %c0 [[0], [1, 2]]
: tensor<2x8xi16> into tensor<2x4x2xi16>
@ -1015,7 +1015,7 @@ func @reshape_splat_constant_int16() -> tensor<2x4x2xi16> {
// -----
func @reshape_splat_constant_float32() -> tensor<2x4x2xf32> {
func.func @reshape_splat_constant_float32() -> tensor<2x4x2xf32> {
%c0 = arith.constant dense<42.0> : tensor<2x8xf32>
%0 = tensor.expand_shape %c0 [[0], [1, 2]]
: tensor<2x8xf32> into tensor<2x4x2xf32>
@ -1028,7 +1028,7 @@ func @reshape_splat_constant_float32() -> tensor<2x4x2xf32> {
// -----
func @reshape_splat_constant_float64() -> tensor<2x4x2xf64> {
func.func @reshape_splat_constant_float64() -> tensor<2x4x2xf64> {
%c0 = arith.constant dense<42.0> : tensor<2x8xf64>
%0 = tensor.expand_shape %c0 [[0], [1, 2]]
: tensor<2x8xf64> into tensor<2x4x2xf64>
@ -1042,7 +1042,7 @@ func @reshape_splat_constant_float64() -> tensor<2x4x2xf64> {
// -----
// CHECK-LABEL: func @fold_rank
func @fold_rank() -> (index) {
func.func @fold_rank() -> (index) {
%const_0 = arith.constant dense<[[[1, -2, 1, 36]], [[0, 2, -1, 64]]]>
: tensor<2x1x4xi32>
@ -1060,7 +1060,7 @@ func @fold_rank() -> (index) {
// CHECK-SAME: %[[ARG0:.*]]: tensor<5x6xf32>
// CHECK-NOT: tensor.pad
// CHECK: return %[[ARG0]]
func @pad_same_static_shape(%arg0: tensor<5x6xf32>, %a: index)
func.func @pad_same_static_shape(%arg0: tensor<5x6xf32>, %a: index)
-> tensor<5x6xf32> {
%cst = arith.constant 0.000000e+00 : f32
%0 = tensor.pad %arg0 low[%a, 0] high[0, %a] {
@ -1076,7 +1076,7 @@ func @pad_same_static_shape(%arg0: tensor<5x6xf32>, %a: index)
// CHECK-SAME: %[[ARG0:.*]]: tensor<5x6xf32>
// CHECK: %[[PAD:.*]] = tensor.pad
// CHECK: return %[[PAD]]
func @pad_nofold_same_static_shape(%arg0: tensor<5x6xf32>, %a: index)
func.func @pad_nofold_same_static_shape(%arg0: tensor<5x6xf32>, %a: index)
-> tensor<5x6xf32> {
%cst = arith.constant 0.000000e+00 : f32
%0 = tensor.pad %arg0 nofold low[%a, 0] high[0, %a] {
@ -1100,7 +1100,7 @@ func @pad_nofold_same_static_shape(%arg0: tensor<5x6xf32>, %a: index)
// CHECK-SAME: tensor<?x64x?x?xf32> to tensor<?x?x?x?xf32>
// CHECK: return %[[DYNAMIC]] : tensor<?x?x?x?xf32>
// CHECK: }
func @pad_after_cast_different_shape(%arg0: tensor<?x64x?x?xf32>)
func.func @pad_after_cast_different_shape(%arg0: tensor<?x64x?x?xf32>)
-> tensor<?x?x?x?xf32> {
%cst = arith.constant 0.000000e+00 : f32
%dynamic = tensor.cast %arg0 : tensor<?x64x?x?xf32> to tensor<?x?x?x?xf32>
@ -1124,7 +1124,7 @@ func @pad_after_cast_different_shape(%arg0: tensor<?x64x?x?xf32>)
// CHECK: } : tensor<?x64x?x?xf32> to tensor<?x?x?x?xf32>
// CHECK: return %[[PADDED:.*]] : tensor<?x?x?x?xf32>
// CHECK: }
func @pad_after_cast_same_shape(%arg0: tensor<?x64x?x?xf32>, %padding : index)
func.func @pad_after_cast_same_shape(%arg0: tensor<?x64x?x?xf32>, %padding : index)
-> tensor<?x?x?x?xf32> {
%cst = arith.constant 0.000000e+00 : f32
%dynamic = tensor.cast %arg0 : tensor<?x64x?x?xf32> to tensor<?x?x?x?xf32>
@ -1141,7 +1141,7 @@ func @pad_after_cast_same_shape(%arg0: tensor<?x64x?x?xf32>, %padding : index)
// CHECK-NOT: tensor.cast
// CHECK: tensor.pad
// CHECK: tensor<8x?xf32> to tensor<8x32xf32>
func @pad_of_cast(%t: tensor<8x?xf32>, %s: index) -> tensor<8x32xf32> {
func.func @pad_of_cast(%t: tensor<8x?xf32>, %s: index) -> tensor<8x32xf32> {
%c0 = arith.constant 0 : index
%cst = arith.constant 0.000000e+00 : f32
%0 = tensor.cast %t : tensor<8x?xf32> to tensor<?x?xf32>
@ -1155,7 +1155,7 @@ func @pad_of_cast(%t: tensor<8x?xf32>, %s: index) -> tensor<8x32xf32> {
// -----
// CHECK-LABEL: @cast_of_pad_more_static
func @cast_of_pad_more_static(%arg0: tensor<?x?xf32>, %padding: index) -> tensor<32x32xf32> {
func.func @cast_of_pad_more_static(%arg0: tensor<?x?xf32>, %padding: index) -> tensor<32x32xf32> {
%cst = arith.constant 0.000000e+00 : f32
// CHECK: %[[PAD:.*]] = tensor.pad
// CHECK: tensor<?x?xf32> to tensor<32x32xf32>
@ -1172,7 +1172,7 @@ func @cast_of_pad_more_static(%arg0: tensor<?x?xf32>, %padding: index) -> tensor
// -----
// CHECK-LABEL: @cast_of_pad_less_static
func @cast_of_pad_less_static(%arg0: tensor<32x?x?xf32>, %padding: index) -> tensor<?x32x32xf32> {
func.func @cast_of_pad_less_static(%arg0: tensor<32x?x?xf32>, %padding: index) -> tensor<?x32x32xf32> {
%cst = arith.constant 0.000000e+00 : f32
// CHECK: tensor.pad
%padded = tensor.pad %arg0 low[%padding, %padding, %padding] high[0, 0, 0] {
@ -1187,7 +1187,7 @@ func @cast_of_pad_less_static(%arg0: tensor<32x?x?xf32>, %padding: index) -> ten
// -----
func @pad_cast_fold(%arg0: tensor<4x4xf32>) -> tensor<4x4xf32> {
func.func @pad_cast_fold(%arg0: tensor<4x4xf32>) -> tensor<4x4xf32> {
%c0 = arith.constant 0 : index
%cst = arith.constant 0.0 : f32
%0 = tensor.cast %arg0 : tensor<4x4xf32> to tensor<?x?xf32>
@ -1207,7 +1207,7 @@ func @pad_cast_fold(%arg0: tensor<4x4xf32>) -> tensor<4x4xf32> {
// CHECK-SAME: %[[ARG0:.*]]: tensor<4x?xf32>
// CHECK-NOT: tensor.cast
// CHECK: %[[RESULT:.*]] = tensor.pad %[[ARG0]]
func @fold_pad_source_cast(%arg0: tensor<4x?xf32>) -> tensor<4x4xf32> {
func.func @fold_pad_source_cast(%arg0: tensor<4x?xf32>) -> tensor<4x4xf32> {
%cst = arith.constant 0.0 : f32
%0 = tensor.cast %arg0 : tensor<4x?xf32> to tensor<?x?xf32>
%1 = tensor.pad %0 low[0, 0] high[0, 1] {
@ -1224,7 +1224,7 @@ func @fold_pad_source_cast(%arg0: tensor<4x?xf32>) -> tensor<4x4xf32> {
// CHECK-NOT: tensor.pad
// CHECK: %[[RESULT:.*]] = tensor.cast %[[ARG0]] : tensor<?x?x?xf32> to tensor<2x3x4xf32>
// CHECK: return %[[RESULT]]
func @pad_static_zero_cast(%arg0: tensor<?x?x?xf32>, %pad_value: f32) -> tensor<2x3x4xf32> {
func.func @pad_static_zero_cast(%arg0: tensor<?x?x?xf32>, %pad_value: f32) -> tensor<2x3x4xf32> {
%c0 = arith.constant 0 : index
%0 = tensor.pad %arg0 low[0, %c0, 0] high[0, 0, %c0] {
^bb0(%arg1: index, %arg2: index, %arg3: index):
@ -1240,7 +1240,7 @@ func @pad_static_zero_cast(%arg0: tensor<?x?x?xf32>, %pad_value: f32) -> tensor<
// CHECK-SAME: %[[ARG0:.*]]: tensor<?x?x?xf32>
// CHECK: %[[PAD:.*]] = tensor.pad
// CHECK: return %[[PAD]]
func @pad_nofold_static_zero(%arg0: tensor<?x?x?xf32>, %pad_value: f32) -> tensor<2x3x4xf32> {
func.func @pad_nofold_static_zero(%arg0: tensor<?x?x?xf32>, %pad_value: f32) -> tensor<2x3x4xf32> {
%c0 = arith.constant 0 : index
%0 = tensor.pad %arg0 nofold low[0, %c0, 0] high[0, 0, %c0] {
^bb0(%arg1: index, %arg2: index, %arg3: index):
@ -1342,7 +1342,7 @@ func.func @dont_fold_pad_chains(%arg0: tensor<64x64xf32>,
// -----
// CHECK-LABEL: func @fold_collapse_shape_from_elements
func @fold_collapse_shape_from_elements(%arg0: i32) -> tensor<i32> {
func.func @fold_collapse_shape_from_elements(%arg0: i32) -> tensor<i32> {
// CHECK: %[[FROM:.+]] = tensor.from_elements %arg0 : tensor<i32>
// CHECK: return %[[FROM]] : tensor<i32>
%0 = tensor.from_elements %arg0 : tensor<1xi32>
@ -1353,7 +1353,7 @@ func @fold_collapse_shape_from_elements(%arg0: i32) -> tensor<i32> {
// -----
// CHECK-LABEL: func @fold_expand_shape_from_elements
func @fold_expand_shape_from_elements(%arg0: i32) -> tensor<1xi32> {
func.func @fold_expand_shape_from_elements(%arg0: i32) -> tensor<1xi32> {
// CHECK: %[[FROM:.+]] = tensor.from_elements %arg0 : tensor<1xi32>
// CHECK: return %[[FROM]] : tensor<1xi32>
%0 = tensor.from_elements %arg0 : tensor<i32>
@ -1364,7 +1364,7 @@ func @fold_expand_shape_from_elements(%arg0: i32) -> tensor<1xi32> {
// -----
// CHECK-LABEL: func @propogate_index_cast
func @propogate_index_cast(%arg0: tensor<1xi32>) -> index {
func.func @propogate_index_cast(%arg0: tensor<1xi32>) -> index {
// CHECK: %[[IDX:.+]] = arith.constant 0
// CHECK: %[[EXT:.+]] = tensor.extract %arg0[%[[IDX]]] : tensor<1xi32>
// CHECK: %[[CAST:.+]] = arith.index_cast %[[EXT]]
@ -1378,7 +1378,7 @@ func @propogate_index_cast(%arg0: tensor<1xi32>) -> index {
// -----
// CHECK-LABEL: func @splat_fold
func @splat_fold() -> tensor<4xf32> {
func.func @splat_fold() -> tensor<4xf32> {
%c = arith.constant 1.0 : f32
%t = tensor.splat %c : tensor<4xf32>
return %t : tensor<4xf32>
@ -1392,7 +1392,7 @@ func @splat_fold() -> tensor<4xf32> {
// There was an issue in cast + insert_slice folding generating invalid ir.
// https://github.com/llvm/llvm-project/issues/53099
// CHECK-LABEL: func @insert_slice_cast
func @insert_slice_cast(%arg0 : tensor<1x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : index, %arg3 : index, %arg4 : index, %arg5 : index, %arg6 : index, %arg7 : index) -> tensor<?x?xf32> {
func.func @insert_slice_cast(%arg0 : tensor<1x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : index, %arg3 : index, %arg4 : index, %arg5 : index, %arg6 : index, %arg7 : index) -> tensor<?x?xf32> {
// CHECK: %[[CAST:.*]] = tensor.cast %{{.*}} : tensor<1x?xf32> to tensor<?x?xf32>
%0 = tensor.cast %arg0 : tensor<1x?xf32> to tensor<?x?xf32>
// CHECK: %[[RES:.*]] = tensor.insert_slice %[[CAST]]

6
mlir/test/Dialect/Tensor/fold-constant-extract-slice.mlir
View File

@ -4,7 +4,7 @@
// CHECK-NOT: tensor.extract_slice
// CHECK: %[[CONST:.+]] = arith.constant dense<1.000000e+01> : tensor<1x1xf32>
// CHECK: return %[[CONST]] : tensor<1x1xf32>
func @slice_constant(%arg0 : tensor<2x1xf32>) -> tensor<1x1xf32>
func.func @slice_constant(%arg0 : tensor<2x1xf32>) -> tensor<1x1xf32>
{
%cst = arith.constant dense<[[10.0], [11.0]]> : tensor<2x1xf32>
%slice = tensor.extract_slice %cst[0, 0] [1, 1] [1, 1] : tensor<2x1xf32> to tensor<1x1xf32>
@ -17,7 +17,7 @@ func @slice_constant(%arg0 : tensor<2x1xf32>) -> tensor<1x1xf32>
// CHECK-NOT: tensor.extract_slice
// CHECK: %[[CONST:.+]] = arith.constant dense<{{\[}}[1.000000e+01, 9.000000e+00], [1.100000e+01, 1.200000e+01]]> : tensor<2x2xf32>
// CHECK: return %[[CONST]] : tensor<2x2xf32>
func @slice_constant_3x4(%arg0 : tensor<3x4xf32>) -> tensor<2x2xf32>
func.func @slice_constant_3x4(%arg0 : tensor<3x4xf32>) -> tensor<2x2xf32>
{
%cst = arith.constant dense<[[10.0, 9.0, 8.0, 7.0], [11.0, 12.0, 13.0, 14.0], [1.0, 3.0, 5.0, 7.0]]> : tensor<3x4xf32>
%slice = tensor.extract_slice %cst[0, 0] [2, 2] [1, 1] : tensor<3x4xf32> to tensor<2x2xf32>
@ -30,7 +30,7 @@ func @slice_constant_3x4(%arg0 : tensor<3x4xf32>) -> tensor<2x2xf32>
// CHECK-NOT: tensor.extract_slice
// CHECK: %[[CONST:.+]] = arith.constant dense<{{\[}}[1.200000e+01, 1.300000e+01], [3.000000e+00, 5.000000e+00]]> : tensor<2x2xf32>
// CHECK: return %[[CONST]] : tensor<2x2xf32>
func @slice_constant_3x4_offsets(%arg0 : tensor<3x4xf32>) -> tensor<2x2xf32>
func.func @slice_constant_3x4_offsets(%arg0 : tensor<3x4xf32>) -> tensor<2x2xf32>
{
%cst = arith.constant dense<[[10.0, 9.0, 8.0, 7.0], [11.0, 12.0, 13.0, 14.0], [1.0, 3.0, 5.0, 7.0]]> : tensor<3x4xf32>
%slice = tensor.extract_slice %cst[1, 1] [2, 2] [1, 1] : tensor<3x4xf32> to tensor<2x2xf32>

78
mlir/test/Dialect/Tensor/invalid.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt <%s -split-input-file -verify-diagnostics
func @dim(%arg : tensor<1x?xf32>) {
func.func @dim(%arg : tensor<1x?xf32>) {
%c2 = arith.constant 2 : index
tensor.dim %arg, %c2 : tensor<1x?xf32> // expected-error {{'tensor.dim' op index is out of range}}
return
@ -8,7 +8,7 @@ func @dim(%arg : tensor<1x?xf32>) {
// -----
func @tensor.cast_mismatching_constants(%arg0: tensor<1xf32>) {
func.func @tensor.cast_mismatching_constants(%arg0: tensor<1xf32>) {
// expected-error@+1 {{operand type 'tensor<1xf32>' and result type 'tensor<2xf32>' are cast incompatible}}
%0 = tensor.cast %arg0 : tensor<1xf32> to tensor<2xf32>
return
@ -16,7 +16,7 @@ func @tensor.cast_mismatching_constants(%arg0: tensor<1xf32>) {
// -----
func @extract_too_many_indices(%arg0: tensor<?xf32>) {
func.func @extract_too_many_indices(%arg0: tensor<?xf32>) {
// expected-error@+1 {{incorrect number of indices for extract_element}}
%0 = tensor.extract %arg0[] : tensor<?xf32>
return
@ -24,7 +24,7 @@ func @extract_too_many_indices(%arg0: tensor<?xf32>) {
// -----
func @insert_too_many_indices(%arg0: f32, %arg1: tensor<?xf32>) {
func.func @insert_too_many_indices(%arg0: f32, %arg1: tensor<?xf32>) {
// expected-error@+1 {{incorrect number of indices}}
%0 = tensor.insert %arg0 into %arg1[] : tensor<?xf32>
return
@ -32,7 +32,7 @@ func @insert_too_many_indices(%arg0: f32, %arg1: tensor<?xf32>) {
// -----
func @tensor.from_elements_wrong_result_type() {
func.func @tensor.from_elements_wrong_result_type() {
// expected-error@+2 {{'result' must be statically shaped tensor of any type values, but got 'tensor<*xi32>'}}
%c0 = arith.constant 0 : i32
%0 = tensor.from_elements %c0 : tensor<*xi32>
@ -41,7 +41,7 @@ func @tensor.from_elements_wrong_result_type() {
// -----
func @tensor.from_elements_wrong_elements_count() {
func.func @tensor.from_elements_wrong_elements_count() {
// expected-error@+2 {{1 operands present, but expected 2}}
%c0 = arith.constant 0 : index
%0 = tensor.from_elements %c0 : tensor<2xindex>
@ -50,7 +50,7 @@ func @tensor.from_elements_wrong_elements_count() {
// -----
func @tensor.generate(%m : index)
func.func @tensor.generate(%m : index)
-> tensor<?x3x?xf32> {
// expected-error @+1 {{must have as many index operands as dynamic extents in the result type}}
%tnsr = tensor.generate %m {
@ -63,7 +63,7 @@ func @tensor.generate(%m : index)
// -----
func @tensor.generate(%m : index, %n : index)
func.func @tensor.generate(%m : index, %n : index)
-> tensor<?x3x?xf32> {
// expected-error @+1 {{must have one body argument per input dimension}}
%tnsr = tensor.generate %m, %n {
@ -76,7 +76,7 @@ func @tensor.generate(%m : index, %n : index)
// -----
func @tensor.generate(%m : index, %n : index)
func.func @tensor.generate(%m : index, %n : index)
-> tensor<?x3x?xf32> {
// expected-error @+1 {{all body arguments must be index}}
%tnsr = tensor.generate %m, %n {
@ -89,7 +89,7 @@ func @tensor.generate(%m : index, %n : index)
// -----
func @tensor.generate(%m : index, %n : index)
func.func @tensor.generate(%m : index, %n : index)
-> tensor<?x3x?xf32> {
// expected-error @+4 {{'func.return' op expects parent op 'func.func'}}
%tnsr = tensor.generate %m, %n {
@ -102,7 +102,7 @@ func @tensor.generate(%m : index, %n : index)
// -----
func @tensor.generate(%m : index, %n : index)
func.func @tensor.generate(%m : index, %n : index)
-> tensor<?x3x?xf32> {
// expected-error @+1 {{body must be terminated with a `yield` operation of the tensor element type}}
%tnsr = tensor.generate %m, %n {
@ -114,7 +114,7 @@ func @tensor.generate(%m : index, %n : index)
}
// -----
func @tensor.reshape_element_type_mismatch(
func.func @tensor.reshape_element_type_mismatch(
%buf: tensor<*xf32>, %shape: tensor<1xi32>) {
// expected-error @+1 {{element types of source and destination tensor types should be the same}}
tensor.reshape %buf(%shape) : (tensor<*xf32>, tensor<1xi32>) -> tensor<?xi32>
@ -122,7 +122,7 @@ func @tensor.reshape_element_type_mismatch(
// -----
func @tensor.reshape_dst_ranked_shape_unranked(
func.func @tensor.reshape_dst_ranked_shape_unranked(
%buf: tensor<*xf32>, %shape: tensor<?xi32>) {
// expected-error @+1 {{cannot use shape operand with dynamic length to reshape to statically-ranked tensor type}}
tensor.reshape %buf(%shape) : (tensor<*xf32>, tensor<?xi32>) -> tensor<?xf32>
@ -130,7 +130,7 @@ func @tensor.reshape_dst_ranked_shape_unranked(
// -----
func @tensor.reshape_dst_shape_rank_mismatch(
func.func @tensor.reshape_dst_shape_rank_mismatch(
%buf: tensor<*xf32>, %shape: tensor<1xi32>) {
// expected-error @+1 {{length of shape operand differs from the result's tensor rank}}
tensor.reshape %buf(%shape)
@ -139,7 +139,7 @@ func @tensor.reshape_dst_shape_rank_mismatch(
// -----
func @tensor.reshape_num_elements_mismatch(
func.func @tensor.reshape_num_elements_mismatch(
%buf: tensor<1xf32>, %shape: tensor<1xi32>) {
// expected-error @+1 {{source and destination tensor should have the same number of elements}}
tensor.reshape %buf(%shape)
@ -148,7 +148,7 @@ func @tensor.reshape_num_elements_mismatch(
// -----
func @extract_slice_wrong_result_rank(%t: tensor<?xf32>, %idx : index) {
func.func @extract_slice_wrong_result_rank(%t: tensor<?xf32>, %idx : index) {
// expected-error @+1 {{expected rank to be smaller or equal to the other rank.}}
%0 = tensor.extract_slice %t[0][4][1] : tensor<?xf32> to tensor<?x?xf32>
@ -157,7 +157,7 @@ func @extract_slice_wrong_result_rank(%t: tensor<?xf32>, %idx : index) {
// -----
func @extract_slice_wrong_result_rank(%t: tensor<?xf32>, %idx : index) {
func.func @extract_slice_wrong_result_rank(%t: tensor<?xf32>, %idx : index) {
// expected-error @+1 {{expected element type to be 'f32'}}
%0 = tensor.extract_slice %t[0][4][1] : tensor<?xf32> to tensor<4xi8>
@ -166,7 +166,7 @@ func @extract_slice_wrong_result_rank(%t: tensor<?xf32>, %idx : index) {
// -----
func @extract_slice_wrong_static_type(%t: tensor<8x16x4xf32>, %idx : index) {
func.func @extract_slice_wrong_static_type(%t: tensor<8x16x4xf32>, %idx : index) {
// expected-error @+1 {{expected type to be 'tensor<?x4x4xf32>' or a rank-reduced version. (size mismatch)}}
%0 = tensor.extract_slice %t[0, 0, 0][%idx, 4, 4][1, 1, 1]
: tensor<8x16x4xf32> to tensor<4x4x4xf32>
@ -176,7 +176,7 @@ func @extract_slice_wrong_static_type(%t: tensor<8x16x4xf32>, %idx : index) {
// -----
func @extract_slice_wrong_dynamic_type(%t: tensor<8x16x4xf32>, %idx : index) {
func.func @extract_slice_wrong_dynamic_type(%t: tensor<8x16x4xf32>, %idx : index) {
// expected-error @+1 {{expected type to be 'tensor<4x4x4xf32>' or a rank-reduced version. (size mismatch)}}
%0 = tensor.extract_slice %t[0, 2, 0][4, 4, 4][1, 1, 1]
: tensor<8x16x4xf32> to tensor<?x4x4xf32>
@ -186,7 +186,7 @@ func @extract_slice_wrong_dynamic_type(%t: tensor<8x16x4xf32>, %idx : index) {
// -----
func @insert_slice_wrong_result_rank(%t1: tensor<?xf32>, %t2: tensor<?x?xf32>, %idx : index) {
func.func @insert_slice_wrong_result_rank(%t1: tensor<?xf32>, %t2: tensor<?x?xf32>, %idx : index) {
// expected-error @+1 {{expected rank to be smaller or equal to the other rank.}}
%0 = tensor.insert_slice %t2 into %t1[0][4][1] : tensor<?x?xf32> into tensor<?xf32>
@ -195,7 +195,7 @@ func @insert_slice_wrong_result_rank(%t1: tensor<?xf32>, %t2: tensor<?x?xf32>, %
// -----
func @insert_slice_wrong_result_rank(%t1: tensor<4xi8>, %t2: tensor<?xf32>, %idx : index) {
func.func @insert_slice_wrong_result_rank(%t1: tensor<4xi8>, %t2: tensor<?xf32>, %idx : index) {
// expected-error @+1 {{expected element type to be 'f32'}}
%0 = tensor.insert_slice %t1 into %t2[0][4][1] : tensor<4xi8> into tensor<?xf32>
@ -204,7 +204,7 @@ func @insert_slice_wrong_result_rank(%t1: tensor<4xi8>, %t2: tensor<?xf32>, %idx
// -----
func @insert_slice_wrong_static_type(%t1: tensor<4x4x4xf32>, %t2: tensor<8x16x4xf32>, %idx : index) {
func.func @insert_slice_wrong_static_type(%t1: tensor<4x4x4xf32>, %t2: tensor<8x16x4xf32>, %idx : index) {
// expected-error @+1 {{expected type to be 'tensor<?x4x4xf32>' or a rank-reduced version. (size mismatch)}}
%0 = tensor.insert_slice %t1 into %t2[0, 0, 0][%idx, 4, 4][1, 1, 1]
: tensor<4x4x4xf32> into tensor<8x16x4xf32>
@ -214,7 +214,7 @@ func @insert_slice_wrong_static_type(%t1: tensor<4x4x4xf32>, %t2: tensor<8x16x4x
// -----
func @insert_slice_wrong_dynamic_type(%t1: tensor<?x4x4xf32>, %t2: tensor<8x16x4xf32>, %idx : index) {
func.func @insert_slice_wrong_dynamic_type(%t1: tensor<?x4x4xf32>, %t2: tensor<8x16x4xf32>, %idx : index) {
// expected-error @+1 {{expected type to be 'tensor<4x4x4xf32>' or a rank-reduced version. (size mismatch)}}
%0 = tensor.insert_slice %t1 into %t2[0, 2, 0][4, 4, 4][1, 1, 1]
: tensor<?x4x4xf32> into tensor<8x16x4xf32>
@ -224,7 +224,7 @@ func @insert_slice_wrong_dynamic_type(%t1: tensor<?x4x4xf32>, %t2: tensor<8x16x4
// -----
func @illegal_expanding_reshape_dynamic_tensor
func.func @illegal_expanding_reshape_dynamic_tensor
(%arg0: tensor<?x?x?xf32>) -> tensor<?x?x?x4x?xf32> {
// expected-error @+1 {{invalid to have a single dimension (2) expanded into multiple dynamic dims (2,4)}}
%0 = tensor.expand_shape %arg0 [[0], [1], [2, 3, 4]]
@ -235,7 +235,7 @@ func @illegal_expanding_reshape_dynamic_tensor
// -----
func @illegal_expanding_reshape_static_tensor
func.func @illegal_expanding_reshape_static_tensor
(%arg0: tensor<2x3x20xf32>) -> tensor<2x3x2x4x5xf32> {
// expected-error @+1 {{expected dimension 2 of collapsed type to be static value of 40}}
%0 = tensor.expand_shape %arg0 [[0], [1], [2, 3, 4]]
@ -245,7 +245,7 @@ func @illegal_expanding_reshape_static_tensor
// -----
func @illegal_collapsing_reshape_static_tensor
func.func @illegal_collapsing_reshape_static_tensor
(%arg0: tensor<2x3x2x4x5xf32>) -> tensor<2x3x20xf32> {
// expected-error @+1 {{expected dimension 2 of collapsed type to be static value of 40}}
%0 = tensor.collapse_shape %arg0 [[0], [1], [2, 3, 4]]
@ -255,7 +255,7 @@ func @illegal_collapsing_reshape_static_tensor
// -----
func @illegal_expanding_reshape_mixed_tensor(%arg0 : tensor<?x?xf32>)
func.func @illegal_expanding_reshape_mixed_tensor(%arg0 : tensor<?x?xf32>)
-> tensor<?x4x5xf32> {
// expected-error @+1 {{expected dimension 1 of collapsed type to be static value of 5}}
%0 = tensor.expand_shape %arg0 [[0, 1], [2]]
@ -265,7 +265,7 @@ func @illegal_expanding_reshape_mixed_tensor(%arg0 : tensor<?x?xf32>)
// -----
func @illegal_expanding_reshape_mixed_tensor_2(%arg0 : tensor<?x?xf32>)
func.func @illegal_expanding_reshape_mixed_tensor_2(%arg0 : tensor<?x?xf32>)
-> tensor<?x4x5xf32> {
// expected-error @+1 {{expected dimension 1 of collapsed type to be static value of 20}}
%0 = tensor.expand_shape %arg0 [[0], [1, 2]]
@ -275,7 +275,7 @@ func @illegal_expanding_reshape_mixed_tensor_2(%arg0 : tensor<?x?xf32>)
// -----
func @illegal_collapsing_reshape_mixed_tensor(%arg0 : tensor<?x4x5xf32>) -> tensor<?x?xf32> {
func.func @illegal_collapsing_reshape_mixed_tensor(%arg0 : tensor<?x4x5xf32>) -> tensor<?x?xf32> {
// expected-error @+1 {{expected dimension 1 of collapsed type to be static value of 5}}
%0 = tensor.collapse_shape %arg0 [[0, 1], [2]]
: tensor<?x4x5xf32> into tensor<?x?xf32>
@ -284,7 +284,7 @@ func @illegal_collapsing_reshape_mixed_tensor(%arg0 : tensor<?x4x5xf32>) -> tens
// -----
func @illegal_collapsing_reshape_mixed_tensor_2(%arg0 : tensor<?x4x5xf32>)
func.func @illegal_collapsing_reshape_mixed_tensor_2(%arg0 : tensor<?x4x5xf32>)
-> tensor<?x?xf32> {
// expected-error @+1 {{expected dimension 1 of collapsed type to be static value of 20}}
%0 = tensor.collapse_shape %arg0 [[0], [1, 2]]
@ -294,7 +294,7 @@ func @illegal_collapsing_reshape_mixed_tensor_2(%arg0 : tensor<?x4x5xf32>)
// -----
func @rank(%0: f32) {
func.func @rank(%0: f32) {
// expected-error@+1 {{'tensor.rank' op operand #0 must be tensor of any type values}}
"tensor.rank"(%0): (f32)->index
return
@ -302,7 +302,7 @@ func @rank(%0: f32) {
// -----
func @illegal_num_offsets(%arg0 : tensor<?x?x?xf32>, %arg1 : index, %arg2 : index) {
func.func @illegal_num_offsets(%arg0 : tensor<?x?x?xf32>, %arg1 : index, %arg2 : index) {
// expected-error@+1 {{expected 3 offset values}}
%0 = tensor.extract_slice %arg0[0, 0] [%arg1, %arg2] [1, 1] : tensor<?x?x?xf32> to tensor<?x?x?xf32>
return
@ -310,7 +310,7 @@ func @illegal_num_offsets(%arg0 : tensor<?x?x?xf32>, %arg1 : index, %arg2 : inde
// -----
func @illegal_num_offsets(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?x?x?xf32>,
func.func @illegal_num_offsets(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?x?x?xf32>,
%arg2 : index, %arg3 : index) {
// expected-error@+1 {{expected 3 offset values}}
%0 = tensor.insert_slice %arg0 into %arg1[0, 0] [%arg2, %arg3] [1, 1] : tensor<?x?xf32> into tensor<?x?x?xf32>
@ -320,7 +320,7 @@ func @illegal_num_offsets(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?x?x?xf32>,
// -----
func @pad_result_type(%arg0: tensor<?x2x3x4xi32>, %arg1: index, %arg2: i32) -> tensor<?x?x?x8xf32> {
func.func @pad_result_type(%arg0: tensor<?x2x3x4xi32>, %arg1: index, %arg2: i32) -> tensor<?x?x?x8xf32> {
// expected-error @+1 {{specified type 'tensor<?x?x?x8xf32>' does not match the inferred type 'tensor<?x?x?x9xi32>}}
%0 = tensor.pad %arg0 low[1, %arg1, 2, 2] high[1, 2, %arg1, 3] {
^bb0(%arg3: index, %arg4: index):
@ -331,7 +331,7 @@ func @pad_result_type(%arg0: tensor<?x2x3x4xi32>, %arg1: index, %arg2: i32) -> t
// -----
func @pad_number_of_block_args(%arg0: tensor<?x4xi32>, %arg1: i32) -> tensor<?x9xi32> {
func.func @pad_number_of_block_args(%arg0: tensor<?x4xi32>, %arg1: i32) -> tensor<?x9xi32> {
// expected-error @+1 {{expected the block to have 2 arguments}}
%0 = tensor.pad %arg0 low[1, 2] high[2, 3] {
^bb0(%arg2: index, %arg3: index, %arg4: index):
@ -342,7 +342,7 @@ func @pad_number_of_block_args(%arg0: tensor<?x4xi32>, %arg1: i32) -> tensor<?x9
// -----
func @pad_block_args(%arg0: tensor<?x4xi32>, %arg1: i32) -> tensor<?x9xi32> {
func.func @pad_block_args(%arg0: tensor<?x4xi32>, %arg1: i32) -> tensor<?x9xi32> {
// expected-error @+1 {{op expected block argument 1 to be an index}}
%0 = tensor.pad %arg0 low[1, 2] high[2, 3] {
^bb0(%arg2: i32, %arg3: i32):
@ -353,7 +353,7 @@ func @pad_block_args(%arg0: tensor<?x4xi32>, %arg1: i32) -> tensor<?x9xi32> {
// -----
func @pad_yield_type(%arg0: tensor<?x4xi32>, %arg1: i8) -> tensor<?x9xi32> {
func.func @pad_yield_type(%arg0: tensor<?x4xi32>, %arg1: i8) -> tensor<?x9xi32> {
// expected-error @+1 {{op expected yield type to match shape element type}}
%0 = tensor.pad %arg0 low[1, 2] high[2, 3] {
^bb0(%arg2: index, %arg3: index):
@ -364,7 +364,7 @@ func @pad_yield_type(%arg0: tensor<?x4xi32>, %arg1: i8) -> tensor<?x9xi32> {
// -----
func @invalid_splat(%v : f32) {
func.func @invalid_splat(%v : f32) {
// expected-error@+1 {{invalid kind of type specified}}
tensor.splat %v : memref<8xf32>
return
@ -372,7 +372,7 @@ func @invalid_splat(%v : f32) {
// -----
func @invalid_splat(%v : vector<8xf32>) {
func.func @invalid_splat(%v : vector<8xf32>) {
// expected-error@+1 {{must be integer/index/float type}}
%w = tensor.splat %v : tensor<8xvector<8xf32>>
return

32
mlir/test/Dialect/Tensor/ops.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt <%s | mlir-opt | FileCheck %s
// CHECK-LABEL: func @cast(
func @cast(%arg0: tensor<*xf32>, %arg1 : tensor<4x4xf32>, %arg2: tensor<?x?xf32>) {
func.func @cast(%arg0: tensor<*xf32>, %arg1 : tensor<4x4xf32>, %arg2: tensor<?x?xf32>) {
// CHECK: tensor.cast %arg0 : tensor<*xf32> to tensor<?x?xf32>
%0 = tensor.cast %arg0 : tensor<*xf32> to tensor<?x?xf32>
// CHECK: tensor.cast %arg1 : tensor<4x4xf32> to tensor<*xf32>
@ -16,7 +16,7 @@ func @cast(%arg0: tensor<*xf32>, %arg1 : tensor<4x4xf32>, %arg2: tensor<?x?xf32>
// CHECK-LABEL: func @extract(
// CHECK-SAME: %[[TENSOR:.*]]: tensor<?x?x?xf32>,
// CHECK-SAME: %[[INDEX:.*]]: index) {
func @extract(%arg0: tensor<?x?x?xf32>, %arg1: index) {
func.func @extract(%arg0: tensor<?x?x?xf32>, %arg1: index) {
// CHECK: tensor.extract %[[TENSOR]][%[[INDEX]], %[[INDEX]], %[[INDEX]]] : tensor<?x?x?xf32>
%0 = tensor.extract %arg0[%arg1, %arg1, %arg1] : tensor<?x?x?xf32>
return
@ -27,7 +27,7 @@ func @extract(%arg0: tensor<?x?x?xf32>, %arg1: index) {
// CHECK-SAME: %[[INDEX:.*]]: index
// CHECK-SAME: %[[DEST1:.*]]: tensor<?x?x?xf32>
// CHECK-SAME: %[[DEST2:.*]]: tensor<*xf32>
func @insert(%arg0: f32, %arg1: index, %arg2: tensor<?x?x?xf32>, %arg3: tensor<*xf32>) {
func.func @insert(%arg0: f32, %arg1: index, %arg2: tensor<?x?x?xf32>, %arg3: tensor<*xf32>) {
// CHECK: tensor.insert %[[SCALAR]] into %[[DEST1]][%[[INDEX]], %[[INDEX]], %[[INDEX]]] : tensor<?x?x?xf32>
%0 = tensor.insert %arg0 into %arg2[%arg1, %arg1, %arg1] : tensor<?x?x?xf32>
// CHECK: tensor.insert %[[SCALAR]] into %[[DEST2]][%[[INDEX]], %[[INDEX]], %[[INDEX]]] : tensor<*xf32>
@ -36,7 +36,7 @@ func @insert(%arg0: f32, %arg1: index, %arg2: tensor<?x?x?xf32>, %arg3: tensor<*
}
// CHECK-LABEL: func @tensor.from_elements() {
func @tensor.from_elements() {
func.func @tensor.from_elements() {
%c0 = "arith.constant"() {value = 0: index} : () -> index
// CHECK: tensor.from_elements %c0 : tensor<1xindex>
%0 = tensor.from_elements %c0 : tensor<1xindex>
@ -62,7 +62,7 @@ func @tensor.from_elements() {
}
// CHECK-LABEL: @tensor.generate
func @tensor.generate(%m : index, %n : index)
func.func @tensor.generate(%m : index, %n : index)
-> tensor<?x3x?xf32> {
%tnsr = tensor.generate %m, %n {
^bb0(%i : index, %j : index, %k : index):
@ -73,7 +73,7 @@ func @tensor.generate(%m : index, %n : index)
}
// CHECK-LABEL: func @tensor_reshape
func @tensor_reshape(%unranked: tensor<*xf32>, %shape1: tensor<1xi32>,
func.func @tensor_reshape(%unranked: tensor<*xf32>, %shape1: tensor<1xi32>,
%shape2: tensor<2xi32>, %shape3: tensor<?xi32>) -> tensor<*xf32> {
%dyn_vec = tensor.reshape %unranked(%shape1)
: (tensor<*xf32>, tensor<1xi32>) -> tensor<?xf32>
@ -85,7 +85,7 @@ func @tensor_reshape(%unranked: tensor<*xf32>, %shape1: tensor<1xi32>,
}
// CHECK-LABEL: func @slice({{.*}}) {
func @slice(%t: tensor<8x16x4xf32>, %idx : index) {
func.func @slice(%t: tensor<8x16x4xf32>, %idx : index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -110,7 +110,7 @@ func @slice(%t: tensor<8x16x4xf32>, %idx : index) {
// -----
// CHECK-LABEL: func @insert_slice({{.*}}) {
func @insert_slice(
func.func @insert_slice(
%t: tensor<8x16x4xf32>,
%td: tensor<8x?x4xf32>,
%t2: tensor<16x32x8xf32>,
@ -145,7 +145,7 @@ func @insert_slice(
// -----
func @tensor_reshape_zero_dim(%arg0 : tensor<1x1xf32>, %arg1 : tensor<f32>)
func.func @tensor_reshape_zero_dim(%arg0 : tensor<1x1xf32>, %arg1 : tensor<f32>)
-> (tensor<f32>, tensor<1x1xf32>) {
%0 = tensor.collapse_shape %arg0 [] : tensor<1x1xf32> into tensor<f32>
%1 = tensor.expand_shape %0 [] : tensor<f32> into tensor<1x1xf32>
@ -155,7 +155,7 @@ func @tensor_reshape_zero_dim(%arg0 : tensor<1x1xf32>, %arg1 : tensor<f32>)
// CHECK: tensor.collapse_shape %{{.*}} [] : tensor<1x1xf32> into tensor<f32>
// CHECK: tensor.expand_shape %{{.*}} [] : tensor<f32> into tensor<1x1xf32>
func @legal_collapsing_reshape_dynamic_tensor
func.func @legal_collapsing_reshape_dynamic_tensor
(%arg0: tensor<?x?x?x4x?xf32>) -> tensor<?x?x?xf32>
{
%0 = tensor.collapse_shape %arg0 [[0], [1], [2, 3, 4]] :
@ -168,7 +168,7 @@ func @legal_collapsing_reshape_dynamic_tensor
// -----
func @rank(%t : tensor<4x4x?xf32>) {
func.func @rank(%t : tensor<4x4x?xf32>) {
// CHECK: %{{.*}} = tensor.rank %{{.*}} : tensor<4x4x?xf32>
%0 = "tensor.rank"(%t) : (tensor<4x4x?xf32>) -> index
@ -179,7 +179,7 @@ func @rank(%t : tensor<4x4x?xf32>) {
// -----
func @pad_dynamic(%arg0: tensor<1x2x2x?xf32>, %low: index, %high: index,
func.func @pad_dynamic(%arg0: tensor<1x2x2x?xf32>, %low: index, %high: index,
%pad_value: f32) -> tensor<6x?x?x?xf32> {
%0 = tensor.pad %arg0 low[2, %low, 3, 3] high[3, 3, %high, 2] {
^bb0(%arg1: index, %arg2: index, %arg3: index, %arg4: index):
@ -198,7 +198,7 @@ func @pad_dynamic(%arg0: tensor<1x2x2x?xf32>, %low: index, %high: index,
// -----
func @pad_static(%arg0: tensor<3x4xf32>, %pad_value: f32) -> tensor<6x9xf32> {
func.func @pad_static(%arg0: tensor<3x4xf32>, %pad_value: f32) -> tensor<6x9xf32> {
%0 = tensor.pad %arg0 low[1, 2] high[2, 3] {
^bb0(%arg1 : index, %arg2 : index):
tensor.yield %pad_value : f32
@ -212,7 +212,7 @@ func @pad_static(%arg0: tensor<3x4xf32>, %pad_value: f32) -> tensor<6x9xf32> {
// -----
func @pad_asymmetrical(%arg0: tensor<2x3xf32>, %ub0: index, %ub1: index,
func.func @pad_asymmetrical(%arg0: tensor<2x3xf32>, %ub0: index, %ub1: index,
%pad_value: f32) -> tensor<?x?xf32> {
%0 = tensor.pad %arg0 low[0, 0] high[%ub0, %ub1] {
^bb0(%arg1: index, %arg2: index):
@ -231,7 +231,7 @@ func @pad_asymmetrical(%arg0: tensor<2x3xf32>, %ub0: index, %ub1: index,
// -----
func @pad_to_static_size(%arg0: tensor<?x?xf32>, %ub0: index, %ub1: index,
func.func @pad_to_static_size(%arg0: tensor<?x?xf32>, %ub0: index, %ub1: index,
%pad_value: f32) -> tensor<2x3xf32> {
%0 = tensor.pad %arg0 low[0, 0] high[%ub0, %ub1] {
^bb0(%arg1: index, %arg2: index):
@ -252,7 +252,7 @@ func @pad_to_static_size(%arg0: tensor<?x?xf32>, %ub0: index, %ub1: index,
// CHECK-LABEL: func @test_splat_op
// CHECK-SAME: [[S:%arg[0-9]+]]: f32
func @test_splat_op(%s : f32) {
func.func @test_splat_op(%s : f32) {
// CHECK: tensor.splat [[S]] : tensor<8xf32>
%v = tensor.splat %s : tensor<8xf32>

16
mlir/test/Dialect/Tensor/resolve-shaped-type-result-dims.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt -resolve-shaped-type-result-dims -split-input-file %s | FileCheck %s
func @insert_slice(
func.func @insert_slice(
%arg0 : tensor<?x?x?xf32>, %arg1 : tensor<?x?x?xf32>,
%arg2 : index, %arg3 : index, %arg4 : index) -> (index, index, index) {
%c0 = arith.constant 0 : index
@ -28,7 +28,7 @@ func @insert_slice(
// -----
func @extract_slice(%arg0 : tensor<?x?x?xf32>, %arg1 : index, %arg2 : index,
func.func @extract_slice(%arg0 : tensor<?x?x?xf32>, %arg1 : index, %arg2 : index,
%arg3 : index) -> (index, index, index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -49,7 +49,7 @@ func @extract_slice(%arg0 : tensor<?x?x?xf32>, %arg1 : index, %arg2 : index,
// -----
func @extract_slice_rank_reduced_1(%arg0 : tensor<?x?x?xf32>,
func.func @extract_slice_rank_reduced_1(%arg0 : tensor<?x?x?xf32>,
%arg1 : index) -> index {
%c0 = arith.constant 0 : index
%0 = tensor.extract_slice %arg0[0, 0, 0] [1, %arg1, 1] [1, 1, 1] :
@ -64,7 +64,7 @@ func @extract_slice_rank_reduced_1(%arg0 : tensor<?x?x?xf32>,
// -----
func @extract_slice_rank_reduced_2(%arg0 : tensor<?x?x?xf32>,
func.func @extract_slice_rank_reduced_2(%arg0 : tensor<?x?x?xf32>,
%arg1 : index) -> index {
%c0 = arith.constant 0 : index
%0 = tensor.extract_slice %arg0[0, 0, 0] [1, %arg1, 1] [1, 1, 1] :
@ -79,7 +79,7 @@ func @extract_slice_rank_reduced_2(%arg0 : tensor<?x?x?xf32>,
// -----
func @extract_slice_rank_reduced_3(%arg0 : tensor<?x?x?xf32>,
func.func @extract_slice_rank_reduced_3(%arg0 : tensor<?x?x?xf32>,
%arg1 : index) -> index {
%c1 = arith.constant 1 : index
%0 = tensor.extract_slice %arg0[0, 0, 0] [1, %arg1, 1] [1, 1, 1] :
@ -94,7 +94,7 @@ func @extract_slice_rank_reduced_3(%arg0 : tensor<?x?x?xf32>,
// -----
func @extract_slice_rank_reduced_4(%arg0 : tensor<?x?x?xf32>,
func.func @extract_slice_rank_reduced_4(%arg0 : tensor<?x?x?xf32>,
%arg1 : index) -> index {
%c1 = arith.constant 1 : index
%0 = tensor.extract_slice %arg0[0, 0, 0] [1, %arg1, 1] [1, 1, 1] :
@ -109,7 +109,7 @@ func @extract_slice_rank_reduced_4(%arg0 : tensor<?x?x?xf32>,
// -----
func @extract_slice_rank_reduced_5(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @extract_slice_rank_reduced_5(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index) -> (index, index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -127,7 +127,7 @@ func @extract_slice_rank_reduced_5(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
// -----
func @extract_slice_rank_reduced_6(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
func.func @extract_slice_rank_reduced_6(%arg0 : tensor<?x?x?xf32>, %arg1 : index,
%arg2 : index) -> (index, index) {
%c0 = arith.constant 0 : index
%c2 = arith.constant 2 : index

4
mlir/test/Dialect/Tensor/split-padding.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt -split-input-file -test-tensor-transform-patterns=test-split-padding-patterns %s | FileCheck %s
// CHECK-LABEL: func @pad_all_zero_sizes
func @pad_all_zero_sizes(%input: tensor<?x?x?xf32>) -> tensor<?x?x?xf32> {
func.func @pad_all_zero_sizes(%input: tensor<?x?x?xf32>) -> tensor<?x?x?xf32> {
%f0 = arith.constant 0.0 : f32
%c0 = arith.constant 0 : index
%0 = tensor.pad %input low[0, %c0, 0] high[%c0, 0, 0] {
@ -18,7 +18,7 @@ func @pad_all_zero_sizes(%input: tensor<?x?x?xf32>) -> tensor<?x?x?xf32> {
// CHECK-LABEL: func @pad_non_zero_sizes
// CHECK-SAME: (%[[INPUT:.+]]: tensor<?x?x8xf32>, %[[LOW0:.+]]: index, %[[HIGH1:.+]]: index)
func @pad_non_zero_sizes(%input: tensor<?x?x8xf32>, %low0: index, %high1: index) -> tensor<?x?x8xf32> {
func.func @pad_non_zero_sizes(%input: tensor<?x?x8xf32>, %low0: index, %high1: index) -> tensor<?x?x8xf32> {
%f0 = arith.constant 0.0 : f32
%0 = tensor.pad %input low[%low0, 0, 0] high[0, %high1, 0] {
^bb0(%dim0: index, %dim1: index, %dim2: index):

44
mlir/test/Dialect/Tosa/broadcast.mlir
View File

@ -2,7 +2,7 @@
// -----
// CHECK-LABEL: broadcast0
func @test_broadcast0(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
func.func @test_broadcast0(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
// CHECK-NOT: reshape
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
return %0 : tensor<1xf32>
@ -10,7 +10,7 @@ func @test_broadcast0(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf3
// -----
// CHECK-LABEL: broadcast1
func @test_broadcast1(%arg0: tensor<1xf32>, %arg1: tensor<2x1xf32>) -> tensor<2x1xf32> {
func.func @test_broadcast1(%arg0: tensor<1xf32>, %arg1: tensor<2x1xf32>) -> tensor<2x1xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1xf32>, tensor<2x1xf32>) -> tensor<2x1xf32>
@ -19,7 +19,7 @@ func @test_broadcast1(%arg0: tensor<1xf32>, %arg1: tensor<2x1xf32>) -> tensor<2x
// -----
// CHECK-LABEL: broadcast2
func @test_broadcast2(%arg0: tensor<2x1xf32>, %arg1: tensor<1xf32>) -> tensor<2x1xf32> {
func.func @test_broadcast2(%arg0: tensor<2x1xf32>, %arg1: tensor<1xf32>) -> tensor<2x1xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<2x1xf32>, tensor<1xf32>) -> tensor<2x1xf32>
@ -28,7 +28,7 @@ func @test_broadcast2(%arg0: tensor<2x1xf32>, %arg1: tensor<1xf32>) -> tensor<2x
// -----
// CHECK-LABEL: broadcast3
func @test_broadcast3(%arg0: tensor<2x1x1x1xf32>, %arg1: tensor<1xf32>) -> tensor<2x1x1x1xf32> {
func.func @test_broadcast3(%arg0: tensor<2x1x1x1xf32>, %arg1: tensor<1xf32>) -> tensor<2x1x1x1xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<2x1x1x1xf32>, tensor<1xf32>) -> tensor<2x1x1x1xf32>
@ -37,7 +37,7 @@ func @test_broadcast3(%arg0: tensor<2x1x1x1xf32>, %arg1: tensor<1xf32>) -> tenso
// -----
// CHECK-LABEL: broadcast4
func @test_broadcast4(%arg0: tensor<1x1x1x2xf32>, %arg1: tensor<1xf32>) -> tensor<1x1x1x2xf32> {
func.func @test_broadcast4(%arg0: tensor<1x1x1x2xf32>, %arg1: tensor<1xf32>) -> tensor<1x1x1x2xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1x1x1x2xf32>, tensor<1xf32>) -> tensor<1x1x1x2xf32>
@ -46,7 +46,7 @@ func @test_broadcast4(%arg0: tensor<1x1x1x2xf32>, %arg1: tensor<1xf32>) -> tenso
// -----
// CHECK-LABEL: broadcast5
func @test_broadcast5(%arg0: tensor<1x1x2x1xf32>, %arg1: tensor<1xf32>) -> tensor<1x1x2x1xf32> {
func.func @test_broadcast5(%arg0: tensor<1x1x2x1xf32>, %arg1: tensor<1xf32>) -> tensor<1x1x2x1xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1x1x2x1xf32>, tensor<1xf32>) -> tensor<1x1x2x1xf32>
@ -55,7 +55,7 @@ func @test_broadcast5(%arg0: tensor<1x1x2x1xf32>, %arg1: tensor<1xf32>) -> tenso
// -----
// CHECK-LABEL: broadcast6
func @test_broadcast6(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<1xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast6(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<1xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<17x16x15x14xf32>, tensor<1xf32>) -> tensor<17x16x15x14xf32>
@ -64,7 +64,7 @@ func @test_broadcast6(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<1xf32>) -> t
// -----
// CHECK-LABEL: broadcast7
func @test_broadcast7(%arg0: tensor<17x16x1x14xf32>, %arg1: tensor<1x1xf32>) -> tensor<17x16x1x14xf32> {
func.func @test_broadcast7(%arg0: tensor<17x16x1x14xf32>, %arg1: tensor<1x1xf32>) -> tensor<17x16x1x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<17x16x1x14xf32>, tensor<1x1xf32>) -> tensor<17x16x1x14xf32>
@ -73,7 +73,7 @@ func @test_broadcast7(%arg0: tensor<17x16x1x14xf32>, %arg1: tensor<1x1xf32>) ->
// -----
// CHECK-LABEL: broadcast8
func @test_broadcast8(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<1x1xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast8(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<1x1xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<17x16x15x14xf32>, tensor<1x1xf32>) -> tensor<17x16x15x14xf32>
@ -82,7 +82,7 @@ func @test_broadcast8(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<1x1xf32>) ->
// -----
// CHECK-LABEL: broadcast9
func @test_broadcast9(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<15x1xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast9(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<15x1xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 15, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<17x16x15x14xf32>, tensor<15x1xf32>) -> tensor<17x16x15x14xf32>
@ -91,7 +91,7 @@ func @test_broadcast9(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<15x1xf32>) -
// -----
// CHECK-LABEL: broadcast10
func @test_broadcast10(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<15x14xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast10(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<15x14xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 15, 14]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<17x16x15x14xf32>, tensor<15x14xf32>) -> tensor<17x16x15x14xf32>
@ -100,7 +100,7 @@ func @test_broadcast10(%arg0: tensor<17x16x15x14xf32>, %arg1: tensor<15x14xf32>)
// -----
// CHECK-LABEL: broadcast13
func @test_broadcast13(%arg0: tensor<1xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast13(%arg0: tensor<1xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1xf32>, tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32>
@ -109,7 +109,7 @@ func @test_broadcast13(%arg0: tensor<1xf32>, %arg1: tensor<17x16x15x14xf32>) ->
// -----
// CHECK-LABEL: broadcast14
func @test_broadcast14(%arg0: tensor<1x1xf32>, %arg1: tensor<17x16x1x14xf32>) -> tensor<17x16x1x14xf32> {
func.func @test_broadcast14(%arg0: tensor<1x1xf32>, %arg1: tensor<17x16x1x14xf32>) -> tensor<17x16x1x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1x1xf32>, tensor<17x16x1x14xf32>) -> tensor<17x16x1x14xf32>
@ -118,7 +118,7 @@ func @test_broadcast14(%arg0: tensor<1x1xf32>, %arg1: tensor<17x16x1x14xf32>) ->
// -----
// CHECK-LABEL: broadcast15
func @test_broadcast15(%arg0: tensor<1x1xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast15(%arg0: tensor<1x1xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<1x1xf32>, tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32>
@ -127,7 +127,7 @@ func @test_broadcast15(%arg0: tensor<1x1xf32>, %arg1: tensor<17x16x15x14xf32>) -
// -----
// CHECK-LABEL: broadcast16
func @test_broadcast16(%arg0: tensor<15x1xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast16(%arg0: tensor<15x1xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 15, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<15x1xf32>, tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32>
@ -136,7 +136,7 @@ func @test_broadcast16(%arg0: tensor<15x1xf32>, %arg1: tensor<17x16x15x14xf32>)
// -----
// CHECK-LABEL: broadcast17
func @test_broadcast17(%arg0: tensor<15x14xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
func.func @test_broadcast17(%arg0: tensor<15x14xf32>, %arg1: tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 15, 14]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<15x14xf32>, tensor<17x16x15x14xf32>) -> tensor<17x16x15x14xf32>
@ -145,7 +145,7 @@ func @test_broadcast17(%arg0: tensor<15x14xf32>, %arg1: tensor<17x16x15x14xf32>)
// -----
// CHECK-LABEL: broadcast18
func @test_broadcast18(%arg0: tensor<14x1xf32>, %arg1: tensor<1x15xf32>) -> tensor<14x15xf32> {
func.func @test_broadcast18(%arg0: tensor<14x1xf32>, %arg1: tensor<1x15xf32>) -> tensor<14x15xf32> {
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<14x1xf32>, tensor<1x15xf32>) -> tensor<14x15xf32>
return %0 : tensor<14x15xf32>
@ -153,7 +153,7 @@ func @test_broadcast18(%arg0: tensor<14x1xf32>, %arg1: tensor<1x15xf32>) -> tens
// -----
// CHECK-LABEL: broadcast19
func @test_broadcast19(%arg0: tensor<64x64x1xf32>, %arg1: tensor<1x17xf32>) -> (tensor<64x64x17xf32> ) {
func.func @test_broadcast19(%arg0: tensor<64x64x1xf32>, %arg1: tensor<1x17xf32>) -> (tensor<64x64x17xf32> ) {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 17]}
// CHECK: %[[VAR1:.*]] = "tosa.sub"(%arg0, %[[VAR0]])
%0 = "tosa.sub"(%arg0, %arg1) : (tensor<64x64x1xf32>, tensor<1x17xf32>) -> tensor<64x64x17xf32>
@ -162,7 +162,7 @@ func @test_broadcast19(%arg0: tensor<64x64x1xf32>, %arg1: tensor<1x17xf32>) -> (
// -----
// CHECK-LABEL: broadcast20
func @test_broadcast20(%arg0: tensor<3x3x4x1xf32>, %arg1: tensor<4x5xf32>) -> (tensor<3x3x4x5xf32> ) {
func.func @test_broadcast20(%arg0: tensor<3x3x4x1xf32>, %arg1: tensor<4x5xf32>) -> (tensor<3x3x4x5xf32> ) {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg1) {new_shape = [1, 1, 4, 5]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%arg0, %[[VAR0]])
%0 = "tosa.add"(%arg0, %arg1) : (tensor<3x3x4x1xf32>, tensor<4x5xf32>) -> tensor<3x3x4x5xf32>
@ -171,7 +171,7 @@ func @test_broadcast20(%arg0: tensor<3x3x4x1xf32>, %arg1: tensor<4x5xf32>) -> (t
// -----
// CHECK-LABEL: broadcast_mul
func @test_broadcast_mul(%arg0: tensor<15x14xi32>, %arg1: tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32> {
func.func @test_broadcast_mul(%arg0: tensor<15x14xi32>, %arg1: tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 15, 14]}
// CHECK: %[[VAR1:.*]] = "tosa.mul"(%[[VAR0]], %arg1)
%0 = "tosa.mul"(%arg0, %arg1) {shift = 1 : i32 } : (tensor<15x14xi32>, tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32>
@ -180,7 +180,7 @@ func @test_broadcast_mul(%arg0: tensor<15x14xi32>, %arg1: tensor<17x16x15x14xi32
// -----
// CHECK-LABEL: broadcast_arithmetic_right_shift
func @test_broadcast_arithmetic_right_shift(%arg0: tensor<15x14xi32>, %arg1: tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32> {
func.func @test_broadcast_arithmetic_right_shift(%arg0: tensor<15x14xi32>, %arg1: tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 15, 14]}
// CHECK: %[[VAR1:.*]] = "tosa.arithmetic_right_shift"(%[[VAR0]], %arg1)
%0 = "tosa.arithmetic_right_shift"(%arg0, %arg1) { round = true } : (tensor<15x14xi32>, tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32>
@ -189,7 +189,7 @@ func @test_broadcast_arithmetic_right_shift(%arg0: tensor<15x14xi32>, %arg1: ten
// -----
// CHECK-LABEL: broadcast_scalar
func @test_broadcast_scalar(%arg0: tensor<i32>, %arg1: tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32> {
func.func @test_broadcast_scalar(%arg0: tensor<i32>, %arg1: tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32> {
// CHECK-DAG: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [1, 1, 1, 1]}
// CHECK: %[[VAR1:.*]] = "tosa.add"(%[[VAR0]], %arg1)
%0 = "tosa.add"(%arg0, %arg1) : (tensor<i32>, tensor<17x16x15x14xi32>) -> tensor<17x16x15x14xi32>

122
mlir/test/Dialect/Tosa/canonicalize.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt --split-input-file --canonicalize %s | FileCheck %s
// CHECK-LABEL: @argmax_nofold
func @argmax_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @argmax_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.argmax"
%0 = "tosa.argmax"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -10,7 +10,7 @@ func @argmax_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @add_zero_different_shape
func @add_zero_different_shape(%arg0: tensor<2x3xi32>) -> tensor<4x2x3xi32> {
func.func @add_zero_different_shape(%arg0: tensor<2x3xi32>) -> tensor<4x2x3xi32> {
// CHECK: tosa.add
%zeros = "tosa.const"() {value = dense<0> : tensor<4x2x3xi32>} : () -> tensor<4x2x3xi32>
%1 = "tosa.add"(%arg0, %zeros) : (tensor<2x3xi32>, tensor<4x2x3xi32>) -> tensor<4x2x3xi32>
@ -21,7 +21,7 @@ func @add_zero_different_shape(%arg0: tensor<2x3xi32>) -> tensor<4x2x3xi32> {
// -----
// CHECK-LABEL: @add_zero_int
func @add_zero_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
func.func @add_zero_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
// CHECK: return %arg0
// CHECK-NOT: tosa.add
%zeros = "tosa.const"() {value = dense<0> : tensor<2x3xi32>} : () -> tensor<2x3xi32>
@ -32,7 +32,7 @@ func @add_zero_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
// -----
// CHECK-LABEL: @cast_fold
func @cast_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @cast_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.cast"(%arg0) : (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -41,7 +41,7 @@ func @cast_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @cast_nofold
func @cast_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xi32> {
func.func @cast_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xi32> {
// CHECK: "tosa.cast"
%0 = "tosa.cast"(%arg0) : (tensor<?x1xf32>) -> tensor<?x1xi32>
return %0 : tensor<?x1xi32>
@ -50,7 +50,7 @@ func @cast_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xi32> {
// -----
// CHECK-LABEL: @clamp_not_noop
func @clamp_not_noop(%arg0: tensor<4xi32>) -> tensor<4xi32> {
func.func @clamp_not_noop(%arg0: tensor<4xi32>) -> tensor<4xi32> {
// CHECK: "tosa.clamp"
%0 = "tosa.clamp"(%arg0) {min_int = 1 : i64, max_int = 4 : i64, min_fp = 1.0 : f32, max_fp = 4.0 : f32} : (tensor<4xi32>) -> tensor<4xi32>
return %0 : tensor<4xi32>
@ -59,7 +59,7 @@ func @clamp_not_noop(%arg0: tensor<4xi32>) -> tensor<4xi32> {
// -----
// CHECK-LABEL: @clamp_float_is_noop
func @clamp_float_is_noop(%arg0: tensor<4xf32>) -> tensor<4xf32> {
func.func @clamp_float_is_noop(%arg0: tensor<4xf32>) -> tensor<4xf32> {
// CHECK: return %arg0
// CHECK-NOT: "tosa.clamp"
%0 = "tosa.clamp"(%arg0) {min_int = -128 : i64, max_int = 127 : i64, min_fp = -3.40282347E+38 : f32, max_fp = 3.40282347E+38 : f32} : (tensor<4xf32>) -> tensor<4xf32>
@ -69,7 +69,7 @@ func @clamp_float_is_noop(%arg0: tensor<4xf32>) -> tensor<4xf32> {
// -----
// CHECK-LABEL: @clamp_int8_is_noop
func @clamp_int8_is_noop(%arg0: tensor<4xi8>) -> tensor<4xi8> {
func.func @clamp_int8_is_noop(%arg0: tensor<4xi8>) -> tensor<4xi8> {
// CHECK: return %arg0
// CHECK-NOT: "tosa.clamp"
%0 = "tosa.clamp"(%arg0) {min_int = -128 : i64, max_int = 127 : i64, min_fp = -3.40282347E+38 : f32, max_fp = 3.40282347E+38 : f32} : (tensor<4xi8>) -> tensor<4xi8>
@ -79,7 +79,7 @@ func @clamp_int8_is_noop(%arg0: tensor<4xi8>) -> tensor<4xi8> {
// -----
// CHECK-LABEL: @clamp_int16_is_noop
func @clamp_int16_is_noop(%arg0: tensor<4xi16>) -> tensor<4xi16> {
func.func @clamp_int16_is_noop(%arg0: tensor<4xi16>) -> tensor<4xi16> {
// CHECK: return %arg0
// CHECK-NOT: "tosa.clamp"
%0 = "tosa.clamp"(%arg0) {min_int = -32768 : i64, max_int = 32767 : i64, min_fp = -3.40282347E+38 : f32, max_fp = 3.40282347E+38 : f32} : (tensor<4xi16>) -> tensor<4xi16>
@ -89,7 +89,7 @@ func @clamp_int16_is_noop(%arg0: tensor<4xi16>) -> tensor<4xi16> {
// -----
// CHECK-LABEL: @clamp_uint8_is_noop
func @clamp_uint8_is_noop(%arg0: tensor<4xui8>) -> tensor<4xui8> {
func.func @clamp_uint8_is_noop(%arg0: tensor<4xui8>) -> tensor<4xui8> {
// CHECK: return %arg0
// CHECK-NOT: "tosa.clamp"
%0 = "tosa.clamp"(%arg0) {min_int = 0 : i64, max_int = 255 : i64, min_fp = -3.40282347E+38 : f32, max_fp = 3.40282347E+38 : f32} : (tensor<4xui8>) -> tensor<4xui8>
@ -99,7 +99,7 @@ func @clamp_uint8_is_noop(%arg0: tensor<4xui8>) -> tensor<4xui8> {
// -----
// CHECK-LABEL: @clamp_twice_is_single_clamp
func @clamp_twice_is_single_clamp(%arg0: tensor<4xi8>) -> tensor<4xi8> {
func.func @clamp_twice_is_single_clamp(%arg0: tensor<4xi8>) -> tensor<4xi8> {
// CHECK: "tosa.clamp"(%arg0) {max_fp = 3.000000e+00 : f32, max_int = 2 : i64, min_fp = -3.000000e+00 : f32, min_int = -2 : i64}
%0 = "tosa.clamp"(%arg0) {max_fp = 3.0 : f32, max_int = 4 : i64, min_fp = -5.0 : f32, min_int = -2 : i64} : (tensor<4xi8>) -> tensor<4xi8>
%1 = "tosa.clamp"(%0) {max_fp = 5.0 : f32, max_int = 2 : i64, min_fp = -3.0 : f32, min_int = -4 : i64} : (tensor<4xi8>) -> tensor<4xi8>
@ -109,7 +109,7 @@ func @clamp_twice_is_single_clamp(%arg0: tensor<4xi8>) -> tensor<4xi8> {
// -----
// CHECK-LABEL: @concat_fold
func @concat_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @concat_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.concat"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -118,7 +118,7 @@ func @concat_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @concat_fold_cast
func @concat_fold_cast(%arg0: tensor<?x1xf32>) -> tensor<?x?xf32> {
func.func @concat_fold_cast(%arg0: tensor<?x1xf32>) -> tensor<?x?xf32> {
// CHECK: %[[VAR0:.*]] = tensor.cast %arg0
// CHECK: return %[[VAR0]]
%0 = "tosa.concat"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x?xf32>
@ -128,7 +128,7 @@ func @concat_fold_cast(%arg0: tensor<?x1xf32>) -> tensor<?x?xf32> {
// -----
// CHECK-LABEL: @conv2d_stride_2
func @conv2d_stride_2(%arg0: tensor<4x10x10x2xf32>) -> tensor<4x10x10x3xf32> {
func.func @conv2d_stride_2(%arg0: tensor<4x10x10x2xf32>) -> tensor<4x10x10x3xf32> {
// CHECK: "tosa.conv2d"
%weight = "tosa.const"() {value = dense<[[[[1.0, 1.0]]], [[[1.0, 1.0]]], [[[1.0, 1.0]]]]> : tensor<3x1x1x2xf32>} : ()-> tensor<3x1x1x2xf32>
%bias = "tosa.const"() {value = dense<0.0> : tensor<3xf32>} : ()-> tensor<3xf32>
@ -139,7 +139,7 @@ func @conv2d_stride_2(%arg0: tensor<4x10x10x2xf32>) -> tensor<4x10x10x3xf32> {
// -----
// CHECK-LABEL: @conv2d_weight_2x2
func @conv2d_weight_2x2(%arg0: tensor<4x10x10x1xf32>) -> tensor<4x10x10x1xf32> {
func.func @conv2d_weight_2x2(%arg0: tensor<4x10x10x1xf32>) -> tensor<4x10x10x1xf32> {
// CHECK: "tosa.conv2d"
%weight = "tosa.const"() {value = dense<[[[[1.0], [1.0]], [[1.0], [1.0]]]]> : tensor<1x2x2x1xf32>} : ()-> tensor<1x2x2x1xf32>
%bias = "tosa.const"() {value = dense<0.0> : tensor<1xf32>} : ()-> tensor<1xf32>
@ -150,7 +150,7 @@ func @conv2d_weight_2x2(%arg0: tensor<4x10x10x1xf32>) -> tensor<4x10x10x1xf32> {
// -----
// CHECK-LABEL: @depthwise_conv2d_stride_2
func @depthwise_conv2d_stride_2(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<1x1x2x3xf32>, %arg2: tensor<6xf32>) -> tensor<4x10x10x6xf32> {
func.func @depthwise_conv2d_stride_2(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<1x1x2x3xf32>, %arg2: tensor<6xf32>) -> tensor<4x10x10x6xf32> {
// CHECK: "tosa.depthwise_conv2d"
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {pad = [0, 0, 0, 0], stride = [2, 2], dilation = [1, 1]} : (tensor<4x10x10x2xf32>, tensor<1x1x2x3xf32>, tensor<6xf32>) -> tensor<4x10x10x6xf32>
return %0 : tensor<4x10x10x6xf32>
@ -159,7 +159,7 @@ func @depthwise_conv2d_stride_2(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<1x1x
// -----
// CHECK-LABEL: @depthwise_conv2d_weight_2x2
func @depthwise_conv2d_weight_2x2(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<2x2x2x3xf32>, %arg2: tensor<6xf32>) -> tensor<4x10x10x6xf32> {
func.func @depthwise_conv2d_weight_2x2(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<2x2x2x3xf32>, %arg2: tensor<6xf32>) -> tensor<4x10x10x6xf32> {
// CHECK: "tosa.depthwise_conv2d"
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1]} : (tensor<4x10x10x2xf32>, tensor<2x2x2x3xf32>, tensor<6xf32>) -> tensor<4x10x10x6xf32>
return %0 : tensor<4x10x10x6xf32>
@ -168,7 +168,7 @@ func @depthwise_conv2d_weight_2x2(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<2x
// -----
// CHECK-LABEL: @max_pool2d_is_noop
func @max_pool2d_is_noop(%arg0: tensor<10x1x1x3xf32>) -> tensor<10x1x1x3xf32> {
func.func @max_pool2d_is_noop(%arg0: tensor<10x1x1x3xf32>) -> tensor<10x1x1x3xf32> {
// CHECK-NOT: "tosa.max_pool2d"
// CHECK: return %arg0
%0 = "tosa.max_pool2d"(%arg0) {kernel = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1]} : (tensor<10x1x1x3xf32>) -> tensor<10x1x1x3xf32>
@ -178,7 +178,7 @@ func @max_pool2d_is_noop(%arg0: tensor<10x1x1x3xf32>) -> tensor<10x1x1x3xf32> {
// -----
// CHECK-LABEL: @pad_noop
func @pad_noop(%arg0: tensor<?x?xf32>) -> tensor<?x?xf32> {
func.func @pad_noop(%arg0: tensor<?x?xf32>) -> tensor<?x?xf32> {
// CHECK: return %arg0
%0 = "tosa.const"() { value = dense<0> : tensor<2x2xi32>} : () -> tensor<2x2xi32>
%1 = "tosa.pad"(%arg0, %0) : (tensor<?x?xf32>, tensor<2x2xi32>) -> tensor<?x?xf32>
@ -188,7 +188,7 @@ func @pad_noop(%arg0: tensor<?x?xf32>) -> tensor<?x?xf32> {
// -----
// CHECK-LABEL: @pad_determine_val_i32
func @pad_determine_val_i32(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi32>) -> tensor<?x?xi32> {
func.func @pad_determine_val_i32(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi32>) -> tensor<?x?xi32> {
// CHECK: %[[ZERO:.+]] = "tosa.const"() {value = dense<0> : tensor<i32>}
// CHECK: "tosa.pad"(%arg0, %arg1, %[[ZERO]])
%0 = "tosa.const"() { value = dense<[[1, 0], [0, 1]]> : tensor<2x2xi32>} : () -> tensor<2x2xi32>
@ -199,7 +199,7 @@ func @pad_determine_val_i32(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi32>) ->
// -----
// CHECK-LABEL: @pad_determine_val_f32
func @pad_determine_val_f32(%arg0: tensor<?x?xf32>, %arg1 : tensor<2x2xi32>) -> tensor<?x?xf32> {
func.func @pad_determine_val_f32(%arg0: tensor<?x?xf32>, %arg1 : tensor<2x2xi32>) -> tensor<?x?xf32> {
// CHECK: %[[ZERO:.+]] = "tosa.const"() {value = dense<0.000000e+00> : tensor<f32>}
// CHECK: "tosa.pad"(%arg0, %arg1, %[[ZERO]])
%0 = "tosa.const"() { value = dense<[[1, 0], [0, 1]]> : tensor<2x2xi32>} : () -> tensor<2x2xi32>
@ -210,7 +210,7 @@ func @pad_determine_val_f32(%arg0: tensor<?x?xf32>, %arg1 : tensor<2x2xi32>) ->
// -----
// CHECK-LABEL: @pad_determine_val_quant
func @pad_determine_val_quant(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi32>) -> tensor<?x?xi32> {
func.func @pad_determine_val_quant(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi32>) -> tensor<?x?xi32> {
// CHECK: %[[ZERO:.+]] = "tosa.const"() {value = dense<42> : tensor<i32>}
// CHECK: "tosa.pad"(%arg0, %arg1, %[[ZERO]])
%0 = "tosa.const"() { value = dense<[[1, 0], [0, 1]]> : tensor<2x2xi32>} : () -> tensor<2x2xi32>
@ -221,7 +221,7 @@ func @pad_determine_val_quant(%arg0: tensor<?x?xi32>, %arg1 : tensor<2x2xi32>) -
// -----
// CHECK-LABEL: @mul_one_different_shape
func @mul_one_different_shape(%arg0: tensor<2x3xf32>) -> tensor<4x2x3xf32> {
func.func @mul_one_different_shape(%arg0: tensor<2x3xf32>) -> tensor<4x2x3xf32> {
// CHECK: tosa.mul
%ones = "tosa.const"() {value = dense<1.0> : tensor<4x2x3xf32>} : () -> tensor<4x2x3xf32>
%1 = "tosa.mul"(%arg0, %ones) {shift = 0 : i32} : (tensor<2x3xf32>, tensor<4x2x3xf32>) -> tensor<4x2x3xf32>
@ -231,7 +231,7 @@ func @mul_one_different_shape(%arg0: tensor<2x3xf32>) -> tensor<4x2x3xf32> {
// -----
// CHECK-LABEL: @mul_one_float
func @mul_one_float(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
func.func @mul_one_float(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
// CHECK: return %arg0
// CHECK-NOT: tosa.mul
%ones = "tosa.const"() {value = dense<1.0> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
@ -242,7 +242,7 @@ func @mul_one_float(%arg0: tensor<2x3xf32>) -> tensor<2x3xf32> {
// -----
// CHECK-LABEL: @mul_one_int
func @mul_one_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
func.func @mul_one_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
// CHECK: return %arg0
// CHECK-NOT: tosa.mul
%ones = "tosa.const"() {value = dense<1> : tensor<2x3xi32>} : () -> tensor<2x3xi32>
@ -253,7 +253,7 @@ func @mul_one_int(%arg0: tensor<2x3xi32>) -> tensor<2x3xi32> {
// -----
// CHECK-LABEL: @select_same_value
func @select_same_value(%arg0: tensor<2x3xi1>, %arg1: tensor<2x3xi32>) -> tensor<2x3xi32> {
func.func @select_same_value(%arg0: tensor<2x3xi1>, %arg1: tensor<2x3xi32>) -> tensor<2x3xi32> {
%0 = "tosa.select"(%arg0, %arg1, %arg1) : (tensor<2x3xi1>, tensor<2x3xi32>, tensor<2x3xi32>) -> tensor<2x3xi32>
// CHECK: return %arg1
// CHECK-NOT: tosa.select
@ -263,7 +263,7 @@ func @select_same_value(%arg0: tensor<2x3xi1>, %arg1: tensor<2x3xi32>) -> tensor
// -----
// CHECK-LABEL: @select_true_value
func @select_true_value(%arg0: tensor<2x3xi32>, %arg1: tensor<2x3xi32>) -> tensor<2x3xi32> {
func.func @select_true_value(%arg0: tensor<2x3xi32>, %arg1: tensor<2x3xi32>) -> tensor<2x3xi32> {
%c1 = "tosa.const"() {value = dense<1> : tensor<2x3xi1>} : () -> tensor<2x3xi1>
%0 = "tosa.select"(%c1, %arg0, %arg1) : (tensor<2x3xi1>, tensor<2x3xi32>, tensor<2x3xi32>) -> tensor<2x3xi32>
// CHECK: return %arg0
@ -274,7 +274,7 @@ func @select_true_value(%arg0: tensor<2x3xi32>, %arg1: tensor<2x3xi32>) -> tenso
// -----
// CHECK-LABEL: @select_false_value
func @select_false_value(%arg0: tensor<2x3xi32>, %arg1: tensor<2x3xi32>) -> tensor<2x3xi32> {
func.func @select_false_value(%arg0: tensor<2x3xi32>, %arg1: tensor<2x3xi32>) -> tensor<2x3xi32> {
%c0 = "tosa.const"() {value = dense<0> : tensor<2x3xi1>} : () -> tensor<2x3xi1>
%0 = "tosa.select"(%c0, %arg0, %arg1) : (tensor<2x3xi1>, tensor<2x3xi32>, tensor<2x3xi32>) -> tensor<2x3xi32>
// CHECK: return %arg1
@ -285,7 +285,7 @@ func @select_false_value(%arg0: tensor<2x3xi32>, %arg1: tensor<2x3xi32>) -> tens
// -----
// CHECK-LABEL: @select_not_pred
func @select_not_pred(%arg0: tensor<2x3xi1>, %arg1: tensor<2x3xi32>, %arg2: tensor<2x3xi32>) -> tensor<2x3xi32> {
func.func @select_not_pred(%arg0: tensor<2x3xi1>, %arg1: tensor<2x3xi32>, %arg2: tensor<2x3xi32>) -> tensor<2x3xi32> {
%0 = "tosa.logical_not"(%arg0) : (tensor<2x3xi1>) -> tensor<2x3xi1>
%1 = "tosa.select"(%0, %arg1, %arg2) : (tensor<2x3xi1>, tensor<2x3xi32>, tensor<2x3xi32>) -> tensor<2x3xi32>
// CHECK: "tosa.select"(%arg0, %arg2, %arg1)
@ -295,7 +295,7 @@ func @select_not_pred(%arg0: tensor<2x3xi1>, %arg1: tensor<2x3xi32>, %arg2: tens
// -----
// CHECK-LABEL: @reduce_all_fold
func @reduce_all_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_all_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.reduce_all"(%arg0) {axis = 1 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -304,7 +304,7 @@ func @reduce_all_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_all_nofold
func @reduce_all_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_all_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.reduce_all"
%0 = "tosa.reduce_all"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -313,7 +313,7 @@ func @reduce_all_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_any_fold
func @reduce_any_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_any_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.reduce_any"(%arg0) {axis = 1 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -322,7 +322,7 @@ func @reduce_any_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_any_nofold
func @reduce_any_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_any_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.reduce_any"
%0 = "tosa.reduce_any"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -331,7 +331,7 @@ func @reduce_any_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_max_fold
func @reduce_max_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_max_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.reduce_max"(%arg0) {axis = 1 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -340,7 +340,7 @@ func @reduce_max_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_max_nofold
func @reduce_max_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_max_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.reduce_max"
%0 = "tosa.reduce_max"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -349,7 +349,7 @@ func @reduce_max_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_min_fold
func @reduce_min_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_min_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.reduce_min"(%arg0) {axis = 1 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -358,7 +358,7 @@ func @reduce_min_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_min_nofold
func @reduce_min_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_min_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.reduce_min"
%0 = "tosa.reduce_min"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -367,7 +367,7 @@ func @reduce_min_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_prod_fold
func @reduce_prod_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_prod_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.reduce_prod"(%arg0) {axis = 1 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -376,7 +376,7 @@ func @reduce_prod_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_prod_nofold
func @reduce_prod_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_prod_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.reduce_prod"
%0 = "tosa.reduce_prod"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -385,7 +385,7 @@ func @reduce_prod_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_sum_fold
func @reduce_sum_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_sum_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: return %arg0
%0 = "tosa.reduce_sum"(%arg0) {axis = 1 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -394,7 +394,7 @@ func @reduce_sum_fold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reduce_sum_nofold
func @reduce_sum_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
func.func @reduce_sum_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// CHECK: "tosa.reduce_sum"
%0 = "tosa.reduce_sum"(%arg0) {axis = 0 : i64}: (tensor<?x1xf32>) -> tensor<?x1xf32>
return %0 : tensor<?x1xf32>
@ -403,7 +403,7 @@ func @reduce_sum_nofold(%arg0: tensor<?x1xf32>) -> tensor<?x1xf32> {
// -----
// CHECK-LABEL: @reshape_canonicalize
func @reshape_canonicalize(%arg0: tensor<?x10xf32>) -> tensor<?x10xf32> {
func.func @reshape_canonicalize(%arg0: tensor<?x10xf32>) -> tensor<?x10xf32> {
// CHECK: return %arg0
%0 = "tosa.reshape"(%arg0) {new_shape = [-1, 10]}: (tensor<?x10xf32>) -> tensor<?x10xf32>
return %0 : tensor<?x10xf32>
@ -412,7 +412,7 @@ func @reshape_canonicalize(%arg0: tensor<?x10xf32>) -> tensor<?x10xf32> {
// -----
// CHECK-LABEL: @reshape_canonicalize_double
func @reshape_canonicalize_double(%arg0: tensor<?x10xf32>) -> tensor<?x5xf32> {
func.func @reshape_canonicalize_double(%arg0: tensor<?x10xf32>) -> tensor<?x5xf32> {
// CHECK: %[[VAR0:.+]] = "tosa.reshape"(%arg0) {new_shape = [-1, 5]}
// CHECK: return %[[VAR0]]
%0 = "tosa.reshape"(%arg0) {new_shape = [5, -1]}: (tensor<?x10xf32>) -> tensor<5x?xf32>
@ -423,7 +423,7 @@ func @reshape_canonicalize_double(%arg0: tensor<?x10xf32>) -> tensor<?x5xf32> {
// -----
// CHECK-LABEL: @reshape_canonicalize_const
func @reshape_canonicalize_const() -> tensor<1x10xi32> {
func.func @reshape_canonicalize_const() -> tensor<1x10xi32> {
// CHECK: %[[VAR0:.+]] = "tosa.const"() {value = dense<0> : tensor<1x10xi32>}
// CHECK: return %[[VAR0]]
%0 = "tosa.const"() {value = dense<0> : tensor<10xi32>} : () -> tensor<10xi32>
@ -434,7 +434,7 @@ func @reshape_canonicalize_const() -> tensor<1x10xi32> {
// -----
// CHECK-LABEL: @reshape_canonicalize_const_spat
func @reshape_canonicalize_const_spat() -> (tensor<10xi32>, tensor<1x10xi32>) {
func.func @reshape_canonicalize_const_spat() -> (tensor<10xi32>, tensor<1x10xi32>) {
// CHECK-DAG: %[[VAR0:.+]] = "tosa.const"() {value = dense<0> : tensor<10xi32>}
// CHECK-DAG: %[[VAR1:.+]] = "tosa.const"() {value = dense<0> : tensor<1x10xi32>}
// CHECK: return %[[VAR0]], %[[VAR1]]
@ -446,7 +446,7 @@ func @reshape_canonicalize_const_spat() -> (tensor<10xi32>, tensor<1x10xi32>) {
// -----
// CHECK-LABEL: @reshape_canonicalize_const_sparse
func @reshape_canonicalize_const_sparse() -> (tensor<3xi32>, tensor<1x3xi32>) {
func.func @reshape_canonicalize_const_sparse() -> (tensor<3xi32>, tensor<1x3xi32>) {
//CHECK: "tosa.reshape"
%0 = "tosa.const"() {value = dense<[1, 2, 3]> : tensor<3xi32>} : ()-> tensor<3xi32>
%1 = "tosa.reshape"(%0) {new_shape = [1, 3]} : (tensor<3xi32>) -> tensor<1x3xi32>
@ -456,7 +456,7 @@ func @reshape_canonicalize_const_sparse() -> (tensor<3xi32>, tensor<1x3xi32>) {
// -----
// CHECK-LABEL: @slice_fold
func @slice_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
func.func @slice_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
// CHECK: return %arg0
%0 = "tosa.slice"(%arg0) { size = [3, 4], start = [0, 0]}: (tensor<3x4xf32>) -> tensor<3x4xf32>
return %0 : tensor<3x4xf32>
@ -465,7 +465,7 @@ func @slice_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
// -----
// CHECK-LABEL: @slice_nofold
func @slice_nofold(%arg0: tensor<?x4xf32>) -> tensor<?x4xf32> {
func.func @slice_nofold(%arg0: tensor<?x4xf32>) -> tensor<?x4xf32> {
// CHECK: "tosa.slice"
%0 = "tosa.slice"(%arg0) { size = [3, 4], start = [0, 0]}: (tensor<?x4xf32>) -> tensor<?x4xf32>
return %0 : tensor<?x4xf32>
@ -474,7 +474,7 @@ func @slice_nofold(%arg0: tensor<?x4xf32>) -> tensor<?x4xf32> {
// -----
// CHECK-LABEL: @tile_fold
func @tile_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
func.func @tile_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
// CHECK: return %arg0
%0 = "tosa.tile"(%arg0) { multiples = [1, 1] }: (tensor<3x4xf32>) -> tensor<3x4xf32>
return %0 : tensor<3x4xf32>
@ -483,7 +483,7 @@ func @tile_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
// -----
// CHECK-LABEL: @tile_nofold
func @tile_nofold(%arg0: tensor<3x4xf32>) -> tensor<3x8xf32> {
func.func @tile_nofold(%arg0: tensor<3x4xf32>) -> tensor<3x8xf32> {
// CHECK: "tosa.tile"
%0 = "tosa.tile"(%arg0) { multiples = [1, 2] }: (tensor<3x4xf32>) -> tensor<3x8xf32>
return %0 : tensor<3x8xf32>
@ -492,7 +492,7 @@ func @tile_nofold(%arg0: tensor<3x4xf32>) -> tensor<3x8xf32> {
// -----
// CHECK-LABEL: @transpose_fold
func @transpose_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
func.func @transpose_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
// CHECK: return %arg0
%0 = arith.constant dense<[0, 1]> : tensor<2xi32>
%1 = "tosa.transpose"(%arg0, %0) { perms = [1, 0] }: (tensor<3x4xf32>, tensor<2xi32>) -> tensor<3x4xf32>
@ -502,7 +502,7 @@ func @transpose_fold(%arg0: tensor<3x4xf32>) -> tensor<3x4xf32> {
// -----
// CHECK-LABEL: @transpose_nofold
func @transpose_nofold(%arg0: tensor<3x3xf32>) -> tensor<3x3xf32> {
func.func @transpose_nofold(%arg0: tensor<3x3xf32>) -> tensor<3x3xf32> {
// CHECK: "tosa.transpose"
%0 = arith.constant dense<[1, 0]> : tensor<2xi32>
%1 = "tosa.transpose"(%arg0, %0) { perms = [1, 0] }: (tensor<3x3xf32>, tensor<2xi32>) -> tensor<3x3xf32>
@ -512,7 +512,7 @@ func @transpose_nofold(%arg0: tensor<3x3xf32>) -> tensor<3x3xf32> {
// -----
// CHECK-LABEL: @transpose_nofold_shape
func @transpose_nofold_shape(%arg0: tensor<3x4xf32>) -> tensor<?x?xf32> {
func.func @transpose_nofold_shape(%arg0: tensor<3x4xf32>) -> tensor<?x?xf32> {
// CHECK: "tosa.transpose"
%0 = arith.constant dense<[1, 0]> : tensor<2xi32>
%1 = "tosa.transpose"(%arg0, %0) { perms = [1, 0] }: (tensor<3x4xf32>, tensor<2xi32>) -> tensor<?x?xf32>
@ -522,7 +522,7 @@ func @transpose_nofold_shape(%arg0: tensor<3x4xf32>) -> tensor<?x?xf32> {
// -----
// CHECK-LABEL: @transpose_fold_splat
func @transpose_fold_splat() -> tensor<3x2xf32> {
func.func @transpose_fold_splat() -> tensor<3x2xf32> {
%input = "tosa.const"() {value = dense<4.0> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
%perms = "tosa.const"() {value = dense<[1, 0]> : tensor<2xi32>} : () -> tensor<2xi32>
// CHECK: %[[CST:.+]] = "tosa.const"()
@ -535,7 +535,7 @@ func @transpose_fold_splat() -> tensor<3x2xf32> {
// -----
// CHECK-LABEL: @transpose_fold_2d_float
func @transpose_fold_2d_float() -> tensor<3x2xf32> {
func.func @transpose_fold_2d_float() -> tensor<3x2xf32> {
%input = "tosa.const"() {value = dense<[[0.0, 1.0, 2.0], [3.0, 4.0, 5.0]]> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
%perms = "tosa.const"() {value = dense<[1, 0]> : tensor<2xi32>} : () -> tensor<2xi32>
// CHECK: %[[CST:.+]] = "tosa.const"()
@ -548,7 +548,7 @@ func @transpose_fold_2d_float() -> tensor<3x2xf32> {
// -----
// CHECK-LABEL: @transpose_fold_4d_int
func @transpose_fold_4d_int() -> tensor<3x1x4x2xi32> {
func.func @transpose_fold_4d_int() -> tensor<3x1x4x2xi32> {
%input = "tosa.const"() {value = dense<[[
[[ 0, 1, 2, 3], [ 4, 5, 6, 7], [ 8, 9, 10, 11]],
[[12, 13, 14, 15], [16, 17, 18, 19], [20, 21, 22, 23]]
@ -568,7 +568,7 @@ func @transpose_fold_4d_int() -> tensor<3x1x4x2xi32> {
// -----
// CHECK-LABEL: @transpose_nofold_non_cst_input
func @transpose_nofold_non_cst_input(%input: tensor<2x3xf32>) -> tensor<3x2xf32> {
func.func @transpose_nofold_non_cst_input(%input: tensor<2x3xf32>) -> tensor<3x2xf32> {
%perms = "tosa.const"() {value = dense<[1, 0]> : tensor<2xi32>} : () -> tensor<2xi32>
// CHECK: tosa.transpose
%1 = "tosa.transpose"(%input, %perms) : (tensor<2x3xf32>, tensor<2xi32>) -> tensor<3x2xf32>
@ -578,7 +578,7 @@ func @transpose_nofold_non_cst_input(%input: tensor<2x3xf32>) -> tensor<3x2xf32>
// -----
// CHECK-LABEL: @transpose_nofold_non_cst_perms
func @transpose_nofold_non_cst_perms(%perms: tensor<2xi32>) -> tensor<3x2xf32> {
func.func @transpose_nofold_non_cst_perms(%perms: tensor<2xi32>) -> tensor<3x2xf32> {
%input = "tosa.const"() {value = dense<[[0.0, 1.0, 2.0], [3.0, 4.0, 5.0]]> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
// CHECK: tosa.transpose
%1 = "tosa.transpose"(%input, %perms) : (tensor<2x3xf32>, tensor<2xi32>) -> tensor<3x2xf32>
@ -588,7 +588,7 @@ func @transpose_nofold_non_cst_perms(%perms: tensor<2xi32>) -> tensor<3x2xf32> {
// -----
// CHECK-LABEL: @transpose_nofold_multi_users
func @transpose_nofold_multi_users() -> (tensor<3x2xf32>, tensor<2x3xf32>) {
func.func @transpose_nofold_multi_users() -> (tensor<3x2xf32>, tensor<2x3xf32>) {
%input = "tosa.const"() {value = dense<[[0.0, 1.0, 2.0], [3.0, 4.0, 5.0]]> : tensor<2x3xf32>} : () -> tensor<2x3xf32>
%perms = "tosa.const"() {value = dense<[1, 0]> : tensor<2xi32>} : () -> tensor<2xi32>
// CHECK: tosa.transpose
@ -599,7 +599,7 @@ func @transpose_nofold_multi_users() -> (tensor<3x2xf32>, tensor<2x3xf32>) {
// -----
// CHECK-LABEL: @transpose_nofold_quantized_types
func @transpose_nofold_quantized_types() -> tensor<1x1x16x1x!quant.uniform<i8<-127:127>:f32:3, {1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,2.100000e+00,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01}>> {
func.func @transpose_nofold_quantized_types() -> tensor<1x1x16x1x!quant.uniform<i8<-127:127>:f32:3, {1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,2.100000e+00,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01,1.000000e-01}>> {
%perms = "tosa.const"() {value = dense<[1, 2, 3, 0]> : tensor<4xi32>} : () -> tensor<4xi32>
%input = "tosa.const"() {value = dense<[[[[-127, 127, 127, -127, -127, -127, -127, -127, -127, 127, 127, 127, 127, 127, -127, 127]]]]> : tensor<1x1x1x16xi8>} : () -> tensor<1x1x1x16xi8>
// CHECK: tosa.transpose
@ -610,7 +610,7 @@ func @transpose_nofold_quantized_types() -> tensor<1x1x16x1x!quant.uniform<i8<-1
// -----
// CHECK-LABEL: @transpose_no_op
func @transpose_no_op(%arg0: tensor<3x4x5x6xf32>) -> tensor<3x4x5x6xf32> {
func.func @transpose_no_op(%arg0: tensor<3x4x5x6xf32>) -> tensor<3x4x5x6xf32> {
// CHECK: return %arg0
// CHECK-NOT: tosa.transpose
%perms = "tosa.const"() {value = dense<[0, 1, 2, 3]> : tensor<4xi32>} : () -> tensor<4xi32>

2
mlir/test/Dialect/Tosa/constant_folding.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt --test-constant-fold %s | FileCheck %s
// CHECK-LABEL: func @test_const
func @test_const(%arg0 : index) -> tensor<4xi32> {
func.func @test_const(%arg0 : index) -> tensor<4xi32> {
// CHECK: "tosa.const"
%0 = "tosa.const"() {value = dense<[3, 0, 1, 2]> : tensor<4xi32>} : () -> tensor<4xi32>
return %0 : tensor<4xi32>

2
mlir/test/Dialect/Tosa/constrained_shapes.mlir
View File

@ -5,7 +5,7 @@
// -----
// Uses argmax as canonical example to validate constrained TOSA tensor shapes.
// CHECK-LABEL: argmax
func @test_argmax(%arg0: tensor<?xf32>) -> tensor<?xi32> {
func.func @test_argmax(%arg0: tensor<?xf32>) -> tensor<?xi32> {
%0 = "tosa.argmax"(%arg0) {axis = 1 : i64} : (tensor<?xf32>) -> tensor<?xi32>
return %0 : tensor<?xi32>
}

12
mlir/test/Dialect/Tosa/inlining.mlir
View File

@ -7,7 +7,7 @@
// Check that both the calls and the functions are eliminated after inlining:
// CHECK-NOT: @add
// CHECK-NOT: @sub
func @inlined_if_fn(%arg0: tensor<f32>, %arg1: tensor<f32>, %arg2: tensor<i1>) -> tensor<f32> {
func.func @inlined_if_fn(%arg0: tensor<f32>, %arg1: tensor<f32>, %arg2: tensor<i1>) -> tensor<f32> {
%0 = "tosa.cond_if"(%arg2, %arg0, %arg1) ({
^bb0(%arg3: tensor<f32>, %arg4: tensor<f32>):
%1 = call @add(%arg3, %arg4) : (tensor<f32>, tensor<f32>) -> tensor<f32>
@ -19,11 +19,11 @@ func @inlined_if_fn(%arg0: tensor<f32>, %arg1: tensor<f32>, %arg2: tensor<i1>) -
}) : (tensor<i1>, tensor<f32>, tensor<f32>) -> tensor<f32>
return %0 : tensor<f32>
}
func private @add(%arg0: tensor<f32>, %arg1: tensor<f32>) -> tensor<f32> {
func.func private @add(%arg0: tensor<f32>, %arg1: tensor<f32>) -> tensor<f32> {
%0 = "tosa.add"(%arg0, %arg1) : (tensor<f32>, tensor<f32>) -> tensor<f32>
return %0 : tensor<f32>
}
func private @sub(%arg0: tensor<f32>, %arg1: tensor<f32>) -> tensor<f32> {
func.func private @sub(%arg0: tensor<f32>, %arg1: tensor<f32>) -> tensor<f32> {
%0 = "tosa.sub"(%arg0, %arg1) : (tensor<f32>, tensor<f32>) -> tensor<f32>
return %0 : tensor<f32>
}
@ -31,7 +31,7 @@ func private @sub(%arg0: tensor<f32>, %arg1: tensor<f32>) -> tensor<f32> {
// -----
// CHECK-LABEL: func @inlined_while_fn
func @inlined_while_fn(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<10xi32>) -> tensor<10xi32> {
func.func @inlined_while_fn(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<10xi32>) -> tensor<10xi32> {
// Check that calls are inlined and functions eliminated:
// CHECK-NOT: @while
%1:4 = "tosa.while_loop"(%arg0, %arg1, %arg2, %arg3) ({
@ -45,12 +45,12 @@ func @inlined_while_fn(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32
}) : (tensor<i32>, tensor<i32>, tensor<i32>, tensor<10xi32>) -> (tensor<i32>, tensor<i32>, tensor<i32>, tensor<10xi32>)
return %1#3 : tensor<10xi32>
}
func private @while_body_50(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<10xi32>) -> (tensor<i32>, tensor<i32>, tensor<i32>, tensor<10xi32>) {
func.func private @while_body_50(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<10xi32>) -> (tensor<i32>, tensor<i32>, tensor<i32>, tensor<10xi32>) {
%1 = "tosa.add"(%arg0, %arg1) : (tensor<i32>, tensor<i32>) -> tensor<i32>
%2 = "tosa.add"(%arg3, %1) : (tensor<10xi32>, tensor<i32>) -> tensor<10xi32>
return %1, %arg1, %arg2, %2: tensor<i32>, tensor<i32>, tensor<i32>, tensor<10xi32>
}
func private @while_cond_40(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<10xi32>) -> tensor<i1> {
func.func private @while_cond_40(%arg0: tensor<i32>, %arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<10xi32>) -> tensor<i1> {
%0 = "tosa.greater_equal"(%arg0, %arg1) : (tensor<i32>, tensor<i32>) -> tensor<i1>
%1 = "tosa.logical_not"(%0) : (tensor<i1>) -> tensor<i1>
return %1 : tensor<i1>

8
mlir/test/Dialect/Tosa/invalid.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt %s -split-input-file -verify-diagnostics
func @test_conv2d(%arg0: tensor<1x29x29x4xf32>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
func.func @test_conv2d(%arg0: tensor<1x29x29x4xf32>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
// expected-error@+1 {{expect both input and weight to be float or not together, got 'f32' and 'i8'}}
%0 = "tosa.conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]}
: (tensor<1x29x29x4xf32>, tensor<16x3x3x4xi8>, tensor<16xi8>) -> tensor<1x27x27x16xi8>
@ -10,7 +10,7 @@ func @test_conv2d(%arg0: tensor<1x29x29x4xf32>, %arg1: tensor<16x3x3x4xi8>, %arg
// -----
func @test_conv2d(%arg0: tensor<*xi8>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
func.func @test_conv2d(%arg0: tensor<*xi8>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
// expected-error@+1 {{expect a ranked tensor for input, got <block argument> of type 'tensor<*xi8>' at index: 0}}
%0 = "tosa.conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]}
: (tensor<*xi8>, tensor<16x3x3x4xi8>, tensor<16xi8>) -> tensor<1x27x27x16xi8>
@ -19,7 +19,7 @@ func @test_conv2d(%arg0: tensor<*xi8>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor
// -----
func @test_conv2d(%arg0: tensor<1x29x29x4xi8>, %arg1: tensor<*xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
func.func @test_conv2d(%arg0: tensor<1x29x29x4xi8>, %arg1: tensor<*xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
// expected-error@+1 {{expect a ranked tensor for weight, got <block argument> of type 'tensor<*xi8>' at index: 1}}
%0 = "tosa.conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]}
: (tensor<1x29x29x4xi8>, tensor<*xi8>, tensor<16xi8>) -> tensor<1x27x27x16xi8>
@ -29,7 +29,7 @@ func @test_conv2d(%arg0: tensor<1x29x29x4xi8>, %arg1: tensor<*xi8>, %arg2: tenso
// -----
func @test_conv2d(%arg0: tensor<1x29x29x4xi8>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
func.func @test_conv2d(%arg0: tensor<1x29x29x4xi8>, %arg1: tensor<16x3x3x4xi8>, %arg2: tensor<16xi8>) -> tensor<1x27x27x16xi8> {
// expected-error@+1 {{'tosa.conv2d' op quantizationattr is required for quantized type, and not allowed for float type}}
%0 = "tosa.conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]}
: (tensor<1x29x29x4xi8>, tensor<16x3x3x4xi8>, tensor<16xi8>) -> tensor<1x27x27x16xi8>

144
mlir/test/Dialect/Tosa/ops.mlir
View File

@ -4,91 +4,91 @@
// -----
// CHECK-LABEL: argmax
func @test_argmax(%arg0: tensor<14x19xf32>) -> tensor<14xi32> {
func.func @test_argmax(%arg0: tensor<14x19xf32>) -> tensor<14xi32> {
%0 = "tosa.argmax"(%arg0) {axis = 1 : i64} : (tensor<14x19xf32>) -> tensor<14xi32>
return %0 : tensor<14xi32>
}
// -----
// CHECK-LABEL: avg_pool2d_f32
func @test_avg_pool2d_f32(%arg0: tensor<1x7x7x9xf32>) -> tensor<1x7x7x9xf32> {
func.func @test_avg_pool2d_f32(%arg0: tensor<1x7x7x9xf32>) -> tensor<1x7x7x9xf32> {
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [2, 2], pad = [0, 1, 0, 1], stride = [1, 1]} : (tensor<1x7x7x9xf32>) -> tensor<1x7x7x9xf32>
return %0 : tensor<1x7x7x9xf32>
}
// -----
// CHECK-LABEL: avg_pool2d_i8
func @test_avg_pool2d_i8(%arg0: tensor<1x7x7x9xi8>) -> tensor<1x7x7x9xi8> {
func.func @test_avg_pool2d_i8(%arg0: tensor<1x7x7x9xi8>) -> tensor<1x7x7x9xi8> {
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [2, 2], pad = [0, 1, 0, 1], stride = [1, 1]} : (tensor<1x7x7x9xi8>) -> tensor<1x7x7x9xi8>
return %0 : tensor<1x7x7x9xi8>
}
// -----
// CHECK-LABEL: avg_pool2d_i16
func @test_avg_pool2d_i16(%arg0: tensor<1x7x7x9xi16>) -> tensor<1x7x7x9xi16> {
func.func @test_avg_pool2d_i16(%arg0: tensor<1x7x7x9xi16>) -> tensor<1x7x7x9xi16> {
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [2, 2], pad = [0, 1, 0, 1], stride = [1, 1]} : (tensor<1x7x7x9xi16>) -> tensor<1x7x7x9xi16>
return %0 : tensor<1x7x7x9xi16>
}
// -----
// CHECK-LABEL: avg_pool2d_q8
func @test_avg_pool2d_q8(%arg0: tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>>) -> tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>> {
func.func @test_avg_pool2d_q8(%arg0: tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>>) -> tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>> {
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [2, 2], pad = [0, 1, 0, 1], stride = [1, 1]} : (tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>>) -> tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>>
return %0 : tensor<1x7x7x9x!quant.uniform<i8:f32, 0.01>>
}
// -----
// CHECK-LABEL: conv2d
func @test_conv2d(%arg0: tensor<1x4x4x4xf32>, %arg1: tensor<8x1x1x4xf32>, %arg2: tensor<8xf32>) -> tensor<1x4x4x8xf32> {
func.func @test_conv2d(%arg0: tensor<1x4x4x4xf32>, %arg1: tensor<8x1x1x4xf32>, %arg2: tensor<8xf32>) -> tensor<1x4x4x8xf32> {
%0 = "tosa.conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<1x4x4x4xf32>, tensor<8x1x1x4xf32>, tensor<8xf32>) -> tensor<1x4x4x8xf32>
return %0 : tensor<1x4x4x8xf32>
}
// -----
// CHECK-LABEL: depthwise_conv2d
func @test_depthwise_conv2d(%arg0: tensor<1x4x4x4xf32>, %arg1: tensor<1x1x4x2xf32>, %arg2: tensor<8xf32>) -> tensor<1x4x4x8xf32> {
func.func @test_depthwise_conv2d(%arg0: tensor<1x4x4x4xf32>, %arg1: tensor<1x1x4x2xf32>, %arg2: tensor<8xf32>) -> tensor<1x4x4x8xf32> {
%2 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<1x4x4x4xf32>, tensor<1x1x4x2xf32>, tensor<8xf32>) -> tensor<1x4x4x8xf32>
return %2 : tensor<1x4x4x8xf32>
}
// -----
// CHECK-LABEL: fully_connected
func @test_fully_connected(%arg0: tensor<14x19xf32>, %arg1: tensor<19x28xf32>, %arg2: tensor<28xf32>) -> tensor<14x28xf32> {
func.func @test_fully_connected(%arg0: tensor<14x19xf32>, %arg1: tensor<19x28xf32>, %arg2: tensor<28xf32>) -> tensor<14x28xf32> {
%0 = "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<14x19xf32>, tensor<19x28xf32>, tensor<28xf32>) -> tensor<14x28xf32>
return %0 : tensor<14x28xf32>
}
// -----
// CHECK-LABEL: test_matmul
func @test_matmul(%arg0: tensor<1x14x19xf32>, %arg1: tensor<1x19x28xf32>) -> tensor<1x14x28xf32> {
func.func @test_matmul(%arg0: tensor<1x14x19xf32>, %arg1: tensor<1x19x28xf32>) -> tensor<1x14x28xf32> {
%0 = "tosa.matmul"(%arg0, %arg1) : (tensor<1x14x19xf32>, tensor<1x19x28xf32>) -> tensor<1x14x28xf32>
return %0 : tensor<1x14x28xf32>
}
// -----
// CHECK-LABEL: max_pool2d
func @test_max_pool2d(%arg0: tensor<1x32x32x8xf32>) -> tensor<1x32x32x8xf32> {
func.func @test_max_pool2d(%arg0: tensor<1x32x32x8xf32>) -> tensor<1x32x32x8xf32> {
%0 = "tosa.max_pool2d"(%arg0) {kernel = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<1x32x32x8xf32>) -> tensor<1x32x32x8xf32>
return %0 : tensor<1x32x32x8xf32>
}
// -----
/// CHECK-LABEL: transpose_conv2d
func @test_transpose_conv2d(%arg0: tensor<1x32x32x8xf32>, %arg1: tensor<16x1x1x8xf32>, %arg2: tensor<16xf32>) -> tensor<1x32x32x16xf32> {
func.func @test_transpose_conv2d(%arg0: tensor<1x32x32x8xf32>, %arg1: tensor<16x1x1x8xf32>, %arg2: tensor<16xf32>) -> tensor<1x32x32x16xf32> {
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [1, 32, 32, 16], stride = [1, 1]} : (tensor<1x32x32x8xf32>, tensor<16x1x1x8xf32>, tensor<16xf32>) -> tensor<1x32x32x16xf32>
return %0 : tensor<1x32x32x16xf32>
}
// -----
// CHECK-LABEL: clamp
func @test_clamp(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_clamp(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.clamp"(%arg0) {min_fp = 0.0 : f32, max_fp = 1.0: f32, min_int = 0 : i64, max_int = 1 : i64} : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: relu
func @test_relu(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_relu(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.reluN"(%arg0) {max_fp = 3.40282347E+38 : f32, max_int = 0 : i64} : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
@ -96,217 +96,217 @@ func @test_relu(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
// -----
// CHECK-LABEL: sigmoid
func @test_sigmoid(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_sigmoid(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.sigmoid"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: tanh
func @test_tanh(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_tanh(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.tanh"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: add
func @test_add(%arg0: tensor<13x21x1xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_add(%arg0: tensor<13x21x1xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.add"(%arg0, %arg1) : (tensor<13x21x1xf32>, tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: arithmetic_right_shift
func @test_arithmetic_right_shift(%arg0: tensor<13x21x1xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_arithmetic_right_shift(%arg0: tensor<13x21x1xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.arithmetic_right_shift"(%arg0, %arg1) { round = false } : (tensor<13x21x1xf32>, tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: bitwise_and
func @test_bitwise_and(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x21x1xi32>) -> tensor<13x21x3xi32> {
func.func @test_bitwise_and(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x21x1xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.bitwise_and"(%arg0, %arg1) : (tensor<13x21x3xi32>, tensor<13x21x1xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: bitwise_or
func @test_bitwise_or(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x1x3xi32>) -> tensor<13x21x3xi32> {
func.func @test_bitwise_or(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x1x3xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.bitwise_or"(%arg0, %arg1) : (tensor<13x21x3xi32>, tensor<13x1x3xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: bitwise_xor
func @test_bitwise_xor(%arg0: tensor<13x21x1xi32>, %arg1: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
func.func @test_bitwise_xor(%arg0: tensor<13x21x1xi32>, %arg1: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.bitwise_xor"(%arg0, %arg1) : (tensor<13x21x1xi32>, tensor<13x21x3xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: div
func @test_div(%arg0: tensor<13x21x1xi32>, %arg1: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
func.func @test_div(%arg0: tensor<13x21x1xi32>, %arg1: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.div"(%arg0, %arg1) : (tensor<13x21x1xi32>, tensor<13x21x3xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: logical_and
func @test_logical_and(%arg0: tensor<13x21x3xi1>, %arg1: tensor<13x21x1xi1>) -> tensor<13x21x3xi1> {
func.func @test_logical_and(%arg0: tensor<13x21x3xi1>, %arg1: tensor<13x21x1xi1>) -> tensor<13x21x3xi1> {
%0 = "tosa.logical_and"(%arg0, %arg1) : (tensor<13x21x3xi1>, tensor<13x21x1xi1>) -> tensor<13x21x3xi1>
return %0 : tensor<13x21x3xi1>
}
// -----
// CHECK-LABEL: logical_left_shift
func @test_logical_left_shift(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x21x1xi32>) -> tensor<13x21x3xi32> {
func.func @test_logical_left_shift(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x21x1xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.logical_left_shift"(%arg0, %arg1) : (tensor<13x21x3xi32>, tensor<13x21x1xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: logical_right_shift
func @test_logical_right_shift(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x21x1xi32>) -> tensor<13x21x3xi32> {
func.func @test_logical_right_shift(%arg0: tensor<13x21x3xi32>, %arg1: tensor<13x21x1xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.logical_right_shift"(%arg0, %arg1) : (tensor<13x21x3xi32>, tensor<13x21x1xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: logical_or
func @test_logical_or(%arg0: tensor<13x1x3xi1>, %arg1: tensor<13x21x3xi1>) -> tensor<13x21x3xi1> {
func.func @test_logical_or(%arg0: tensor<13x1x3xi1>, %arg1: tensor<13x21x3xi1>) -> tensor<13x21x3xi1> {
%0 = "tosa.logical_or"(%arg0, %arg1) : (tensor<13x1x3xi1>, tensor<13x21x3xi1>) -> tensor<13x21x3xi1>
return %0 : tensor<13x21x3xi1>
}
// -----
// CHECK-LABEL: logical_xor
func @test_logical_xor(%arg0: tensor<13x1x3xi1>, %arg1: tensor<13x21x3xi1>) -> tensor<13x21x3xi1> {
func.func @test_logical_xor(%arg0: tensor<13x1x3xi1>, %arg1: tensor<13x21x3xi1>) -> tensor<13x21x3xi1> {
%0 = "tosa.logical_xor"(%arg0, %arg1) : (tensor<13x1x3xi1>, tensor<13x21x3xi1>) -> tensor<13x21x3xi1>
return %0 : tensor<13x21x3xi1>
}
// -----
// CHECK-LABEL: maximum
func @test_max(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x21x1xf32>) -> tensor<13x21x3xf32> {
func.func @test_max(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x21x1xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.maximum"(%arg0, %arg1) : (tensor<13x21x3xf32>, tensor<13x21x1xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: minimum
func @test_min(%arg0: tensor<13x21x3xf32>, %arg1: tensor<1x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_min(%arg0: tensor<13x21x3xf32>, %arg1: tensor<1x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.minimum"(%arg0, %arg1) : (tensor<13x21x3xf32>, tensor<1x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: mul
func @test_mul(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x1x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_mul(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x1x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.mul"(%arg0, %arg1) { shift = 1 : i32 } : (tensor<13x21x3xf32>, tensor<13x1x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: pow
func @test_pow(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x21x1xf32>) -> tensor<13x21x3xf32> {
func.func @test_pow(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x21x1xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.pow"(%arg0, %arg1) : (tensor<13x21x3xf32>, tensor<13x21x1xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: sub
func @test_sub(%arg0: tensor<1x21x3xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_sub(%arg0: tensor<1x21x3xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.sub"(%arg0, %arg1) : (tensor<1x21x3xf32>, tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: table
func @main(%arg0: tensor<64xi32>, %arg1: tensor<513x!quant.uniform<i16:f32, 1.0:0>>) -> tensor<64x!quant.uniform<i16:f32, 1.0:0>> {
func.func @main(%arg0: tensor<64xi32>, %arg1: tensor<513x!quant.uniform<i16:f32, 1.0:0>>) -> tensor<64x!quant.uniform<i16:f32, 1.0:0>> {
%0 = "tosa.table"(%arg0, %arg1) : (tensor<64xi32>, tensor<513x!quant.uniform<i16:f32, 1.0:0>>) -> tensor<64x!quant.uniform<i16:f32, 1.0:0>>
return %0 : tensor<64x!quant.uniform<i16:f32, 1.0:0>>
}
// -----
// CHECK-LABEL: abs
func @test_abs(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_abs(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.abs"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: bitwise_not
func @test_bitwise_not(%arg0: tensor<13x21x1xi32>) -> tensor<13x21x1xi32> {
func.func @test_bitwise_not(%arg0: tensor<13x21x1xi32>) -> tensor<13x21x1xi32> {
%0 = "tosa.bitwise_not"(%arg0) : (tensor<13x21x1xi32>) -> tensor<13x21x1xi32>
return %0 : tensor<13x21x1xi32>
}
// -----
// CHECK-LABEL: ceil
func @test_ceil(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_ceil(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.ceil"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: clz
func @test_clz(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
func.func @test_clz(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.clz"(%arg0) : (tensor<13x21x3xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: exp
func @test_exp(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_exp(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.exp"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: floor
func @test_floor(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_floor(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.floor"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: log
func @test_log(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_log(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.log"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: logical_not
func @test_logical_not(%arg0: tensor<1x21x3xi1>) -> tensor<1x21x3xi1> {
func.func @test_logical_not(%arg0: tensor<1x21x3xi1>) -> tensor<1x21x3xi1> {
%0 = "tosa.logical_not"(%arg0) : (tensor<1x21x3xi1>) -> tensor<1x21x3xi1>
return %0 : tensor<1x21x3xi1>
}
// -----
// CHECK-LABEL: negate
func @test_negate(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_negate(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.negate"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: reciprocal
func @test_reciprocal(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_reciprocal(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.reciprocal"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: rsqrt
func @test_rsqrt(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_rsqrt(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.rsqrt"(%arg0) : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: select
func @test_select(%arg0: tensor<1x1x1xi1>, %arg1: tensor<13x21x3xf32>, %arg2: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_select(%arg0: tensor<1x1x1xi1>, %arg1: tensor<13x21x3xf32>, %arg2: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.select"(%arg0, %arg1, %arg2) : (tensor<1x1x1xi1>, tensor<13x21x3xf32>, tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
@ -314,28 +314,28 @@ func @test_select(%arg0: tensor<1x1x1xi1>, %arg1: tensor<13x21x3xf32>, %arg2: te
// -----
// CHECK-LABEL: equal
func @test_equal(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x1x3xf32>) -> tensor<13x21x3xi1> {
func.func @test_equal(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x1x3xf32>) -> tensor<13x21x3xi1> {
%0 = "tosa.equal"(%arg0, %arg1) : (tensor<13x21x3xf32>, tensor<13x1x3xf32>) -> tensor<13x21x3xi1>
return %0 : tensor<13x21x3xi1>
}
// -----
// CHECK-LABEL: greater
func @test_greater(%arg0: tensor<13x21x1xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xi1> {
func.func @test_greater(%arg0: tensor<13x21x1xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xi1> {
%0 = "tosa.greater"(%arg0, %arg1) : (tensor<13x21x1xf32>, tensor<13x21x3xf32>) -> tensor<13x21x3xi1>
return %0 : tensor<13x21x3xi1>
}
// -----
// CHECK-LABEL: greater_equal
func @test_greater_equal(%arg0: tensor<13x1x3xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xi1> {
func.func @test_greater_equal(%arg0: tensor<13x1x3xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<13x21x3xi1> {
%0 = "tosa.greater_equal"(%arg0, %arg1) : (tensor<13x1x3xf32>, tensor<13x21x3xf32>) -> tensor<13x21x3xi1>
return %0 : tensor<13x21x3xi1>
}
// -----
// CHECK-LABEL: reduce_all
func @test_reduce_all(%arg0: tensor<13x21x3xi1>) -> tensor<21x3xi1> {
func.func @test_reduce_all(%arg0: tensor<13x21x3xi1>) -> tensor<21x3xi1> {
%0 = "tosa.reduce_all"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xi1>) -> tensor<1x21x3xi1>
%1 = "tosa.reshape"(%0) {new_shape = [21, 3]} : (tensor<1x21x3xi1>) -> tensor<21x3xi1>
return %1 : tensor<21x3xi1>
@ -343,7 +343,7 @@ func @test_reduce_all(%arg0: tensor<13x21x3xi1>) -> tensor<21x3xi1> {
// -----
// CHECK-LABEL: reduce_any
func @test_reduce_any(%arg0: tensor<13x21x3xi1>) -> tensor<21x3xi1> {
func.func @test_reduce_any(%arg0: tensor<13x21x3xi1>) -> tensor<21x3xi1> {
%0 = "tosa.reduce_any"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xi1>) -> tensor<1x21x3xi1>
%1 = "tosa.reshape"(%0) {new_shape = [21, 3]} : (tensor<1x21x3xi1>) -> tensor<21x3xi1>
return %1 : tensor<21x3xi1>
@ -351,7 +351,7 @@ func @test_reduce_any(%arg0: tensor<13x21x3xi1>) -> tensor<21x3xi1> {
// -----
// CHECK-LABEL: reduce_max
func @test_reduce_max(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
func.func @test_reduce_max(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
%0 = "tosa.reduce_max"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xf32>) -> tensor<1x21x3xf32>
%1 = "tosa.reshape"(%0) {new_shape = [21, 3]} : (tensor<1x21x3xf32>) -> tensor<21x3xf32>
return %1 : tensor<21x3xf32>
@ -359,7 +359,7 @@ func @test_reduce_max(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
// -----
// CHECK-LABEL: reduce_min
func @test_reduce_min(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
func.func @test_reduce_min(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
%0 = "tosa.reduce_min"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xf32>) -> tensor<1x21x3xf32>
%1 = "tosa.reshape"(%0) {new_shape = [21, 3]} : (tensor<1x21x3xf32>) -> tensor<21x3xf32>
return %1 : tensor<21x3xf32>
@ -367,7 +367,7 @@ func @test_reduce_min(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
// -----
// CHECK-LABEL: reduce_product
func @test_reduce_product(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
func.func @test_reduce_product(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
%0 = "tosa.reduce_prod"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xf32>) -> tensor<1x21x3xf32>
%1 = "tosa.reshape"(%0) {new_shape = [21, 3]} : (tensor<1x21x3xf32>) -> tensor<21x3xf32>
return %1 : tensor<21x3xf32>
@ -375,7 +375,7 @@ func @test_reduce_product(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
// -----
// CHECK-LABEL: reduce_sum
func @test_reduce_sum(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
func.func @test_reduce_sum(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
%0 = "tosa.reduce_sum"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xf32>) -> tensor<1x21x3xf32>
%1 = "tosa.reshape"(%0) {new_shape = [21, 3]} : (tensor<1x21x3xf32>) -> tensor<21x3xf32>
return %1 : tensor<21x3xf32>
@ -383,21 +383,21 @@ func @test_reduce_sum(%arg0: tensor<13x21x3xf32>) -> tensor<21x3xf32> {
// -----
// CHECK-LABEL: concat
func @test_concat(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<26x21x3xf32> {
func.func @test_concat(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x21x3xf32>) -> tensor<26x21x3xf32> {
%0 = "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<13x21x3xf32>, tensor<13x21x3xf32>) -> tensor<26x21x3xf32>
return %0 : tensor<26x21x3xf32>
}
// -----
// CHECK-LABEL: pad
func @test_pad(%arg0: tensor<13x21x3xf32>, %arg1: tensor<3x2xi32>) -> tensor<13x21x3xf32> {
func.func @test_pad(%arg0: tensor<13x21x3xf32>, %arg1: tensor<3x2xi32>) -> tensor<13x21x3xf32> {
%0 = "tosa.pad"(%arg0, %arg1) : (tensor<13x21x3xf32>, tensor<3x2xi32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: pad_explicit_value
func @test_pad_explicit_value(%arg0: tensor<13x21x3xf32>, %arg1: tensor<3x2xi32>) -> tensor<13x21x3xf32> {
func.func @test_pad_explicit_value(%arg0: tensor<13x21x3xf32>, %arg1: tensor<3x2xi32>) -> tensor<13x21x3xf32> {
%0 = "tosa.const"() {value = dense<3.14> : tensor<f32>} : () -> tensor<f32>
%1 = "tosa.pad"(%arg0, %arg1, %0) : (tensor<13x21x3xf32>, tensor<3x2xi32>, tensor<f32>) -> tensor<13x21x3xf32>
return %1 : tensor<13x21x3xf32>
@ -405,35 +405,35 @@ func @test_pad_explicit_value(%arg0: tensor<13x21x3xf32>, %arg1: tensor<3x2xi32>
// -----
// CHECK-LABEL: reshape
func @test_reshape(%arg0: tensor<13x21x3xf32>) -> tensor<1x819xf32> {
func.func @test_reshape(%arg0: tensor<13x21x3xf32>) -> tensor<1x819xf32> {
%0 = "tosa.reshape"(%arg0) {new_shape = [1, 819]} : (tensor<13x21x3xf32>) -> tensor<1x819xf32>
return %0 : tensor<1x819xf32>
}
// -----
// CHECK-LABEL: reverse
func @test_reverse(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_reverse(%arg0: tensor<13x21x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.reverse"(%arg0) {axis = 0 : i64} : (tensor<13x21x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: slice
func @test_slice(%arg0: tensor<13x21x3xf32>) -> tensor<4x11x1xf32> {
func.func @test_slice(%arg0: tensor<13x21x3xf32>) -> tensor<4x11x1xf32> {
%0 = "tosa.slice"(%arg0) {start = [6, 8, 0], size = [4, 11, 1]} : (tensor<13x21x3xf32>) -> tensor<4x11x1xf32>
return %0 : tensor<4x11x1xf32>
}
// -----
// CHECK-LABEL: tile
func @test_tile(%arg0: tensor<13x21x3xf32>) -> tensor<39x21x6xf32> {
func.func @test_tile(%arg0: tensor<13x21x3xf32>) -> tensor<39x21x6xf32> {
%0 = "tosa.tile"(%arg0) {multiples = [3, 1, 2]} : (tensor<13x21x3xf32>) -> tensor<39x21x6xf32>
return %0 : tensor<39x21x6xf32>
}
// -----
// CHECK-LABEL: transpose
func @test_transpose(%arg0: tensor<13x21x3xf32>) -> tensor<3x13x21xf32> {
func.func @test_transpose(%arg0: tensor<13x21x3xf32>) -> tensor<3x13x21xf32> {
%0 = "tosa.const"() {value = dense<[2, 0, 1]> : tensor<3xi32>} : () -> tensor<3xi32>
%1 = "tosa.transpose"(%arg0, %0) : (tensor<13x21x3xf32>, tensor<3xi32>) -> tensor<3x13x21xf32>
return %1 : tensor<3x13x21xf32>
@ -441,70 +441,70 @@ func @test_transpose(%arg0: tensor<13x21x3xf32>) -> tensor<3x13x21xf32> {
// -----
// CHECK-LABEL: gather
func @test_gather(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x26xi32>) -> tensor<13x26x3xf32> {
func.func @test_gather(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x26xi32>) -> tensor<13x26x3xf32> {
%0 = "tosa.gather"(%arg0, %arg1) : (tensor<13x21x3xf32>, tensor<13x26xi32>) -> tensor<13x26x3xf32>
return %0 : tensor<13x26x3xf32>
}
// -----
// CHECK-LABEL: scatter
func @test_scatter(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x26xi32>, %arg2: tensor<13x26x3xf32>) -> tensor<13x21x3xf32> {
func.func @test_scatter(%arg0: tensor<13x21x3xf32>, %arg1: tensor<13x26xi32>, %arg2: tensor<13x26x3xf32>) -> tensor<13x21x3xf32> {
%0 = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<13x21x3xf32>, tensor<13x26xi32>, tensor<13x26x3xf32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: resize
func @test_resize(%arg0: tensor<1x32x32x8xf32>) -> tensor<1x64x64x8xf32> {
func.func @test_resize(%arg0: tensor<1x32x32x8xf32>) -> tensor<1x64x64x8xf32> {
%1 = "tosa.resize"(%arg0) {output_size = [64, 64], stride = [1024, 1024], offset = [0, 0], shift = 10 : i32, stride_fp = [0.0 : f32, 0.0 : f32], offset_fp = [0.0 : f32, 0.0 : f32], mode = "BILINEAR"} : (tensor<1x32x32x8xf32>) -> tensor<1x64x64x8xf32>
return %1 : tensor<1x64x64x8xf32>
}
// -----
// CHECK-LABEL: cast
func @test_cast1(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3xf32> {
func.func @test_cast1(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3xf32> {
%0 = "tosa.cast"(%arg0) : (tensor<13x21x3xi32>) -> tensor<13x21x3xf32>
return %0 : tensor<13x21x3xf32>
}
// -----
// CHECK-LABEL: cast2
func @test_cast2(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3x!quant.uniform<u8:f32, 0.078431375324726104:128>> {
func.func @test_cast2(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3x!quant.uniform<u8:f32, 0.078431375324726104:128>> {
%0 = "tosa.cast"(%arg0) : (tensor<13x21x3xi32>) -> tensor<13x21x3x!quant.uniform<u8:f32, 0.078431375324726104:128>>
return %0 : tensor<13x21x3x!quant.uniform<u8:f32, 0.078431375324726104:128>>
}
// -----
// CHECK-LABEL: cast3
func @test_cast3(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3x!quant.uniform<i16:f32, 0.078431375324726104:128>> {
func.func @test_cast3(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3x!quant.uniform<i16:f32, 0.078431375324726104:128>> {
%0 = "tosa.cast"(%arg0) : (tensor<13x21x3xi32>) -> tensor<13x21x3x!quant.uniform<i16:f32, 0.078431375324726104:128>>
return %0 : tensor<13x21x3x!quant.uniform<i16:f32, 0.078431375324726104:128>>
}
// -----
// CHECK-LABEL: rescale
func @test_rescale(%arg0: tensor<13x21x3x!quant.uniform<u8:f32, 0.015655439347028732:127>>) -> tensor<13x21x3x!quant.uniform<i8:f32, 0.015655439347028732:-1>> {
func.func @test_rescale(%arg0: tensor<13x21x3x!quant.uniform<u8:f32, 0.015655439347028732:127>>) -> tensor<13x21x3x!quant.uniform<i8:f32, 0.015655439347028732:-1>> {
%0 = "tosa.rescale"(%arg0) {double_round = false, input_zp = 127 : i32, multiplier = [1073741824 : i32], output_zp = -1 : i32, per_channel = false, scale32 = true, shift = [30 : i32]} : (tensor<13x21x3x!quant.uniform<u8:f32, 0.015655439347028732:127>>) -> tensor<13x21x3x!quant.uniform<i8:f32, 0.015655439347028732:-1>>
return %0 : tensor<13x21x3x!quant.uniform<i8:f32, 0.015655439347028732:-1>>
}
// -----
// CHECK-LABEL: const
func @test_const(%arg0 : index) -> tensor<4xi32> {
func.func @test_const(%arg0 : index) -> tensor<4xi32> {
%0 = "tosa.const"() {value = dense<[3, 0, 1, 2]> : tensor<4xi32>} : () -> tensor<4xi32>
return %0 : tensor<4xi32>
}
// -----
// CHECK-LABEL: identity
func @test_identity(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
func.func @test_identity(%arg0: tensor<13x21x3xi32>) -> tensor<13x21x3xi32> {
%0 = "tosa.identity"(%arg0) : (tensor<13x21x3xi32>) -> tensor<13x21x3xi32>
return %0 : tensor<13x21x3xi32>
}
// -----
// CHECK-LABEL: cond_if
func @test_cond_if(%arg0: tensor<f32>, %arg1: tensor<f32>, %arg2: tensor<i1>) -> tensor<f32> {
func.func @test_cond_if(%arg0: tensor<f32>, %arg1: tensor<f32>, %arg2: tensor<i1>) -> tensor<f32> {
%0 = "tosa.cond_if"(%arg2, %arg0, %arg1) ({
^bb0(%arg3: tensor<f32>, %arg4: tensor<f32>):
%1 = "tosa.add"(%arg3, %arg4) : (tensor<f32>, tensor<f32>) -> tensor<f32>
@ -519,7 +519,7 @@ func @test_cond_if(%arg0: tensor<f32>, %arg1: tensor<f32>, %arg2: tensor<i1>) ->
// -----
// CHECK-LABEL: while_loop
func @test_while_loop(%arg0: tensor<10xi32>, %arg1: tensor<i32>) {
func.func @test_while_loop(%arg0: tensor<10xi32>, %arg1: tensor<i32>) {
%0 = "tosa.const"() {value = dense<0> : tensor<i32>} : () -> tensor<i32>
%1:3 = "tosa.while_loop"(%0, %0, %arg0) ({
^bb0(%arg2: tensor<i32>, %arg3: tensor<i32>, %arg4: tensor<10xi32>):

4
mlir/test/Dialect/Tosa/quant-test.mlir
View File

@ -2,7 +2,7 @@
// -----
// CHECK-LABEL: test_build_qtype
func @test_build_qtype(%arg0 : tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>>) -> tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>> {
func.func @test_build_qtype(%arg0 : tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>>) -> tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>> {
// CHECK: tosa.negate
%0 = "tosa.negate"(%arg0) : (tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>>) -> tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>>
return %0 : tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.015680249780416489:128>>
@ -10,7 +10,7 @@ func @test_build_qtype(%arg0 : tensor<16x1x1x8x!quant.uniform<u8<1:255>:f32, 0.0
// -----
// CHECK-LABEL: test_build_mult_and_shift
func @test_build_mult_and_shift(%arg0: tensor<1x32x32x8x!quant.uniform<i8:f32, 0.015684768557548523>>, %arg1 : tensor<16x1x1x8x!quant.uniform<i8<-127:127>:f32, 0.015680249780416489>>, %arg2 : tensor<16xi32>) -> tensor<1x32x32x16x!quant.uniform<i8:f32, 0.078431375324726104>> {
func.func @test_build_mult_and_shift(%arg0: tensor<1x32x32x8x!quant.uniform<i8:f32, 0.015684768557548523>>, %arg1 : tensor<16x1x1x8x!quant.uniform<i8<-127:127>:f32, 0.015680249780416489>>, %arg2 : tensor<16xi32>) -> tensor<1x32x32x16x!quant.uniform<i8:f32, 0.078431375324726104>> {
// CHECK: tosa.conv2d
%0 = "tosa.conv2d"(%arg0, %arg1, %arg2) {pad = [1, 1, 2, 2], dilation = [2, 1], stride = [1, 1], quantization_info = {input_zp = -1 : i32, weight_zp = 0 : i32}} : (tensor<1x32x32x8x!quant.uniform<i8:f32, 0.015684768557548523>>, tensor<16x1x1x8x!quant.uniform<i8<-127:127>:f32, 0.015680249780416489>>, tensor<16xi32>) -> tensor<1x32x32x16x!quant.uniform<i8:f32, 0.078431375324726104>>
return %0 : tensor<1x32x32x16x!quant.uniform<i8:f32, 0.078431375324726104>>

4
mlir/test/Dialect/Tosa/tosa-decompose-conv2d.mlir
View File

@ -3,7 +3,7 @@
// -----
// CHECK-LABEL: @conv2d_as_fully_connected
func @conv2d_as_fully_connected(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<3x1x1x2xf32>, %arg2: tensor<3xf32>) -> tensor<4x10x10x3xf32> {
func.func @conv2d_as_fully_connected(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<3x1x1x2xf32>, %arg2: tensor<3xf32>) -> tensor<4x10x10x3xf32> {
// CHECK-NOT: "tosa.conv2d"
// CHECK: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [400, 2]}
// CHECK-SAME: -> tensor<400x2xf32>
@ -21,7 +21,7 @@ func @conv2d_as_fully_connected(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<3x1x
// -----
// CHECK-LABEL: @conv2d_as_fully_connected_quant
func @conv2d_as_fully_connected_quant(%arg0: tensor<4x10x10x2xi8>, %arg1: tensor<3x1x1x2xi8>, %arg2: tensor<3xi32>) -> tensor<4x10x10x3xi32> {
func.func @conv2d_as_fully_connected_quant(%arg0: tensor<4x10x10x2xi8>, %arg1: tensor<3x1x1x2xi8>, %arg2: tensor<3xi32>) -> tensor<4x10x10x3xi32> {
// CHECK-NOT: "tosa.conv2d"
// CHECK: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [400, 2]}
// CHECK-SAME: -> tensor<400x2xi8>

4
mlir/test/Dialect/Tosa/tosa-decompose-depthwise.mlir
View File

@ -3,7 +3,7 @@
// -----
// CHECK-LABEL: @depthwise_conv2d_as_mul
func @depthwise_conv2d_as_mul(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<1x1x2x3xf32>, %arg2: tensor<6xf32>) -> tensor<4x10x10x6xf32> {
func.func @depthwise_conv2d_as_mul(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<1x1x2x3xf32>, %arg2: tensor<6xf32>) -> tensor<4x10x10x6xf32> {
// CHECK-NOT: "tosa.depthwise_conv2d"
// CHECK: %[[VAR0:.*]] = "tosa.reshape"(%arg0) {new_shape = [4, 10, 10, 2, 1]}
// CHECK-SAME: -> tensor<4x10x10x2x1xf32>
@ -23,7 +23,7 @@ func @depthwise_conv2d_as_mul(%arg0: tensor<4x10x10x2xf32>, %arg1: tensor<1x1x2x
// -----
// CHECK-LABEL: @depthwise_conv2d_as_mul_q
func @depthwise_conv2d_as_mul_q(%arg0: tensor<4x10x10x2xi8>, %arg1: tensor<1x1x2x3xi8>, %arg2: tensor<6xi32>) -> tensor<4x10x10x6xi32> {
func.func @depthwise_conv2d_as_mul_q(%arg0: tensor<4x10x10x2xi8>, %arg1: tensor<1x1x2x3xi8>, %arg2: tensor<6xi32>) -> tensor<4x10x10x6xi32> {
// CHECK: "tosa.depthwise_conv2d"
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1], quantization_info = {input_zp = 0 : i32, weight_zp = 0 : i32}} : (tensor<4x10x10x2xi8>, tensor<1x1x2x3xi8>, tensor<6xi32>) -> tensor<4x10x10x6xi32>
return %0 : tensor<4x10x10x6xi32>

10
mlir/test/Dialect/Tosa/tosa-decompose-transpose-conv.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt --split-input-file --tosa-optional-decompositions %s | FileCheck %s
// CHECK-LABEL: @transpose_conv2d
func @transpose_conv2d(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) -> tensor<2x?x?x5xf32> {
func.func @transpose_conv2d(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) -> tensor<2x?x?x5xf32> {
// CHECK: %[[REV1:.+]] = "tosa.reverse"(%arg1) {axis = 1 : i64}
// CHECK: %[[REV2:.+]] = "tosa.reverse"(%[[REV1]]) {axis = 2 : i64}
// CHECK: "tosa.conv2d"(%arg0, %[[REV2]], %arg2) {dilation = [1, 1], pad = [2, 2, 5, 5], stride = [1, 1]}
@ -13,7 +13,7 @@ func @transpose_conv2d(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>,
// -----
// CHECK-LABEL: @transpose_conv2d_quantized
func @transpose_conv2d_quantized(%arg0: tensor<2x16x14x3xi8>, %arg1: tensor<5x3x6x3xi8>, %arg2: tensor<5xi32>) -> (tensor<2x18x19x5xi32>) {
func.func @transpose_conv2d_quantized(%arg0: tensor<2x16x14x3xi8>, %arg1: tensor<5x3x6x3xi8>, %arg2: tensor<5xi32>) -> (tensor<2x18x19x5xi32>) {
// CHECK: %[[REV1:.+]] = "tosa.reverse"(%arg1) {axis = 1 : i64}
// CHECK: %[[REV2:.+]] = "tosa.reverse"(%[[REV1]]) {axis = 2 : i64}
// CHECK: "tosa.conv2d"(%arg0, %[[REV2]], %arg2) {dilation = [1, 1], pad = [2, 2, 5, 5], quantization_info = {input_zp = -22 : i32, weight_zp = 42 : i32}, stride = [1, 1]}
@ -24,7 +24,7 @@ func @transpose_conv2d_quantized(%arg0: tensor<2x16x14x3xi8>, %arg1: tensor<5x3x
// ----
// CHECK-LABEL: @transpose_conv2d_dilated
func @transpose_conv2d_dilated(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) -> tensor<2x?x?x5xf32> {
func.func @transpose_conv2d_dilated(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) -> tensor<2x?x?x5xf32> {
// CHECK: %[[REV1:.+]] = "tosa.reverse"(%arg1) {axis = 1 : i64}
// CHECK: %[[REV2:.+]] = "tosa.reverse"(%[[REV1]]) {axis = 2 : i64}
// CHECK: "tosa.conv2d"(%arg0, %[[REV2]], %arg2) {dilation = [2, 3], pad = [4, 4, 15, 15], stride = [1, 1]}
@ -36,7 +36,7 @@ func @transpose_conv2d_dilated(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6
// ----
// CHECK-LABEL: @transpose_conv2d_strided
func @transpose_conv2d_strided(%arg0: tensor<2x17x15x3xf32>, %arg1: tensor<5x3x5x3xf32>, %arg2: tensor<5xf32>) -> tensor<2x?x?x5xf32> {
func.func @transpose_conv2d_strided(%arg0: tensor<2x17x15x3xf32>, %arg1: tensor<5x3x5x3xf32>, %arg2: tensor<5xf32>) -> tensor<2x?x?x5xf32> {
// Manipulate the weight matrix to handle striding.
// CHECK-DAG: %[[PADV:.+]] = "tosa.const"() {value = dense<{{\[\[}}0, 0], [0, 1], [0, 1], [0, 0]]> : tensor<4x2xi32>}
// CHECK-DAG: %[[TRANSV:.+]] = "tosa.const"() {value = dense<[2, 4, 0, 1, 3, 5]> : tensor<6xi32>}
@ -68,7 +68,7 @@ func @transpose_conv2d_strided(%arg0: tensor<2x17x15x3xf32>, %arg1: tensor<5x3x5
// ----
// CHECK-LABEL: @transpose_conv2d_strided_quantized
func @transpose_conv2d_strided_quantized(%arg0: tensor<2x17x15x3xi8>, %arg1: tensor<5x3x5x3xi8>, %arg2: tensor<5xi32>) -> (tensor<2x35x47x5xi32>) {
func.func @transpose_conv2d_strided_quantized(%arg0: tensor<2x17x15x3xi8>, %arg1: tensor<5x3x5x3xi8>, %arg2: tensor<5xi32>) -> (tensor<2x35x47x5xi32>) {
// Manipulate the weight matrix to handle striding.
// CHECK-DAG: %[[PADV:.+]] = "tosa.const"() {value = dense<{{\[\[}}0, 0], [0, 1], [0, 1], [0, 0]]> : tensor<4x2xi32>}
// CHECK-DAG: %[[TRANSV:.+]] = "tosa.const"() {value = dense<[2, 4, 0, 1, 3, 5]> : tensor<6xi32>}

194
mlir/test/Dialect/Tosa/tosa-infer-shapes.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt --split-input-file --tosa-infer-shapes %s | FileCheck %s
// CHECK-LABEL: @test_return
func @test_return(%arg0 : tensor<4xf32>) -> tensor<*xf32> {
func.func @test_return(%arg0 : tensor<4xf32>) -> tensor<*xf32> {
// CHECK: [[LOG:%.+]] = "tosa.log"(%arg0) : (tensor<4xf32>) -> tensor<4xf32>
// CHECK: tensor.cast [[LOG]] : tensor<4xf32> to tensor<*xf32>
%0 = "tosa.log"(%arg0) : (tensor<4xf32>) -> tensor<*xf32>
@ -11,7 +11,7 @@ func @test_return(%arg0 : tensor<4xf32>) -> tensor<*xf32> {
// -----
// CHECK-LABEL: @test_multiple
func @test_multiple(%arg0 : tensor<4xf32>, %arg1 : tensor<1xf32>, %arg2 : tensor<f32>) -> tensor<*xf32> {
func.func @test_multiple(%arg0 : tensor<4xf32>, %arg1 : tensor<1xf32>, %arg2 : tensor<f32>) -> tensor<*xf32> {
// CHECK: [[ADD:%.+]] = "tosa.add"(%arg0, %arg1) : (tensor<4xf32>, tensor<1xf32>) -> tensor<4xf32>
%0 = "tosa.add"(%arg0, %arg1) : (tensor<4xf32>, tensor<1xf32>) -> tensor<*xf32>
@ -26,7 +26,7 @@ func @test_multiple(%arg0 : tensor<4xf32>, %arg1 : tensor<1xf32>, %arg2 : tensor
// -----
// CHECK-LABEL: @test_unary_f32
func @test_unary_f32(%arg0 : tensor<4xf32>) -> () {
func.func @test_unary_f32(%arg0 : tensor<4xf32>) -> () {
// CHECK: "tosa.abs"(%arg0) : (tensor<4xf32>) -> tensor<4xf32>
%0 = "tosa.abs"(%arg0) : (tensor<4xf32>) -> tensor<*xf32>
@ -74,7 +74,7 @@ func @test_unary_f32(%arg0 : tensor<4xf32>) -> () {
// -----
// CHECK-LABEL: @test_unary_i32
func @test_unary_i32(%arg0 : tensor<4xi32>) -> () {
func.func @test_unary_i32(%arg0 : tensor<4xi32>) -> () {
// CHECK: "tosa.abs"(%arg0) : (tensor<4xi32>) -> tensor<4xi32>
%0 = "tosa.abs"(%arg0) : (tensor<4xi32>) -> tensor<*xi32>
@ -107,7 +107,7 @@ func @test_unary_i32(%arg0 : tensor<4xi32>) -> () {
// -----
// CHECK-LABEL: @test_unary_i1
func @test_unary_i1(%arg0 : tensor<4xi1>) -> () {
func.func @test_unary_i1(%arg0 : tensor<4xi1>) -> () {
// CHECK: "tosa.logical_not"(%arg0) : (tensor<4xi1>) -> tensor<4xi1>
%0 = "tosa.logical_not"(%arg0) : (tensor<4xi1>) -> tensor<*xi1>
return
@ -116,7 +116,7 @@ func @test_unary_i1(%arg0 : tensor<4xi1>) -> () {
// -----
// CHECK-LABEL: @test_binary_scalar_f32
func @test_binary_scalar_f32(%arg0 : tensor<4xf32>, %arg1 : tensor<f32>) -> () {
func.func @test_binary_scalar_f32(%arg0 : tensor<4xf32>, %arg1 : tensor<f32>) -> () {
// CHECK: "tosa.add"(%arg0, %arg1) : (tensor<4xf32>, tensor<f32>) -> tensor<4xf32>
%0 = "tosa.add"(%arg0, %arg1) : (tensor<4xf32>, tensor<f32>) -> tensor<*xf32>
@ -150,7 +150,7 @@ func @test_binary_scalar_f32(%arg0 : tensor<4xf32>, %arg1 : tensor<f32>) -> () {
// -----
// CHECK-LABEL: @test_binary_broadcast_f32
func @test_binary_broadcast_f32(%arg0 : tensor<4xf32>, %arg1 : tensor<1xf32>) -> () {
func.func @test_binary_broadcast_f32(%arg0 : tensor<4xf32>, %arg1 : tensor<1xf32>) -> () {
// CHECK: "tosa.add"(%arg0, %arg1) : (tensor<4xf32>, tensor<1xf32>) -> tensor<4xf32>
%0 = "tosa.add"(%arg0, %arg1) : (tensor<4xf32>, tensor<1xf32>) -> tensor<*xf32>
@ -184,7 +184,7 @@ func @test_binary_broadcast_f32(%arg0 : tensor<4xf32>, %arg1 : tensor<1xf32>) ->
// -----
// CHECK-LABEL: @test_binary_i32
func @test_binary_i32(%arg0 : tensor<4xi32>, %arg1 : tensor<i32>) -> () {
func.func @test_binary_i32(%arg0 : tensor<4xi32>, %arg1 : tensor<i32>) -> () {
// CHECK: "tosa.add"(%arg0, %arg1) : (tensor<4xi32>, tensor<i32>) -> tensor<4xi32>
%0 = "tosa.add"(%arg0, %arg1) : (tensor<4xi32>, tensor<i32>) -> tensor<*xi32>
@ -233,7 +233,7 @@ func @test_binary_i32(%arg0 : tensor<4xi32>, %arg1 : tensor<i32>) -> () {
// -----
// CHECK-LABEL: @test_binary_i1
func @test_binary_i1(%arg0 : tensor<4xi1>, %arg1 : tensor<i1>) -> () {
func.func @test_binary_i1(%arg0 : tensor<4xi1>, %arg1 : tensor<i1>) -> () {
// CHECK "tosa.logical_and"(%arg0, %arg1) : (tensor<4xi1>, tensor<i1>) -> tensor<4xi1>
%0 = "tosa.logical_and"(%arg0, %arg1): (tensor<4xi1>, tensor<i1>) -> tensor<*xi1>
@ -249,7 +249,7 @@ func @test_binary_i1(%arg0 : tensor<4xi1>, %arg1 : tensor<i1>) -> () {
// -----
// CHECK-LABEL: @test_select_i32
func @test_select_i32(%arg0 : tensor<4xi1>, %arg1 : tensor<i32>, %arg2 : tensor<4xi32>) -> () {
func.func @test_select_i32(%arg0 : tensor<4xi1>, %arg1 : tensor<i32>, %arg2 : tensor<4xi32>) -> () {
// CHECK: "tosa.select"(%arg0, %arg1, %arg2) : (tensor<4xi1>, tensor<i32>, tensor<4xi32>) -> tensor<4xi32>
%0 = "tosa.select"(%arg0, %arg1, %arg2): (tensor<4xi1>, tensor<i32>, tensor<4xi32>) -> tensor<*xi32>
@ -259,7 +259,7 @@ func @test_select_i32(%arg0 : tensor<4xi1>, %arg1 : tensor<i32>, %arg2 : tensor<
// -----
// CHECK-LABEL: @test_static_argmax
func @test_static_argmax(%arg0 : tensor<2x3xi32>) -> () {
func.func @test_static_argmax(%arg0 : tensor<2x3xi32>) -> () {
// CHECK: "tosa.argmax"(%arg0) {axis = 0 : i64} : (tensor<2x3xi32>) -> tensor<3xi32>
%0 = "tosa.argmax"(%arg0) {axis = 0 : i64} : (tensor<2x3xi32>) -> tensor<?xi32>
@ -271,7 +271,7 @@ func @test_static_argmax(%arg0 : tensor<2x3xi32>) -> () {
// -----
// CHECK-LABEL: @test_dynamic_argmax
func @test_dynamic_argmax(%arg0 : tensor<2x?xi32>) -> () {
func.func @test_dynamic_argmax(%arg0 : tensor<2x?xi32>) -> () {
// CHECK: "tosa.argmax"(%arg0) {axis = 0 : i64} : (tensor<2x?xi32>) -> tensor<?xi32>
%0 = "tosa.argmax"(%arg0) {axis = 0 : i64} : (tensor<2x?xi32>) -> tensor<?xi32>
@ -283,7 +283,7 @@ func @test_dynamic_argmax(%arg0 : tensor<2x?xi32>) -> () {
// -----
// CHECK-LABEL: @test_static_fully_connected
func @test_static_fully_connected(%arg0 : tensor<3x4xf32>, %arg1 : tensor<5x4xf32>, %arg2 : tensor<5xf32>) -> () {
func.func @test_static_fully_connected(%arg0 : tensor<3x4xf32>, %arg1 : tensor<5x4xf32>, %arg2 : tensor<5xf32>) -> () {
// CHECK: "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<3x4xf32>, tensor<5x4xf32>, tensor<5xf32>) -> tensor<3x5xf32>
%0 = "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<3x4xf32>, tensor<5x4xf32>, tensor<5xf32>) -> tensor<?x?xf32>
return
@ -292,7 +292,7 @@ func @test_static_fully_connected(%arg0 : tensor<3x4xf32>, %arg1 : tensor<5x4xf3
// -----
// CHECK-LABEL: @test_static_input_fully_connected
func @test_static_input_fully_connected(%arg0 : tensor<3x4xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<?xf32>) -> () {
func.func @test_static_input_fully_connected(%arg0 : tensor<3x4xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<?xf32>) -> () {
// CHECK: "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<3x4xf32>, tensor<?x?xf32>, tensor<?xf32>) -> tensor<3x?xf32>
%0 = "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<3x4xf32>, tensor<?x?xf32>, tensor<?xf32>) -> tensor<?x?xf32>
return
@ -301,7 +301,7 @@ func @test_static_input_fully_connected(%arg0 : tensor<3x4xf32>, %arg1 : tensor<
// -----
// CHECK-LABEL: @test_static_weight_fully_connected
func @test_static_weight_fully_connected(%arg0 : tensor<?x?xf32>, %arg1 : tensor<5x4xf32>, %arg2 : tensor<?xf32>) -> () {
func.func @test_static_weight_fully_connected(%arg0 : tensor<?x?xf32>, %arg1 : tensor<5x4xf32>, %arg2 : tensor<?xf32>) -> () {
// CHECK: "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<?x?xf32>, tensor<5x4xf32>, tensor<?xf32>) -> tensor<?x5xf32>
%0 = "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<?x?xf32>, tensor<5x4xf32>, tensor<?xf32>) -> tensor<?x?xf32>
return
@ -310,7 +310,7 @@ func @test_static_weight_fully_connected(%arg0 : tensor<?x?xf32>, %arg1 : tensor
// -----
// CHECK-LABEL: @test_static_bias_fully_connected
func @test_static_bias_fully_connected(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<5xf32>) -> () {
func.func @test_static_bias_fully_connected(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<5xf32>) -> () {
// CHECK: "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<?x?xf32>, tensor<?x?xf32>, tensor<5xf32>) -> tensor<?x5xf32>
%0 = "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<?x?xf32>, tensor<?x?xf32>, tensor<5xf32>) -> tensor<?x?xf32>
return
@ -319,7 +319,7 @@ func @test_static_bias_fully_connected(%arg0 : tensor<?x?xf32>, %arg1 : tensor<?
// -----
// CHECK-LABEL: @test_static_out_fully_connected
func @test_static_out_fully_connected(%arg0 : tensor<3x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<5xf32>) -> () {
func.func @test_static_out_fully_connected(%arg0 : tensor<3x?xf32>, %arg1 : tensor<?x?xf32>, %arg2 : tensor<5xf32>) -> () {
// CHECK: "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<3x?xf32>, tensor<?x?xf32>, tensor<5xf32>) -> tensor<3x5xf32>
%0 = "tosa.fully_connected"(%arg0, %arg1, %arg2) : (tensor<3x?xf32>, tensor<?x?xf32>, tensor<5xf32>) -> tensor<?x?xf32>
return
@ -328,7 +328,7 @@ func @test_static_out_fully_connected(%arg0 : tensor<3x?xf32>, %arg1 : tensor<?x
// -----
// CHECK-LABEL: @test_static_matmul
func @test_static_matmul(%arg0 : tensor<2x3x4xi32>, %arg1 : tensor<2x4x5xi32>) -> () {
func.func @test_static_matmul(%arg0 : tensor<2x3x4xi32>, %arg1 : tensor<2x4x5xi32>) -> () {
// CHECK: "tosa.matmul"(%arg0, %arg1) : (tensor<2x3x4xi32>, tensor<2x4x5xi32>) -> tensor<2x3x5xi32>
%0 = "tosa.matmul"(%arg0, %arg1) : (tensor<2x3x4xi32>, tensor<2x4x5xi32>) -> tensor<?x?x?xi32>
@ -338,7 +338,7 @@ func @test_static_matmul(%arg0 : tensor<2x3x4xi32>, %arg1 : tensor<2x4x5xi32>) -
// -----
// CHECK-LABEL: @test_dynamic_lhs_matmul
func @test_dynamic_lhs_matmul(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<2x4x5xi32>) -> () {
func.func @test_dynamic_lhs_matmul(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<2x4x5xi32>) -> () {
// CHECK: "tosa.matmul"(%arg0, %arg1) : (tensor<?x?x?xi32>, tensor<2x4x5xi32>) -> tensor<2x?x5xi32>
%0 = "tosa.matmul"(%arg0, %arg1) : (tensor<?x?x?xi32>, tensor<2x4x5xi32>) -> tensor<?x?x?xi32>
@ -348,7 +348,7 @@ func @test_dynamic_lhs_matmul(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<2x4x5xi3
// -----
// CHECK-LABEL: @test_dynamic_rhs_matmul
func @test_dynamic_rhs_matmul(%arg0 : tensor<2x3x4xi32>, %arg1 : tensor<?x?x?xi32>) -> () {
func.func @test_dynamic_rhs_matmul(%arg0 : tensor<2x3x4xi32>, %arg1 : tensor<?x?x?xi32>) -> () {
// CHECK: "tosa.matmul"(%arg0, %arg1) : (tensor<2x3x4xi32>, tensor<?x?x?xi32>) -> tensor<2x3x?xi32>
%0 = "tosa.matmul"(%arg0, %arg1) : (tensor<2x3x4xi32>, tensor<?x?x?xi32>) -> tensor<?x?x?xi32>
@ -358,7 +358,7 @@ func @test_dynamic_rhs_matmul(%arg0 : tensor<2x3x4xi32>, %arg1 : tensor<?x?x?xi3
// -----
// CHECK-LABEL: @test_dynamic_mixed_matmul
func @test_dynamic_mixed_matmul(%arg0 : tensor<?x3x?xi32>, %arg1 : tensor<?x?x5xi32>) -> () {
func.func @test_dynamic_mixed_matmul(%arg0 : tensor<?x3x?xi32>, %arg1 : tensor<?x?x5xi32>) -> () {
// CHECK: "tosa.matmul"(%arg0, %arg1) : (tensor<?x3x?xi32>, tensor<?x?x5xi32>) -> tensor<?x3x5xi32>
%0 = "tosa.matmul"(%arg0, %arg1) : (tensor<?x3x?xi32>, tensor<?x?x5xi32>) -> tensor<?x?x?xi32>
@ -368,7 +368,7 @@ func @test_dynamic_mixed_matmul(%arg0 : tensor<?x3x?xi32>, %arg1 : tensor<?x?x5x
// -----
// CHECK-LABLE: @test_table_static
func @test_table_static(%arg0 : tensor<4x5xi16>, %arg1 : tensor<513xi16>) -> () {
func.func @test_table_static(%arg0 : tensor<4x5xi16>, %arg1 : tensor<513xi16>) -> () {
// CHECK:"tosa.table"(%arg0, %arg1) : (tensor<4x5xi16>, tensor<513xi16>) -> tensor<4x5xi16>
%0 = "tosa.table"(%arg0, %arg1) : (tensor<4x5xi16>, tensor<513xi16>) -> tensor<?x?xi16>
return
@ -377,7 +377,7 @@ func @test_table_static(%arg0 : tensor<4x5xi16>, %arg1 : tensor<513xi16>) -> ()
// -----
// CHECK-LABLE: @test_table_dynamic
func @test_table_dynamic(%arg0 : tensor<4x?xi16>, %arg1 : tensor<513xi16>) -> () {
func.func @test_table_dynamic(%arg0 : tensor<4x?xi16>, %arg1 : tensor<513xi16>) -> () {
// CHECK:"tosa.table"(%arg0, %arg1) : (tensor<4x?xi16>, tensor<513xi16>) -> tensor<4x?xi16>
%0 = "tosa.table"(%arg0, %arg1) : (tensor<4x?xi16>, tensor<513xi16>) -> tensor<?x?xi16>
return
@ -386,7 +386,7 @@ func @test_table_dynamic(%arg0 : tensor<4x?xi16>, %arg1 : tensor<513xi16>) -> ()
// -----
// CHECK-LABEL: @test_static_reshape
func @test_static_reshape(%arg0 : tensor<4x4xi32>) -> () {
func.func @test_static_reshape(%arg0 : tensor<4x4xi32>) -> () {
// CHECK: "tosa.reshape"(%arg0) {new_shape = [16]} : (tensor<4x4xi32>) -> tensor<16xi32>
%0 = "tosa.reshape"(%arg0) {new_shape = [16]} : (tensor<4x4xi32>) -> tensor<?xi32>
@ -401,7 +401,7 @@ func @test_static_reshape(%arg0 : tensor<4x4xi32>) -> () {
// -----
// CHECK-LABEL: @test_dynamic_reshape
func @test_dynamic_reshape(%arg0 : tensor<4x?xi32>) -> () {
func.func @test_dynamic_reshape(%arg0 : tensor<4x?xi32>) -> () {
// CHECK: %0 = "tosa.reshape"(%arg0) {new_shape = [16]} : (tensor<4x?xi32>) -> tensor<16xi32>
%0 = "tosa.reshape"(%arg0) {new_shape = [16]} : (tensor<4x?xi32>) -> tensor<?xi32>
@ -417,7 +417,7 @@ func @test_dynamic_reshape(%arg0 : tensor<4x?xi32>) -> () {
// -----
// CHECK: @test_reduce_binary
func @test_reduce_binary(%arg0 : tensor<2x3x?x?xi1>) -> () {
func.func @test_reduce_binary(%arg0 : tensor<2x3x?x?xi1>) -> () {
// CHECK: "tosa.reduce_all"(%arg0) {axis = 0 : i64} : (tensor<2x3x?x?xi1>) -> tensor<1x3x?x?xi1>
%0 = "tosa.reduce_all"(%arg0) {axis = 0 : i64} : (tensor<2x3x?x?xi1>) -> tensor<?x?x?x?xi1>
@ -439,7 +439,7 @@ func @test_reduce_binary(%arg0 : tensor<2x3x?x?xi1>) -> () {
// -----
// CHECK: @test_reduce_float
func @test_reduce_float(%arg0 : tensor<2x3x?x?xf32>) -> () {
func.func @test_reduce_float(%arg0 : tensor<2x3x?x?xf32>) -> () {
// CHECK: "tosa.reduce_sum"(%arg0) {axis = 0 : i64} : (tensor<2x3x?x?xf32>) -> tensor<1x3x?x?xf32>
%0 = "tosa.reduce_sum"(%arg0) {axis = 0 : i64} : (tensor<2x3x?x?xf32>) -> tensor<?x?x?x?xf32>
@ -467,7 +467,7 @@ func @test_reduce_float(%arg0 : tensor<2x3x?x?xf32>) -> () {
// -----
// CHECK-LABEL: @test_concat
func @test_concat(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x2xf32>) -> () {
func.func @test_concat(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x2xf32>) -> () {
// CHECK: "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<1x2xf32>, tensor<2x2xf32>) -> tensor<3x2xf32>
%0 = "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<1x2xf32>, tensor<2x2xf32>) -> tensor<?x?xf32>
@ -477,7 +477,7 @@ func @test_concat(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x2xf32>) -> () {
// -----
// CHECK-LABEL: @test_concat_dynamic
func @test_concat_dynamic(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x?xf32>) -> () {
func.func @test_concat_dynamic(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x?xf32>) -> () {
// CHECK: "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<1x2xf32>, tensor<2x?xf32>) -> tensor<3x2xf32>
%0 = "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<1x2xf32>, tensor<2x?xf32>) -> tensor<?x?xf32>
@ -487,7 +487,7 @@ func @test_concat_dynamic(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x?xf32>) -> (
// -----
// CHECK-LABEL: @test_concat_dynamic_axis
func @test_concat_dynamic_axis(%arg0 : tensor<?x2xf32>, %arg1 : tensor<2x2xf32>) -> () {
func.func @test_concat_dynamic_axis(%arg0 : tensor<?x2xf32>, %arg1 : tensor<2x2xf32>) -> () {
// CHECK: "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<?x2xf32>, tensor<2x2xf32>) -> tensor<?x2xf32>
%0 = "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<?x2xf32>, tensor<2x2xf32>) -> tensor<?x?xf32>
@ -497,7 +497,7 @@ func @test_concat_dynamic_axis(%arg0 : tensor<?x2xf32>, %arg1 : tensor<2x2xf32>)
// -----
// CHECK-LABEL: @test_concat_axis_1
func @test_concat_axis_1(%arg0 : tensor<2x1xf32>, %arg1 : tensor<2x2xf32>) -> () {
func.func @test_concat_axis_1(%arg0 : tensor<2x1xf32>, %arg1 : tensor<2x2xf32>) -> () {
// CHECK: "tosa.concat"(%arg0, %arg1) {axis = 1 : i64} : (tensor<2x1xf32>, tensor<2x2xf32>) -> tensor<2x3xf32>
%0 = "tosa.concat"(%arg0, %arg1) {axis = 1 : i64} : (tensor<2x1xf32>, tensor<2x2xf32>) -> tensor<?x?xf32>
@ -507,7 +507,7 @@ func @test_concat_axis_1(%arg0 : tensor<2x1xf32>, %arg1 : tensor<2x2xf32>) -> ()
// -----
// CHECK-LABEL: @test_concat_failure
func @test_concat_failure(%arg0 : tensor<2x1xf32>, %arg1 : tensor<2x2xf32>) -> () {
func.func @test_concat_failure(%arg0 : tensor<2x1xf32>, %arg1 : tensor<2x2xf32>) -> () {
// CHECK: "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<2x1xf32>, tensor<2x2xf32>) -> tensor<?x?xf32>
%0 = "tosa.concat"(%arg0, %arg1) {axis = 0 : i64} : (tensor<2x1xf32>, tensor<2x2xf32>) -> tensor<?x?xf32>
@ -517,7 +517,7 @@ func @test_concat_failure(%arg0 : tensor<2x1xf32>, %arg1 : tensor<2x2xf32>) -> (
// -----
// CHECK-LABEL: @test_padding_no_const
func @test_padding_no_const(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x2xi32>) -> () {
func.func @test_padding_no_const(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x2xi32>) -> () {
// CHECK: "tosa.pad"(%arg0, %arg1) : (tensor<1x2xf32>, tensor<2x2xi32>) -> tensor<?x?xf32>
%0 = "tosa.pad"(%arg0, %arg1) : (tensor<1x2xf32>, tensor<2x2xi32>) -> (tensor<?x?xf32>)
return
@ -526,7 +526,7 @@ func @test_padding_no_const(%arg0 : tensor<1x2xf32>, %arg1 : tensor<2x2xi32>) ->
// -----
// CHECK-LABEL:@test_padding_dynamic_input
func @test_padding_dynamic_input(%arg0 : tensor<1x?xf32>) -> () {
func.func @test_padding_dynamic_input(%arg0 : tensor<1x?xf32>) -> () {
%0 = arith.constant dense<[[1, 2], [3, 4]]> : tensor<2x2xi32>
// CHECK: "tosa.pad"(%arg0, %cst) : (tensor<1x?xf32>, tensor<2x2xi32>) -> tensor<4x?xf32>
%1 = "tosa.pad"(%arg0, %0) : (tensor<1x?xf32>, tensor<2x2xi32>) -> (tensor<?x?xf32>)
@ -536,7 +536,7 @@ func @test_padding_dynamic_input(%arg0 : tensor<1x?xf32>) -> () {
// -----
// CHECK-LABEL: @test_padding_simple
func @test_padding_simple(%arg0 : tensor<1x2xf32>) -> () {
func.func @test_padding_simple(%arg0 : tensor<1x2xf32>) -> () {
%0 = arith.constant dense<[[1, 2], [3, 4]]> : tensor<2x2xi32>
// CHECK: "tosa.pad"(%arg0, %cst) : (tensor<1x2xf32>, tensor<2x2xi32>) -> tensor<4x9xf32>
%1 = "tosa.pad"(%arg0, %0) : (tensor<1x2xf32>, tensor<2x2xi32>) -> (tensor<?x?xf32>)
@ -546,7 +546,7 @@ func @test_padding_simple(%arg0 : tensor<1x2xf32>) -> () {
// -----
// CHECK-LABEL: @test_slice
func @test_slice(%arg0 : tensor<?xi32>) -> () {
func.func @test_slice(%arg0 : tensor<?xi32>) -> () {
// CHECK: "tosa.slice"(%arg0) {size = [2], start = [1]} : (tensor<?xi32>) -> tensor<2xi32>
%0 = "tosa.slice"(%arg0) { size = [2], start = [1] } : (tensor<?xi32>) -> tensor<?xi32>
return
@ -555,7 +555,7 @@ func @test_slice(%arg0 : tensor<?xi32>) -> () {
// -----
// CHECK-LABEL: @test_tile
func @test_tile(%arg0 : tensor<2x3x?xi32>) -> () {
func.func @test_tile(%arg0 : tensor<2x3x?xi32>) -> () {
// CHECK: "tosa.tile"(%arg0) {multiples = [2, 1, 5]} : (tensor<2x3x?xi32>) -> tensor<4x3x?xi32>
%0 = "tosa.tile"(%arg0) {multiples = [2, 1, 5]} : (tensor<2x3x?xi32>) -> (tensor<?x?x?xi32>)
return
@ -564,7 +564,7 @@ func @test_tile(%arg0 : tensor<2x3x?xi32>) -> () {
// -----
// CHECK-LABEL: @test_transpose_same
func @test_transpose_same(%arg0 : tensor<4x4x4xi32>, %arg1 : tensor<3xi32>) -> () {
func.func @test_transpose_same(%arg0 : tensor<4x4x4xi32>, %arg1 : tensor<3xi32>) -> () {
// CHECK: "tosa.transpose"(%arg0, %arg1) : (tensor<4x4x4xi32>, tensor<3xi32>) -> tensor<4x4x4xi32>
%0 = "tosa.transpose"(%arg0, %arg1) : (tensor<4x4x4xi32>, tensor<3xi32>) -> (tensor<?x?x?xi32>)
return
@ -573,7 +573,7 @@ func @test_transpose_same(%arg0 : tensor<4x4x4xi32>, %arg1 : tensor<3xi32>) -> (
// -----
// CHECK-LABEL: @test_transpose_perm_unknown
func @test_transpose_perm_unknown(%arg0 : tensor<4x4x5xi32>, %arg1 : tensor<3xi32>) -> () {
func.func @test_transpose_perm_unknown(%arg0 : tensor<4x4x5xi32>, %arg1 : tensor<3xi32>) -> () {
// CHECK: "tosa.transpose"(%arg0, %arg1) : (tensor<4x4x5xi32>, tensor<3xi32>) -> tensor<?x?x?xi32>
%0 = "tosa.transpose"(%arg0, %arg1) : (tensor<4x4x5xi32>, tensor<3xi32>) -> (tensor<?x?x?xi32>)
return
@ -582,7 +582,7 @@ func @test_transpose_perm_unknown(%arg0 : tensor<4x4x5xi32>, %arg1 : tensor<3xi3
// -----
// CHECK-LABEL: @test_transpose_static
func @test_transpose_static(%arg0 : tensor<3x4x5xi32>) -> () {
func.func @test_transpose_static(%arg0 : tensor<3x4x5xi32>) -> () {
%0 = arith.constant dense<[2, 1, 0]> : tensor<3xi32>
// CHECK: "tosa.transpose"(%arg0, %cst) : (tensor<3x4x5xi32>, tensor<3xi32>) -> tensor<5x4x3xi32>
%1 = "tosa.transpose"(%arg0, %0) : (tensor<3x4x5xi32>, tensor<3xi32>) -> (tensor<?x?x?xi32>)
@ -592,7 +592,7 @@ func @test_transpose_static(%arg0 : tensor<3x4x5xi32>) -> () {
// -----
// CHECK-LABEL: @gather_static
func @gather_static(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<3x6xi32>) {
func.func @gather_static(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<3x6xi32>) {
// CHECK: "tosa.gather"(%arg0, %arg1) : (tensor<3x4x5xi32>, tensor<3x6xi32>) -> tensor<3x6x5xi32>
%0 = "tosa.gather"(%arg0, %arg1) : (tensor<3x4x5xi32>, tensor<3x6xi32>) -> (tensor<?x?x?xi32>)
return
@ -601,7 +601,7 @@ func @gather_static(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<3x6xi32>) {
// -----
// CHECK-LABEL: @gather_dynamic_values
func @gather_dynamic_values(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<3x6xi32>) {
func.func @gather_dynamic_values(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<3x6xi32>) {
// CHECK: "tosa.gather"(%arg0, %arg1) : (tensor<?x?x?xi32>, tensor<3x6xi32>) -> tensor<3x6x?xi32>
%0 = "tosa.gather"(%arg0, %arg1) : (tensor<?x?x?xi32>, tensor<3x6xi32>) -> (tensor<?x?x?xi32>)
return
@ -610,7 +610,7 @@ func @gather_dynamic_values(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<3x6xi32>)
// -----
// CHECK-LABEL: @gather_dynamic_indices
func @gather_dynamic_indices(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<?x?xi32>) {
func.func @gather_dynamic_indices(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<?x?xi32>) {
// CHECK: "tosa.gather"(%arg0, %arg1) : (tensor<3x4x5xi32>, tensor<?x?xi32>) -> tensor<3x?x5xi32>
%0 = "tosa.gather"(%arg0, %arg1) : (tensor<3x4x5xi32>, tensor<?x?xi32>) -> (tensor<?x?x?xi32>)
return
@ -619,7 +619,7 @@ func @gather_dynamic_indices(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<?x?xi32>)
// -----
// CHECK-LABEL: @gather_minimum_info
func @gather_minimum_info(%arg0 : tensor<3x?x5xi32>, %arg1 : tensor<?x6xi32>) {
func.func @gather_minimum_info(%arg0 : tensor<3x?x5xi32>, %arg1 : tensor<?x6xi32>) {
// CHECK: "tosa.gather"(%arg0, %arg1) : (tensor<3x?x5xi32>, tensor<?x6xi32>) -> tensor<3x6x5xi32>
%0 = "tosa.gather"(%arg0, %arg1) : (tensor<3x?x5xi32>, tensor<?x6xi32>) -> (tensor<?x?x?xi32>)
return
@ -628,7 +628,7 @@ func @gather_minimum_info(%arg0 : tensor<3x?x5xi32>, %arg1 : tensor<?x6xi32>) {
// -----
// CHECK-LABEL: @scatter_static
func @scatter_static(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<3x6xi32>, %arg2 : tensor<3x6x5xi32>) {
func.func @scatter_static(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<3x6xi32>, %arg2 : tensor<3x6x5xi32>) {
// CHECK: "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<3x4x5xi32>, tensor<3x6xi32>, tensor<3x6x5xi32>) -> tensor<3x4x5xi32>
%0 = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<3x4x5xi32>, tensor<3x6xi32>, tensor<3x6x5xi32>) -> (tensor<?x?x?xi32>)
return
@ -637,7 +637,7 @@ func @scatter_static(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<3x6xi32>, %arg2 :
// -----
// CHECK-LABEL: @scatter_static_values
func @scatter_static_values(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<?x?xi32>, %arg2 : tensor<?x?x?xi32>) {
func.func @scatter_static_values(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<?x?xi32>, %arg2 : tensor<?x?x?xi32>) {
// CHECK: "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<3x4x5xi32>, tensor<?x?xi32>, tensor<?x?x?xi32>) -> tensor<3x4x5xi32>
%0 = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<3x4x5xi32>, tensor<?x?xi32>, tensor<?x?x?xi32>) -> (tensor<?x?x?xi32>)
return
@ -646,7 +646,7 @@ func @scatter_static_values(%arg0 : tensor<3x4x5xi32>, %arg1 : tensor<?x?xi32>,
// -----
// CHECK-LABEL: @scatter_static_indices
func @scatter_static_indices(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<3x6xi32>, %arg2 : tensor<?x?x?xi32>) {
func.func @scatter_static_indices(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<3x6xi32>, %arg2 : tensor<?x?x?xi32>) {
// CHECK: "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<?x?x?xi32>, tensor<3x6xi32>, tensor<?x?x?xi32>) -> tensor<3x?x?xi32>
%0 = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<?x?x?xi32>, tensor<3x6xi32>, tensor<?x?x?xi32>) -> (tensor<?x?x?xi32>)
return
@ -655,7 +655,7 @@ func @scatter_static_indices(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<3x6xi32>,
// -----
// CHECK-LABEL: @scatter_static_input
func @scatter_static_input(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<?x?xi32>, %arg2 : tensor<3x6x5xi32>) {
func.func @scatter_static_input(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<?x?xi32>, %arg2 : tensor<3x6x5xi32>) {
// CHECK: "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<?x?x?xi32>, tensor<?x?xi32>, tensor<3x6x5xi32>) -> tensor<3x?x5xi32>
%0 = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<?x?x?xi32>, tensor<?x?xi32>, tensor<3x6x5xi32>) -> (tensor<?x?x?xi32>)
return
@ -664,7 +664,7 @@ func @scatter_static_input(%arg0 : tensor<?x?x?xi32>, %arg1 : tensor<?x?xi32>, %
// -----
// CHECK-LABEL: @scatter_minimum_static
func @scatter_minimum_static(%arg0 : tensor<?x4x?xi32>, %arg1 : tensor<3x?xi32>, %arg2 : tensor<?x?x5xi32>) {
func.func @scatter_minimum_static(%arg0 : tensor<?x4x?xi32>, %arg1 : tensor<3x?xi32>, %arg2 : tensor<?x?x5xi32>) {
// CHECK: "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<?x4x?xi32>, tensor<3x?xi32>, tensor<?x?x5xi32>) -> tensor<3x4x5xi32>
%0 = "tosa.scatter"(%arg0, %arg1, %arg2) : (tensor<?x4x?xi32>, tensor<3x?xi32>, tensor<?x?x5xi32>) -> (tensor<?x?x?xi32>)
return
@ -673,7 +673,7 @@ func @scatter_minimum_static(%arg0 : tensor<?x4x?xi32>, %arg1 : tensor<3x?xi32>,
// -----
// CHECK-LABEL: @test_pool_static
func @test_pool_static(%arg0: tensor<3x5x6x7xf32>) {
func.func @test_pool_static(%arg0: tensor<3x5x6x7xf32>) {
// CHECK: -> tensor<3x2x4x7xf32>
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [4, 3], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<3x5x6x7xf32>) -> tensor<?x?x?x?xf32>
@ -685,7 +685,7 @@ func @test_pool_static(%arg0: tensor<3x5x6x7xf32>) {
// -----
// CHECK-LABEL: @conv2d_static
func @conv2d_static(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
func.func @conv2d_static(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
// CHECK: -> tensor<2x6x4x5xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -694,7 +694,7 @@ func @conv2d_static(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>,
// -----
// CHECK-LABEL: @conv2d_dynamic_input
func @conv2d_dynamic_input(%input: tensor<?x?x?x?xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
func.func @conv2d_dynamic_input(%input: tensor<?x?x?x?xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
// CHECK: -> tensor<?x?x?x5xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1]} : (tensor<?x?x?x?xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -703,7 +703,7 @@ func @conv2d_dynamic_input(%input: tensor<?x?x?x?xf32>, %weights: tensor<5x3x6x3
// -----
// CHECK-LABEL: @test_pool_dynamic_input
func @test_pool_dynamic_input(%arg0: tensor<?x?x?x?xf32>) {
func.func @test_pool_dynamic_input(%arg0: tensor<?x?x?x?xf32>) {
// CHECK: -> tensor<?x?x?x?xf32>
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [4, 3], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<?x?x?x?xf32>) -> tensor<?x?x?x?xf32>
@ -715,7 +715,7 @@ func @test_pool_dynamic_input(%arg0: tensor<?x?x?x?xf32>) {
// -----
// CHECK-LABEL: @test_pool_padded
func @test_pool_padded(%arg0: tensor<3x5x6x7xf32>) {
func.func @test_pool_padded(%arg0: tensor<3x5x6x7xf32>) {
// CHECK: -> tensor<3x5x11x7xf32>
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [4, 3], pad = [1, 2, 3, 4], stride = [1, 1]} : (tensor<3x5x6x7xf32>) -> tensor<?x?x?x?xf32>
@ -727,7 +727,7 @@ func @test_pool_padded(%arg0: tensor<3x5x6x7xf32>) {
// -----
// CHECK-LABEL: @conv2d_dynamic_weight
func @conv2d_dynamic_weight(%input: tensor<2x8x9x3xf32>, %weights: tensor<?x?x?x?xf32>, %bias: tensor<5xf32>) -> () {
func.func @conv2d_dynamic_weight(%input: tensor<2x8x9x3xf32>, %weights: tensor<?x?x?x?xf32>, %bias: tensor<5xf32>) -> () {
// CHECK: -> tensor<2x?x?x5xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<?x?x?x?xf32>, tensor<5xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -736,7 +736,7 @@ func @conv2d_dynamic_weight(%input: tensor<2x8x9x3xf32>, %weights: tensor<?x?x?x
// -----
// CHECK-LABEL: @conv2d_dynamic_bias
func @conv2d_dynamic_bias(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<?xf32>) -> () {
func.func @conv2d_dynamic_bias(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<?xf32>) -> () {
// CHECK: -> tensor<2x6x4x5xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<5x3x6x3xf32>, tensor<?xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -745,7 +745,7 @@ func @conv2d_dynamic_bias(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3x
// -----
// CHECK-LABEL: @test_pool_stride
func @test_pool_stride(%arg0: tensor<3x11x12x7xf32>) {
func.func @test_pool_stride(%arg0: tensor<3x11x12x7xf32>) {
// CHECK: -> tensor<3x4x4x7xf32>
%0 = "tosa.avg_pool2d"(%arg0) {kernel = [4, 3], pad = [0, 0, 0, 0], stride = [2, 3]} : (tensor<3x11x12x7xf32>) -> tensor<?x?x?x?xf32>
@ -757,7 +757,7 @@ func @test_pool_stride(%arg0: tensor<3x11x12x7xf32>) {
// -----
// CHECK-LABEL: @conv2d_padded
func @conv2d_padded(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
func.func @conv2d_padded(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
// CHECK: -> tensor<2x9x11x5xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [1, 2, 3, 4], stride = [1, 1], dilation = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -766,7 +766,7 @@ func @conv2d_padded(%input: tensor<2x8x9x3xf32>, %weights: tensor<5x3x6x3xf32>,
// -----
// CHECK-LABEL: @conv2d_dilated
func @conv2d_dilated(%input: tensor<2x12x14x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
func.func @conv2d_dilated(%input: tensor<2x12x14x3xf32>, %weights: tensor<5x3x6x3xf32>, %bias: tensor<5xf32>) -> () {
// CHECK: -> tensor<2x6x4x5xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [0, 0, 0, 0], stride = [1, 1], dilation = [3, 2]} : (tensor<2x12x14x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -775,7 +775,7 @@ func @conv2d_dilated(%input: tensor<2x12x14x3xf32>, %weights: tensor<5x3x6x3xf32
// -----
// CHECK-LABEL: @conv2d_strided
func @conv2d_strided(%input: tensor<1x13x14x1xf32>, %weights: tensor<1x1x1x1xf32>, %bias: tensor<1xf32>) -> () {
func.func @conv2d_strided(%input: tensor<1x13x14x1xf32>, %weights: tensor<1x1x1x1xf32>, %bias: tensor<1xf32>) -> () {
// CHECK: -> tensor<1x5x7x1xf32>
%0 = "tosa.conv2d"(%input, %weights, %bias) {pad = [0, 0, 0, 0], stride = [3, 2], dilation = [1, 1]} : (tensor<1x13x14x1xf32>, tensor<1x1x1x1xf32>, tensor<1xf32>) -> (tensor<?x?x?x?xf32>)
return
@ -784,7 +784,7 @@ func @conv2d_strided(%input: tensor<1x13x14x1xf32>, %weights: tensor<1x1x1x1xf32
// -----
// CHECK-LABEL: @conv3d_static
func @conv3d_static(%input: tensor<2x8x9x10x3xf32>, %weights: tensor<5x3x6x4x3xf32>, %bias: tensor<5xf32>) -> () {
func.func @conv3d_static(%input: tensor<2x8x9x10x3xf32>, %weights: tensor<5x3x6x4x3xf32>, %bias: tensor<5xf32>) -> () {
// CHECK: -> tensor<2x6x4x7x5xf32>
%0 = "tosa.conv3d"(%input, %weights, %bias) {dilation = [1, 1, 1], pad = [0, 0, 0, 0, 0, 0], stride = [1, 1, 1]} : (tensor<2x8x9x10x3xf32>, tensor<5x3x6x4x3xf32>, tensor<5xf32>) -> (tensor<?x?x?x?x?xf32>)
return
@ -793,7 +793,7 @@ func @conv3d_static(%input: tensor<2x8x9x10x3xf32>, %weights: tensor<5x3x6x4x3xf
// -----
// CHECK-LABEL: @conv3d_dynamic_input
func @conv3d_dynamic_input(%arg0: tensor<?x?x?x?x?xf32>, %arg1: tensor<5x3x6x4x3xf32>, %arg2: tensor<5xf32>) {
func.func @conv3d_dynamic_input(%arg0: tensor<?x?x?x?x?xf32>, %arg1: tensor<5x3x6x4x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<?x?x?x?x5xf32>
%0 = "tosa.conv3d"(%arg0, %arg1, %arg2) {dilation = [1, 1, 1], pad = [0, 0, 0, 0, 0, 0], stride = [1, 1, 1]} : (tensor<?x?x?x?x?xf32>, tensor<5x3x6x4x3xf32>, tensor<5xf32>) -> tensor<?x?x?x?x?xf32>
return
@ -802,7 +802,7 @@ func @conv3d_dynamic_input(%arg0: tensor<?x?x?x?x?xf32>, %arg1: tensor<5x3x6x4x3
// -----
// CHECK-LABEL: @conv3d_dynamic_weight
func @conv3d_dynamic_weight(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<?x?x?x?x?xf32>, %arg2: tensor<5xf32>) {
func.func @conv3d_dynamic_weight(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<?x?x?x?x?xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x?x?x?x5xf32>
%0 = "tosa.conv3d"(%arg0, %arg1, %arg2) {dilation = [1, 1, 1], pad = [0, 0, 0, 0, 0, 0], stride = [1, 1, 1]} : (tensor<2x8x9x10x3xf32>, tensor<?x?x?x?x?xf32>, tensor<5xf32>) -> tensor<?x?x?x?x?xf32>
return
@ -811,7 +811,7 @@ func @conv3d_dynamic_weight(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<?x?x?x?
// -----
// CHECK-LABEL: @conv3d_dynamic_bias
func @conv3d_dynamic_bias(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<5x3x6x4x3xf32>, %arg2: tensor<?xf32>) {
func.func @conv3d_dynamic_bias(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<5x3x6x4x3xf32>, %arg2: tensor<?xf32>) {
// CHECK: -> tensor<2x6x4x7x5xf32>
%0 = "tosa.conv3d"(%arg0, %arg1, %arg2) {dilation = [1, 1, 1], pad = [0, 0, 0, 0, 0, 0], stride = [1, 1, 1]} : (tensor<2x8x9x10x3xf32>, tensor<5x3x6x4x3xf32>, tensor<?xf32>) -> tensor<?x?x?x?x?xf32>
return
@ -820,7 +820,7 @@ func @conv3d_dynamic_bias(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<5x3x6x4x3
// -----
// CHECK-LABEL: @conv3d_padded
func @conv3d_padded(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<5x3x6x4x3xf32>, %arg2: tensor<5xf32>) {
func.func @conv3d_padded(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<5x3x6x4x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x9x11x18x5xf32>
%0 = "tosa.conv3d"(%arg0, %arg1, %arg2) {dilation = [1, 1, 1], pad = [1, 2, 3, 4, 5, 6], stride = [1, 1, 1]} : (tensor<2x8x9x10x3xf32>, tensor<5x3x6x4x3xf32>, tensor<5xf32>) -> tensor<?x?x?x?x?xf32>
return
@ -829,7 +829,7 @@ func @conv3d_padded(%arg0: tensor<2x8x9x10x3xf32>, %arg1: tensor<5x3x6x4x3xf32>,
// -----
// CHECK-LABEL: @conv3d_dilated
func @conv3d_dilated(%arg0: tensor<2x12x14x16x3xf32>, %arg1: tensor<5x3x6x2x3xf32>, %arg2: tensor<5xf32>) {
func.func @conv3d_dilated(%arg0: tensor<2x12x14x16x3xf32>, %arg1: tensor<5x3x6x2x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x6x4x12x5xf32>
%0 = "tosa.conv3d"(%arg0, %arg1, %arg2) {dilation = [3, 2, 4], pad = [0, 0, 0, 0, 0, 0], stride = [1, 1, 1]} : (tensor<2x12x14x16x3xf32>, tensor<5x3x6x2x3xf32>, tensor<5xf32>) -> tensor<?x?x?x?x?xf32>
return
@ -838,7 +838,7 @@ func @conv3d_dilated(%arg0: tensor<2x12x14x16x3xf32>, %arg1: tensor<5x3x6x2x3xf3
// -----
// CHECK-LABEL: @conv3d_strided
func @conv3d_strided(%arg0: tensor<1x13x14x15x1xf32>, %arg1: tensor<1x1x1x1x1xf32>, %arg2: tensor<1xf32>) {
func.func @conv3d_strided(%arg0: tensor<1x13x14x15x1xf32>, %arg1: tensor<1x1x1x1x1xf32>, %arg2: tensor<1xf32>) {
// CHECK: -> tensor<1x5x7x4x1xf32>
%0 = "tosa.conv3d"(%arg0, %arg1, %arg2) {dilation = [1, 1, 1], pad = [0, 0, 0, 0, 0, 0], stride = [3, 2, 4]} : (tensor<1x13x14x15x1xf32>, tensor<1x1x1x1x1xf32>, tensor<1xf32>) -> tensor<?x?x?x?x?xf32>
return
@ -847,7 +847,7 @@ func @conv3d_strided(%arg0: tensor<1x13x14x15x1xf32>, %arg1: tensor<1x1x1x1x1xf3
// -----
// CHECK-LABEL: @depthwise_conv2d_static
func @depthwise_conv2d_static(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
func.func @depthwise_conv2d_static(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
// CHECK: -> tensor<2x6x4x15xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<3x6x3x5xf32>, tensor<15xf32>) -> tensor<2x6x4x15xf32>
return
@ -856,7 +856,7 @@ func @depthwise_conv2d_static(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5x
// -----
// CHECK-LABEL: @depthwise_conv2d_dynamic_input
func @depthwise_conv2d_dynamic_input(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
func.func @depthwise_conv2d_dynamic_input(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
// CHECK: -> tensor<?x?x?x15xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<?x?x?x?xf32>, tensor<3x6x3x5xf32>, tensor<15xf32>) -> tensor<?x?x?x15xf32>
return
@ -865,7 +865,7 @@ func @depthwise_conv2d_dynamic_input(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<3
// -----
// CHECK-LABEL: @depthwise_conv2d_dynamic_weight
func @depthwise_conv2d_dynamic_weight(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<?x?x?x?xf32>, %arg2: tensor<15xf32>) {
func.func @depthwise_conv2d_dynamic_weight(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<?x?x?x?xf32>, %arg2: tensor<15xf32>) {
// CHECK: -> tensor<2x?x?x15xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<?x?x?x?xf32>, tensor<15xf32>) -> tensor<2x?x?x15xf32>
return
@ -874,7 +874,7 @@ func @depthwise_conv2d_dynamic_weight(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<
// -----
// CHECK-LABEL: @depthwise_conv2d_dynamic_bias
func @depthwise_conv2d_dynamic_bias(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<?xf32>) {
func.func @depthwise_conv2d_dynamic_bias(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<?xf32>) {
// CHECK: -> tensor<2x6x4x15xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<3x6x3x5xf32>, tensor<?xf32>) -> tensor<2x6x4x15xf32>
return
@ -883,7 +883,7 @@ func @depthwise_conv2d_dynamic_bias(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x
// -----
// CHECK-LABEL: @depthwise_conv2d_padded
func @depthwise_conv2d_padded(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
func.func @depthwise_conv2d_padded(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
// CHECK: -> tensor<2x9x11x15xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [1, 2, 3, 4], stride = [1, 1]} : (tensor<2x8x9x3xf32>, tensor<3x6x3x5xf32>, tensor<15xf32>) -> tensor<2x9x11x15xf32>
return
@ -892,7 +892,7 @@ func @depthwise_conv2d_padded(%arg0: tensor<2x8x9x3xf32>, %arg1: tensor<3x6x3x5x
// -----
// CHECK-LABEL: @depthwise_conv2d_dilated
func @depthwise_conv2d_dilated(%arg0: tensor<2x12x14x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
func.func @depthwise_conv2d_dilated(%arg0: tensor<2x12x14x3xf32>, %arg1: tensor<3x6x3x5xf32>, %arg2: tensor<15xf32>) {
// CHECK: -> tensor<2x6x4x15xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [3, 2], pad = [0, 0, 0, 0], stride = [1, 1]} : (tensor<2x12x14x3xf32>, tensor<3x6x3x5xf32>, tensor<15xf32>) -> tensor<2x6x4x15xf32>
return
@ -901,7 +901,7 @@ func @depthwise_conv2d_dilated(%arg0: tensor<2x12x14x3xf32>, %arg1: tensor<3x6x3
// -----
// CHECK-LABEL: @depthwise_conv2d_strided
func @depthwise_conv2d_strided(%arg0: tensor<1x13x14x1xf32>, %arg1: tensor<1x1x1x1xf32>, %arg2: tensor<1xf32>) {
func.func @depthwise_conv2d_strided(%arg0: tensor<1x13x14x1xf32>, %arg1: tensor<1x1x1x1xf32>, %arg2: tensor<1xf32>) {
// CHECK: -> tensor<1x5x7x1xf32>
%0 = "tosa.depthwise_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], pad = [0, 0, 0, 0], stride = [3, 2]} : (tensor<1x13x14x1xf32>, tensor<1x1x1x1xf32>, tensor<1xf32>) -> tensor<1x5x7x1xf32>
return
@ -910,7 +910,7 @@ func @depthwise_conv2d_strided(%arg0: tensor<1x13x14x1xf32>, %arg1: tensor<1x1x1
// -----
// CHECK-LABEL: @transpose_conv2d_out_shape
func @transpose_conv2d_out_shape(%arg0: tensor<2x?x?x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_out_shape(%arg0: tensor<2x?x?x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x8x9x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, 8, 9, -1], stride = [1, 1]} : (tensor<2x?x?x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<2x8x9x5xf32>
return
@ -919,7 +919,7 @@ func @transpose_conv2d_out_shape(%arg0: tensor<2x?x?x3xf32>, %arg1: tensor<5x3x6
// -----
// CHECK-LABEL: @transpose_conv2d_static
func @transpose_conv2d_static(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_static(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x18x19x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<2x16x14x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<2x?x?x5xf32>
return
@ -928,7 +928,7 @@ func @transpose_conv2d_static(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x
// -----
// CHECK-LABEL: @transpose_conv2d_static_dilated
func @transpose_conv2d_static_dilated(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_static_dilated(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x20x29x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [2, 3], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<2x16x14x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<2x?x?x5xf32>
return
@ -937,7 +937,7 @@ func @transpose_conv2d_static_dilated(%arg0: tensor<2x16x14x3xf32>, %arg1: tenso
// -----
// CHECK-LABEL: @transpose_conv2d_static_strided
func @transpose_conv2d_static_strided(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_static_strided(%arg0: tensor<2x16x14x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x33x45x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [2, 3]} : (tensor<2x16x14x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<2x?x?x5xf32>
return
@ -946,7 +946,7 @@ func @transpose_conv2d_static_strided(%arg0: tensor<2x16x14x3xf32>, %arg1: tenso
// -----
// CHECK-LABEL: @transpose_conv2d_dynamic_input
func @transpose_conv2d_dynamic_input(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_dynamic_input(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<?x?x?x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<?x?x?x?xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<?x?x?x5xf32>
return
@ -955,7 +955,7 @@ func @transpose_conv2d_dynamic_input(%arg0: tensor<?x?x?x?xf32>, %arg1: tensor<5
// -----
// CHECK-LABEL: @transpose_conv2d_dynamic_weights
func @transpose_conv2d_dynamic_weights(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<?x?x?x?xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_dynamic_weights(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<?x?x?x?xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x?x?x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<2x6x4x3xf32>, tensor<?x?x?x?xf32>, tensor<5xf32>) -> tensor<2x?x?x5xf32>
return
@ -964,7 +964,7 @@ func @transpose_conv2d_dynamic_weights(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor
// -----
// CHECK-LABEL: @transpose_conv2d_dynamic_bias
func @transpose_conv2d_dynamic_bias(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<?xf32>) {
func.func @transpose_conv2d_dynamic_bias(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<?xf32>) {
// CHECK: -> tensor<2x8x9x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<2x6x4x3xf32>, tensor<5x3x6x3xf32>, tensor<?xf32>) -> tensor<2x8x9x5xf32>
return
@ -973,7 +973,7 @@ func @transpose_conv2d_dynamic_bias(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<5x
// -----
// CHECK-LABEL: @transpose_conv2d_padded
func @transpose_conv2d_padded(%arg0: tensor<2x9x11x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_padded(%arg0: tensor<2x9x11x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x10x13x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [1, 3], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<2x9x11x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<2x10x13x5xf32>
return
@ -982,7 +982,7 @@ func @transpose_conv2d_padded(%arg0: tensor<2x9x11x3xf32>, %arg1: tensor<5x3x6x3
// -----
// CHECK-LABEL: @transpose_conv2d_dilated
func @transpose_conv2d_dilated(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
func.func @transpose_conv2d_dilated(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<5x3x6x3xf32>, %arg2: tensor<5xf32>) {
// CHECK: -> tensor<2x12x14x5xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [3, 2], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [1, 1]} : (tensor<2x6x4x3xf32>, tensor<5x3x6x3xf32>, tensor<5xf32>) -> tensor<2x12x14x5xf32>
return
@ -991,7 +991,7 @@ func @transpose_conv2d_dilated(%arg0: tensor<2x6x4x3xf32>, %arg1: tensor<5x3x6x3
// -----
// CHECK-LABEL: @transpose_conv2d_strided
func @transpose_conv2d_strided(%arg0: tensor<1x5x7x1xf32>, %arg1: tensor<1x1x1x1xf32>, %arg2: tensor<1xf32>) {
func.func @transpose_conv2d_strided(%arg0: tensor<1x5x7x1xf32>, %arg1: tensor<1x1x1x1xf32>, %arg2: tensor<1xf32>) {
// CHECK: -> tensor<1x13x13x1xf32>
%0 = "tosa.transpose_conv2d"(%arg0, %arg1, %arg2) {dilation = [1, 1], out_pad = [0, 0], out_shape = [-1, -1, -1, -1], stride = [3, 2]} : (tensor<1x5x7x1xf32>, tensor<1x1x1x1xf32>, tensor<1xf32>) -> tensor<1x13x13x1xf32>
return
@ -1000,7 +1000,7 @@ func @transpose_conv2d_strided(%arg0: tensor<1x5x7x1xf32>, %arg1: tensor<1x1x1x1
// -----
// CHECK-LABEL: @resize_output_size
func @resize_output_size(%arg0: tensor<2x?x?x3xi32>) {
func.func @resize_output_size(%arg0: tensor<2x?x?x3xi32>) {
// CHECK: -> tensor<2x4x5x3xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [0, 1], offset_fp = [0.000000e+00 : f32, 0.000000e+00 : f32], output_size = [4, 5], shift = 8 : i32, stride = [1, 1], stride_fp = [0.000000e+00 : f32, 0.000000e+00 : f32]} : (tensor<2x?x?x3xi32>) -> tensor<?x?x?x?xi32>
return
@ -1009,7 +1009,7 @@ func @resize_output_size(%arg0: tensor<2x?x?x3xi32>) {
// -----
// CHECK-LABEL: @resize_int_horizontal
func @resize_int_horizontal(%arg0: tensor<1x2x4x1xi32>) {
func.func @resize_int_horizontal(%arg0: tensor<1x2x4x1xi32>) {
// CHECK: -> tensor<1x2x7x1xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [0, 0], offset_fp = [0.000000e+00 : f32, 0.000000e+00 : f32], output_size = [-1, -1], shift = 8 : i32, stride = [256, 128], stride_fp = [0.000000e+00 : f32, 0.000000e+00 : f32]} : (tensor<1x2x4x1xi32>) -> tensor<?x?x?x?xi32>
return
@ -1018,7 +1018,7 @@ func @resize_int_horizontal(%arg0: tensor<1x2x4x1xi32>) {
// -----
// CHECK-LABEL: @resize_int_vertical
func @resize_int_vertical(%arg0: tensor<1x2x4x1xi32>) {
func.func @resize_int_vertical(%arg0: tensor<1x2x4x1xi32>) {
// CHECK: -> tensor<1x3x4x1xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [0, 0], offset_fp = [0.000000e+00 : f32, 0.000000e+00 : f32], output_size = [-1, -1], shift = 8 : i32, stride = [128, 256], stride_fp = [0.000000e+00 : f32, 0.000000e+00 : f32]} : (tensor<1x2x4x1xi32>) -> tensor<?x?x?x?xi32>
return
@ -1027,7 +1027,7 @@ func @resize_int_vertical(%arg0: tensor<1x2x4x1xi32>) {
// -----
// CHECK-LABEL: @resize_int_offsetted
func @resize_int_offsetted(%arg0: tensor<1x2x4x1xi32>) {
func.func @resize_int_offsetted(%arg0: tensor<1x2x4x1xi32>) {
// CHECK: -> tensor<1x4x6x1xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [64, 64], offset_fp = [0.000000e+00 : f32, 0.000000e+00 : f32], output_size = [-1, -1], shift = 8 : i32, stride = [64, 128], stride_fp = [0.000000e+00 : f32, 0.000000e+00 : f32]} : (tensor<1x2x4x1xi32>) -> tensor<?x?x?x?xi32>
return
@ -1036,7 +1036,7 @@ func @resize_int_offsetted(%arg0: tensor<1x2x4x1xi32>) {
// -----
// CHECK-LABEL: @resize_fp_horizontal
func @resize_fp_horizontal(%arg0: tensor<1x2x4x1xi32>) {
func.func @resize_fp_horizontal(%arg0: tensor<1x2x4x1xi32>) {
// CHECK: -> tensor<1x2x7x1xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [0, 0], offset_fp = [0.000000e+00 : f32, 0.000000e+00 : f32], output_size = [-1, -1], shift = 0 : i32, stride = [0, 0], stride_fp = [1.000000e+00 : f32, 5.000000e-01 : f32]} : (tensor<1x2x4x1xi32>) -> tensor<?x?x?x?xi32>
return
@ -1045,13 +1045,13 @@ func @resize_fp_horizontal(%arg0: tensor<1x2x4x1xi32>) {
// -----
// CHECK-LABEL: @resize_fp_vertical
func @resize_fp_vertical(%arg0: tensor<1x2x4x1xi32>) {
func.func @resize_fp_vertical(%arg0: tensor<1x2x4x1xi32>) {
// CHECK: -> tensor<1x3x4x1xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [0, 0], offset_fp = [0.000000e+00 : f32, 0.000000e+00 : f32], output_size = [-1, -1], shift = 0 : i32, stride = [0, 0], stride_fp = [5.000000e-01 : f32, 1.000000e+00 : f32]} : (tensor<1x2x4x1xi32>) -> tensor<?x?x?x?xi32>
return
}
// CHECK-LABEL: @resize_fp_offsetted
func @resize_fp_offsetted(%arg0: tensor<1x2x4x1xi32>) {
func.func @resize_fp_offsetted(%arg0: tensor<1x2x4x1xi32>) {
// CHECK: -> tensor<1x4x6x1xi32>
%0 = "tosa.resize"(%arg0) {mode = "NEAREST_NEIGHBOR", offset = [0, 0], offset_fp = [2.500000e-01 : f32, 2.500000e-01 : f32], output_size = [-1, -1], shift = 0 : i32, stride = [0, 0], stride_fp = [2.500000e-01 : f32, 5.000000e-01 : f32]} : (tensor<1x2x4x1xi32>) -> tensor<?x?x?x?xi32>
return
@ -1060,7 +1060,7 @@ func @resize_fp_offsetted(%arg0: tensor<1x2x4x1xi32>) {
// -----
// CHECK-LABEL: @if_test_simple
func @if_test_simple(%arg0 : tensor<f32>, %arg1 : tensor<f32>, %arg2 : tensor<i1>) -> () {
func.func @if_test_simple(%arg0 : tensor<f32>, %arg1 : tensor<f32>, %arg2 : tensor<i1>) -> () {
%a = "tosa.log"(%arg0) : (tensor<f32>) -> tensor<*xf32>
%b = "tosa.log"(%arg1) : (tensor<f32>) -> tensor<*xf32>
// CHECK: (tensor<i1>, tensor<f32>, tensor<f32>) -> tensor<f32>
@ -1077,7 +1077,7 @@ func @if_test_simple(%arg0 : tensor<f32>, %arg1 : tensor<f32>, %arg2 : tensor<i1
// -----
// CHECK-LABEL: @if_test_dynamic
func @if_test_dynamic(%arg0 : tensor<2xf32>, %arg1 : tensor<3xf32>, %arg2 : tensor<i1>) -> () {
func.func @if_test_dynamic(%arg0 : tensor<2xf32>, %arg1 : tensor<3xf32>, %arg2 : tensor<i1>) -> () {
// CHECK: (tensor<i1>, tensor<2xf32>, tensor<3xf32>) -> tensor<?xf32>
%0 = "tosa.cond_if"(%arg2, %arg0, %arg1) ({
^bb1(%arg3 : tensor<2xf32>, %arg4 : tensor<3xf32>):
@ -1092,7 +1092,7 @@ func @if_test_dynamic(%arg0 : tensor<2xf32>, %arg1 : tensor<3xf32>, %arg2 : tens
// -----
// CHECK-LABEL: @if_test_unranked
func @if_test_unranked(%arg0 : tensor<f32>, %arg1 : tensor<3xf32>, %arg2 : tensor<i1>) -> () {
func.func @if_test_unranked(%arg0 : tensor<f32>, %arg1 : tensor<3xf32>, %arg2 : tensor<i1>) -> () {
// CHECK: (tensor<i1>, tensor<f32>, tensor<3xf32>) -> tensor<*xf32>
%0 = "tosa.cond_if"(%arg2, %arg0, %arg1) ({
^bb1(%arg3 : tensor<f32>, %arg4 : tensor<3xf32>):
@ -1107,7 +1107,7 @@ func @if_test_unranked(%arg0 : tensor<f32>, %arg1 : tensor<3xf32>, %arg2 : tenso
// -----
// CHECK-LABEL: @if_test_propagate
func @if_test_propagate(%arg0 : tensor<f32>, %arg1 : tensor<f32>, %arg2 : tensor<i1>) -> () {
func.func @if_test_propagate(%arg0 : tensor<f32>, %arg1 : tensor<f32>, %arg2 : tensor<i1>) -> () {
// CHECK: (tensor<i1>, tensor<f32>, tensor<f32>) -> tensor<f32>
%0 = "tosa.cond_if"(%arg2, %arg0, %arg1) ({
^bb1(%arg3 : tensor<*xf32>, %arg4 : tensor<*xf32>):
@ -1124,7 +1124,7 @@ func @if_test_propagate(%arg0 : tensor<f32>, %arg1 : tensor<f32>, %arg2 : tensor
// -----
// CHECK-LABEL: @while_test
func @while_test(%arg0 : tensor<i32>) -> (tensor<*xi32>) {
func.func @while_test(%arg0 : tensor<i32>) -> (tensor<*xi32>) {
// CHECK: "tosa.add"
// CHECK-SAME: (tensor<i32>, tensor<i32>) -> tensor<i32>
%0 = "tosa.add"(%arg0, %arg0) : (tensor<i32>, tensor<i32>) -> tensor<*xi32>
@ -1166,7 +1166,7 @@ func @while_test(%arg0 : tensor<i32>) -> (tensor<*xi32>) {
// -----
// CHECK-LABEL: @while_test
func @while_test(%arg0 : tensor<i32>, %arg1 : tensor<1xi32>) -> () {
func.func @while_test(%arg0 : tensor<i32>, %arg1 : tensor<1xi32>) -> () {
// CHECK: "tosa.while_loop"
%1:2 = "tosa.while_loop"(%arg0, %arg1) ({

4
mlir/test/Dialect/Vector/bufferize.mlir
View File

@ -5,7 +5,7 @@
// CHECK: %[[m:.*]] = bufferization.to_memref %[[t]] : memref<?x?xf32>
// CHECK: %[[r:.*]] = vector.transfer_read %[[m]][%[[o1]], %[[o2]]], %[[pad]] {in_bounds = [true, false]} : memref<?x?xf32>, vector<5x6xf32>
// CHECK: return %[[r]]
func @transfer_read(%t: tensor<?x?xf32>, %o1: index,
func.func @transfer_read(%t: tensor<?x?xf32>, %o1: index,
%o2: index, %pad: f32) -> vector<5x6xf32> {
%0 = vector.transfer_read %t[%o1, %o2], %pad {in_bounds = [true, false]}
: tensor<?x?xf32>, vector<5x6xf32>
@ -22,7 +22,7 @@ func @transfer_read(%t: tensor<?x?xf32>, %o1: index,
// CHECK: vector.transfer_write %[[vec]], %[[alloc]][%[[o1]], %[[o2]]] {in_bounds = [true, false]} : vector<5x6xf32>, memref<?x?xf32>
// CHECK: %[[r:.*]] = bufferization.to_tensor %[[alloc]] : memref<?x?xf32>
// CHECK: return %[[r]]
func @transfer_write(%t: tensor<?x?xf32>, %o1: index,
func.func @transfer_write(%t: tensor<?x?xf32>, %o1: index,
%o2: index, %vec: vector<5x6xf32>) -> tensor<?x?xf32> {
%0 = vector.transfer_write %vec, %t[%o1, %o2] {in_bounds = [true, false]}
: vector<5x6xf32>, tensor<?x?xf32>

188
mlir/test/Dialect/Vector/canonicalize.mlir
View File

@ -3,7 +3,7 @@
// -----
// CHECK-LABEL: create_vector_mask_to_constant_mask
func @create_vector_mask_to_constant_mask() -> (vector<4x3xi1>) {
func.func @create_vector_mask_to_constant_mask() -> (vector<4x3xi1>) {
%c2 = arith.constant 2 : index
%c3 = arith.constant 3 : index
// CHECK: vector.constant_mask [3, 2] : vector<4x3xi1>
@ -14,7 +14,7 @@ func @create_vector_mask_to_constant_mask() -> (vector<4x3xi1>) {
// -----
// CHECK-LABEL: create_scalable_vector_mask_to_constant_mask
func @create_scalable_vector_mask_to_constant_mask() -> (vector<[8]xi1>) {
func.func @create_scalable_vector_mask_to_constant_mask() -> (vector<[8]xi1>) {
%c-1 = arith.constant -1 : index
// CHECK: vector.constant_mask [0] : vector<[8]xi1>
%0 = vector.create_mask %c-1 : vector<[8]xi1>
@ -24,7 +24,7 @@ func @create_scalable_vector_mask_to_constant_mask() -> (vector<[8]xi1>) {
// -----
// CHECK-LABEL: create_vector_mask_to_constant_mask_truncation
func @create_vector_mask_to_constant_mask_truncation() -> (vector<4x3xi1>) {
func.func @create_vector_mask_to_constant_mask_truncation() -> (vector<4x3xi1>) {
%c2 = arith.constant 2 : index
%c5 = arith.constant 5 : index
// CHECK: vector.constant_mask [4, 2] : vector<4x3xi1>
@ -35,7 +35,7 @@ func @create_vector_mask_to_constant_mask_truncation() -> (vector<4x3xi1>) {
// -----
// CHECK-LABEL: create_vector_mask_to_constant_mask_truncation_neg
func @create_vector_mask_to_constant_mask_truncation_neg() -> (vector<4x3xi1>) {
func.func @create_vector_mask_to_constant_mask_truncation_neg() -> (vector<4x3xi1>) {
%cneg2 = arith.constant -2 : index
%c5 = arith.constant 5 : index
// CHECK: vector.constant_mask [0, 0] : vector<4x3xi1>
@ -46,7 +46,7 @@ func @create_vector_mask_to_constant_mask_truncation_neg() -> (vector<4x3xi1>) {
// -----
// CHECK-LABEL: create_vector_mask_to_constant_mask_truncation_zero
func @create_vector_mask_to_constant_mask_truncation_zero() -> (vector<4x3xi1>) {
func.func @create_vector_mask_to_constant_mask_truncation_zero() -> (vector<4x3xi1>) {
%c2 = arith.constant 2 : index
%c0 = arith.constant 0 : index
// CHECK: vector.constant_mask [0, 0] : vector<4x3xi1>
@ -56,7 +56,7 @@ func @create_vector_mask_to_constant_mask_truncation_zero() -> (vector<4x3xi1>)
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [0, 0], sizes = [2, 2], strides = [1, 1]}
@ -67,7 +67,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [1, 0], sizes = [2, 2], strides = [1, 1]}
@ -78,7 +78,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [0, 1], sizes = [2, 2], strides = [1, 1]}
@ -89,7 +89,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [2, 0], sizes = [2, 2], strides = [1, 1]}
@ -100,7 +100,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x2xi1>) {
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [0, 2], sizes = [2, 1], strides = [1, 1]}
@ -111,7 +111,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [0, 1], sizes = [2, 1], strides = [1, 1]}
@ -122,7 +122,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
// -----
func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
func.func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
%0 = vector.constant_mask [2, 2] : vector<4x3xi1>
%1 = vector.extract_strided_slice %0
{offsets = [1, 1], sizes = [2, 1], strides = [1, 1]}
@ -136,7 +136,7 @@ func @extract_strided_slice_of_constant_mask() -> (vector<2x1xi1>) {
// CHECK-LABEL: extract_strided_fold
// CHECK-SAME: (%[[ARG:.*]]: vector<4x3xi1>)
// CHECK-NEXT: return %[[ARG]] : vector<4x3xi1>
func @extract_strided_fold(%arg : vector<4x3xi1>) -> (vector<4x3xi1>) {
func.func @extract_strided_fold(%arg : vector<4x3xi1>) -> (vector<4x3xi1>) {
%0 = vector.extract_strided_slice %arg
{offsets = [0, 0], sizes = [4, 3], strides = [1, 1]}
: vector<4x3xi1> to vector<4x3xi1>
@ -148,7 +148,7 @@ func @extract_strided_fold(%arg : vector<4x3xi1>) -> (vector<4x3xi1>) {
// CHECK-LABEL: extract_strided_fold_insert
// CHECK-SAME: (%[[ARG:.*]]: vector<4x4xf32>
// CHECK-NEXT: return %[[ARG]] : vector<4x4xf32>
func @extract_strided_fold_insert(%a: vector<4x4xf32>, %b: vector<8x16xf32>)
func.func @extract_strided_fold_insert(%a: vector<4x4xf32>, %b: vector<8x16xf32>)
-> (vector<4x4xf32>) {
%0 = vector.insert_strided_slice %a, %b {offsets = [2, 2], strides = [1, 1]}
: vector<4x4xf32> into vector<8x16xf32>
@ -167,7 +167,7 @@ func @extract_strided_fold_insert(%a: vector<4x4xf32>, %b: vector<8x16xf32>)
// CHECK-SAME: {offsets = [0, 0], sizes = [4, 4], strides = [1, 1]}
// CHECK-SAME: : vector<6x4xf32> to vector<4x4xf32>
// CHECK-NEXT: return %[[EXT]] : vector<4x4xf32>
func @extract_strided_fold_insert(%a: vector<6x4xf32>, %b: vector<8x16xf32>)
func.func @extract_strided_fold_insert(%a: vector<6x4xf32>, %b: vector<8x16xf32>)
-> (vector<4x4xf32>) {
%0 = vector.insert_strided_slice %a, %b {offsets = [2, 2], strides = [1, 1]}
: vector<6x4xf32> into vector<8x16xf32>
@ -189,7 +189,7 @@ func @extract_strided_fold_insert(%a: vector<6x4xf32>, %b: vector<8x16xf32>)
// CHECK-SAME: {offsets = [2, 2], sizes = [6, 4], strides = [1, 1]}
// CHECK-SAME: : vector<8x16xf32> to vector<6x4xf32>
// CHECK-NEXT: return %[[EXT]] : vector<6x4xf32>
func @extract_strided_fold_negative(%a: vector<4x4xf32>, %b: vector<8x16xf32>)
func.func @extract_strided_fold_negative(%a: vector<4x4xf32>, %b: vector<8x16xf32>)
-> (vector<6x4xf32>) {
%0 = vector.insert_strided_slice %a, %b {offsets = [2, 2], strides = [1, 1]}
: vector<4x4xf32> into vector<8x16xf32>
@ -208,7 +208,7 @@ func @extract_strided_fold_negative(%a: vector<4x4xf32>, %b: vector<8x16xf32>)
// CHECK-SAME: {offsets = [0, 0], sizes = [1, 1], strides = [1, 1]}
// CHECK-SAME: : vector<1x4xf32> to vector<1x1xf32>
// CHECK-NEXT: return %[[EXT]] : vector<1x1xf32>
func @extract_strided_fold_insert(%a: vector<2x8xf32>, %b: vector<1x4xf32>,
func.func @extract_strided_fold_insert(%a: vector<2x8xf32>, %b: vector<1x4xf32>,
%c : vector<1x4xf32>) -> (vector<1x1xf32>) {
%0 = vector.insert_strided_slice %b, %a {offsets = [0, 1], strides = [1, 1]}
: vector<1x4xf32> into vector<2x8xf32>
@ -224,7 +224,7 @@ func @extract_strided_fold_insert(%a: vector<2x8xf32>, %b: vector<1x4xf32>,
// CHECK-LABEL: transpose_1D_identity
// CHECK-SAME: ([[ARG:%.*]]: vector<4xf32>)
func @transpose_1D_identity(%arg : vector<4xf32>) -> vector<4xf32> {
func.func @transpose_1D_identity(%arg : vector<4xf32>) -> vector<4xf32> {
// CHECK-NOT: transpose
%0 = vector.transpose %arg, [0] : vector<4xf32> to vector<4xf32>
// CHECK-NEXT: return [[ARG]]
@ -235,7 +235,7 @@ func @transpose_1D_identity(%arg : vector<4xf32>) -> vector<4xf32> {
// CHECK-LABEL: transpose_2D_identity
// CHECK-SAME: ([[ARG:%.*]]: vector<4x3xf32>)
func @transpose_2D_identity(%arg : vector<4x3xf32>) -> vector<4x3xf32> {
func.func @transpose_2D_identity(%arg : vector<4x3xf32>) -> vector<4x3xf32> {
// CHECK-NOT: transpose
%0 = vector.transpose %arg, [0, 1] : vector<4x3xf32> to vector<4x3xf32>
// CHECK-NEXT: return [[ARG]]
@ -246,7 +246,7 @@ func @transpose_2D_identity(%arg : vector<4x3xf32>) -> vector<4x3xf32> {
// CHECK-LABEL: transpose_3D_identity
// CHECK-SAME: ([[ARG:%.*]]: vector<4x3x2xf32>)
func @transpose_3D_identity(%arg : vector<4x3x2xf32>) -> vector<4x3x2xf32> {
func.func @transpose_3D_identity(%arg : vector<4x3x2xf32>) -> vector<4x3x2xf32> {
// CHECK-NOT: transpose
%0 = vector.transpose %arg, [0, 1, 2] : vector<4x3x2xf32> to vector<4x3x2xf32>
// CHECK-NEXT: return [[ARG]]
@ -257,7 +257,7 @@ func @transpose_3D_identity(%arg : vector<4x3x2xf32>) -> vector<4x3x2xf32> {
// CHECK-LABEL: transpose_2D_sequence
// CHECK-SAME: ([[ARG:%.*]]: vector<4x3xf32>)
func @transpose_2D_sequence(%arg : vector<4x3xf32>) -> vector<4x3xf32> {
func.func @transpose_2D_sequence(%arg : vector<4x3xf32>) -> vector<4x3xf32> {
// CHECK-NOT: transpose
%0 = vector.transpose %arg, [1, 0] : vector<4x3xf32> to vector<3x4xf32>
%1 = vector.transpose %0, [0, 1] : vector<3x4xf32> to vector<3x4xf32>
@ -273,7 +273,7 @@ func @transpose_2D_sequence(%arg : vector<4x3xf32>) -> vector<4x3xf32> {
// CHECK-LABEL: transpose_3D_sequence
// CHECK-SAME: ([[ARG:%.*]]: vector<4x3x2xf32>)
func @transpose_3D_sequence(%arg : vector<4x3x2xf32>) -> vector<4x3x2xf32> {
func.func @transpose_3D_sequence(%arg : vector<4x3x2xf32>) -> vector<4x3x2xf32> {
// CHECK: [[T0:%.*]] = vector.transpose [[ARG]], [2, 1, 0]
%0 = vector.transpose %arg, [1, 2, 0] : vector<4x3x2xf32> to vector<3x2x4xf32>
%1 = vector.transpose %0, [1, 0, 2] : vector<3x2x4xf32> to vector<2x3x4xf32>
@ -295,7 +295,7 @@ func @transpose_3D_sequence(%arg : vector<4x3x2xf32>) -> vector<4x3x2xf32> {
// -----
// CHECK-LABEL: cast_transfers
func @cast_transfers(%A: memref<4x8xf32>) -> (vector<4x8xf32>) {
func.func @cast_transfers(%A: memref<4x8xf32>) -> (vector<4x8xf32>) {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
%0 = memref.cast %A : memref<4x8xf32> to memref<?x?xf32>
@ -311,7 +311,7 @@ func @cast_transfers(%A: memref<4x8xf32>) -> (vector<4x8xf32>) {
// -----
// CHECK-LABEL: cast_transfers
func @cast_transfers(%A: tensor<4x8xf32>) -> (vector<4x8xf32>) {
func.func @cast_transfers(%A: tensor<4x8xf32>) -> (vector<4x8xf32>) {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
%0 = tensor.cast %A : tensor<4x8xf32> to tensor<?x?xf32>
@ -330,7 +330,7 @@ func @cast_transfers(%A: tensor<4x8xf32>) -> (vector<4x8xf32>) {
// CHECK-SAME: %[[F1:[a-zA-Z0-9]*]]: f32,
// CHECK-SAME: %[[F2:[a-zA-Z0-9]*]]: f32,
// CHECK-SAME: %[[F3:[a-zA-Z0-9]*]]: f32
func @insert_extract_transpose_2d(
func.func @insert_extract_transpose_2d(
%v: vector<2x3xf32>, %f0: f32, %f1: f32, %f2: f32, %f3: f32)
-> (f32, f32, f32)
{
@ -363,7 +363,7 @@ func @insert_extract_transpose_2d(
// CHECK-SAME: %[[V234:[a-zA-Z0-9]*]]: vector<2x3x4xf32>
// CHECK-SAME: %[[V34:[a-zA-Z0-9]*]]: vector<3x4xf32>
// CHECK-SAME: %[[V4:[a-zA-Z0-9]*]]: vector<4xf32>
func @insert_extract_chain(%v234: vector<2x3x4xf32>, %v34: vector<3x4xf32>, %v4: vector<4xf32>)
func.func @insert_extract_chain(%v234: vector<2x3x4xf32>, %v34: vector<3x4xf32>, %v4: vector<4xf32>)
-> (vector<4xf32>, vector<4xf32>, vector<3x4xf32>, vector<3x4xf32>) {
// CHECK-NEXT: %[[A34:.*]] = vector.insert
%A34 = vector.insert %v34, %v234[0]: vector<3x4xf32> into vector<2x3x4xf32>
@ -404,7 +404,7 @@ func @insert_extract_chain(%v234: vector<2x3x4xf32>, %v34: vector<3x4xf32>, %v4:
// CHECK-LABEL: func @insert_extract_transpose_3d(
// CHECK-SAME: %[[V234:[a-zA-Z0-9]*]]: vector<2x3x4xf32>
func @insert_extract_transpose_3d(
func.func @insert_extract_transpose_3d(
%v234: vector<2x3x4xf32>, %v43: vector<4x3xf32>, %f0: f32)
-> (vector<4xf32>, vector<4xf32>, vector<4xf32>, vector<3x4xf32>) {
@ -462,7 +462,7 @@ func @insert_extract_transpose_3d(
// CHECK-LABEL: fold_extracts
// CHECK-SAME: %[[A:[a-zA-Z0-9]*]]: vector<3x4x5x6xf32>
func @fold_extracts(%a : vector<3x4x5x6xf32>) -> (f32, vector<4x5x6xf32>) {
func.func @fold_extracts(%a : vector<3x4x5x6xf32>) -> (f32, vector<4x5x6xf32>) {
%b = vector.extract %a[0] : vector<3x4x5x6xf32>
%c = vector.extract %b[1, 2] : vector<4x5x6xf32>
// CHECK-NEXT: vector.extract %[[A]][0, 1, 2, 3] : vector<3x4x5x6xf32>
@ -480,7 +480,7 @@ func @fold_extracts(%a : vector<3x4x5x6xf32>) -> (f32, vector<4x5x6xf32>) {
// CHECK-LABEL: fold_extract_transpose
// CHECK-SAME: %[[A:[a-zA-Z0-9]*]]: vector<3x4x5x6xf32>
// CHECK-SAME: %[[B:[a-zA-Z0-9]*]]: vector<3x6x5x6xf32>
func @fold_extract_transpose(
func.func @fold_extract_transpose(
%a : vector<3x4x5x6xf32>, %b : vector<3x6x5x6xf32>) -> (
vector<6xf32>, vector<6xf32>, vector<6xf32>) {
// [3] is a proper most minor identity map in transpose.
@ -513,7 +513,7 @@ func @fold_extract_transpose(
// CHECK-LABEL: fold_extract_broadcast
// CHECK-SAME: %[[A:.*]]: f32
// CHECK: return %[[A]] : f32
func @fold_extract_broadcast(%a : f32) -> f32 {
func.func @fold_extract_broadcast(%a : f32) -> f32 {
%b = vector.broadcast %a : f32 to vector<1x2x4xf32>
%r = vector.extract %b[0, 1, 2] : vector<1x2x4xf32>
return %r : f32
@ -524,7 +524,7 @@ func @fold_extract_broadcast(%a : f32) -> f32 {
// CHECK-LABEL: fold_extract_broadcast_negative
// CHECK: vector.broadcast %{{.*}} : vector<1x1xf32> to vector<1x1x4xf32>
// CHECK: vector.extract %{{.*}}[0, 0] : vector<1x1x4xf32>
func @fold_extract_broadcast_negative(%a : vector<1x1xf32>) -> vector<4xf32> {
func.func @fold_extract_broadcast_negative(%a : vector<1x1xf32>) -> vector<4xf32> {
%b = vector.broadcast %a : vector<1x1xf32> to vector<1x1x4xf32>
%r = vector.extract %b[0, 0] : vector<1x1x4xf32>
return %r : vector<4xf32>
@ -535,7 +535,7 @@ func @fold_extract_broadcast_negative(%a : vector<1x1xf32>) -> vector<4xf32> {
// CHECK-LABEL: fold_extract_splat
// CHECK-SAME: %[[A:.*]]: f32
// CHECK: return %[[A]] : f32
func @fold_extract_splat(%a : f32) -> f32 {
func.func @fold_extract_splat(%a : f32) -> f32 {
%b = vector.splat %a : vector<1x2x4xf32>
%r = vector.extract %b[0, 1, 2] : vector<1x2x4xf32>
return %r : f32
@ -546,7 +546,7 @@ func @fold_extract_splat(%a : f32) -> f32 {
// CHECK-LABEL: fold_extract_broadcast_vector
// CHECK-SAME: %[[A:.*]]: vector<4xf32>
// CHECK: return %[[A]] : vector<4xf32>
func @fold_extract_broadcast_vector(%a : vector<4xf32>) -> vector<4xf32> {
func.func @fold_extract_broadcast_vector(%a : vector<4xf32>) -> vector<4xf32> {
%b = vector.broadcast %a : vector<4xf32> to vector<1x2x4xf32>
%r = vector.extract %b[0, 1] : vector<1x2x4xf32>
return %r : vector<4xf32>
@ -558,7 +558,7 @@ func @fold_extract_broadcast_vector(%a : vector<4xf32>) -> vector<4xf32> {
// CHECK-SAME: %[[A:.*]]: vector<4xf32>
// CHECK: %[[R:.*]] = vector.extract %[[A]][2] : vector<4xf32>
// CHECK: return %[[R]] : f32
func @fold_extract_broadcast(%a : vector<4xf32>) -> f32 {
func.func @fold_extract_broadcast(%a : vector<4xf32>) -> f32 {
%b = vector.broadcast %a : vector<4xf32> to vector<1x2x4xf32>
%r = vector.extract %b[0, 1, 2] : vector<1x2x4xf32>
return %r : f32
@ -569,7 +569,7 @@ func @fold_extract_broadcast(%a : vector<4xf32>) -> f32 {
// CHECK-LABEL: fold_extract_broadcast
// CHECK: %[[B:.*]] = vector.broadcast %{{.*}} : f32 to vector<4xf32>
// CHECK: return %[[B]] : vector<4xf32>
func @fold_extract_broadcast(%a : f32) -> vector<4xf32> {
func.func @fold_extract_broadcast(%a : f32) -> vector<4xf32> {
%b = vector.broadcast %a : f32 to vector<1x2x4xf32>
%r = vector.extract %b[0, 1] : vector<1x2x4xf32>
return %r : vector<4xf32>
@ -581,7 +581,7 @@ func @fold_extract_broadcast(%a : f32) -> vector<4xf32> {
// CHECK-SAME: %[[A:.*]]: vector<1xf32>
// CHECK: %[[R:.*]] = vector.broadcast %[[A]] : vector<1xf32> to vector<8xf32>
// CHECK: return %[[R]] : vector<8xf32>
func @fold_extract_broadcast(%a : vector<1xf32>) -> vector<8xf32> {
func.func @fold_extract_broadcast(%a : vector<1xf32>) -> vector<8xf32> {
%b = vector.broadcast %a : vector<1xf32> to vector<1x8xf32>
%r = vector.extract %b[0] : vector<1x8xf32>
return %r : vector<8xf32>
@ -595,7 +595,7 @@ func @fold_extract_broadcast(%a : vector<1xf32>) -> vector<8xf32> {
// CHECK: %[[R1:.*]] = vector.extract %[[A0]][1, 0, 2] : vector<5x1x3x2xf32>
// CHECK: %[[R2:.*]] = vector.extract %[[A1]][7] : vector<8x4x2xf32>
// CHECK: return %[[R0]], %[[R1]], %[[R2]], %[[A1]] : f32, vector<2xf32>, vector<4x2xf32>, vector<8x4x2xf32>
func @fold_extract_shapecast(%arg0 : vector<5x1x3x2xf32>,
func.func @fold_extract_shapecast(%arg0 : vector<5x1x3x2xf32>,
%arg1 : vector<8x4x2xf32>)
-> (f32, vector<2xf32>, vector<4x2xf32>, vector<8x4x2xf32>) {
%0 = vector.shape_cast %arg0 : vector<5x1x3x2xf32> to vector<15x2xf32>
@ -614,7 +614,7 @@ func @fold_extract_shapecast(%arg0 : vector<5x1x3x2xf32>,
// CHECK: %[[V:.*]] = vector.shape_cast %{{.*}} : vector<16xf32> to vector<2x4x2xf32>
// CHECK: %[[R:.*]] = vector.extract %[[V]][1] : vector<2x4x2xf32>
// CHECK: return %[[R]] : vector<4x2xf32>
func @fold_extract_shapecast_negative(%arg0 : vector<16xf32>,
func.func @fold_extract_shapecast_negative(%arg0 : vector<16xf32>,
%arg1 : vector<8x4x2xf32>) -> vector<4x2xf32> {
%0 = vector.shape_cast %arg0 : vector<16xf32> to vector<2x4x2xf32>
%r = vector.extract %0[1] : vector<2x4x2xf32>
@ -627,7 +627,7 @@ func @fold_extract_shapecast_negative(%arg0 : vector<16xf32>,
// CHECK: %[[A:.*]] = vector.shape_cast %{{.*}} : vector<1x1x64xf32> to vector<1x1x8x8xf32>
// CHECK: %[[B:.*]] = vector.shape_cast %{{.*}} : vector<1x1x8x8xf32> to vector<8x8xf32>
// CHECK: return %[[B]] : vector<8x8xf32>
func @dont_fold_expand_collapse(%arg0: vector<1x1x64xf32>) -> vector<8x8xf32> {
func.func @dont_fold_expand_collapse(%arg0: vector<1x1x64xf32>) -> vector<8x8xf32> {
%0 = vector.shape_cast %arg0 : vector<1x1x64xf32> to vector<1x1x8x8xf32>
%1 = vector.shape_cast %0 : vector<1x1x8x8xf32> to vector<8x8xf32>
return %1 : vector<8x8xf32>
@ -636,7 +636,7 @@ func @dont_fold_expand_collapse(%arg0: vector<1x1x64xf32>) -> vector<8x8xf32> {
// -----
// CHECK-LABEL: fold_vector_transfers
func @fold_vector_transfers(%A: memref<?x8xf32>) -> (vector<4x8xf32>, vector<4x9xf32>) {
func.func @fold_vector_transfers(%A: memref<?x8xf32>) -> (vector<4x8xf32>, vector<4x9xf32>) {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
@ -666,7 +666,7 @@ func @fold_vector_transfers(%A: memref<?x8xf32>) -> (vector<4x8xf32>, vector<4x9
// CHECK-SAME: %[[A:.*]]: vector<4x8xf32>
// CHECK-SAME: %[[B:.*]]: vector<2xi32>
// CHECK: return %[[A]], %[[B]] : vector<4x8xf32>, vector<2xi32>
func @bitcast_folding(%I1: vector<4x8xf32>, %I2: vector<2xi32>) -> (vector<4x8xf32>, vector<2xi32>) {
func.func @bitcast_folding(%I1: vector<4x8xf32>, %I2: vector<2xi32>) -> (vector<4x8xf32>, vector<2xi32>) {
%0 = vector.bitcast %I1 : vector<4x8xf32> to vector<4x8xf32>
%1 = vector.bitcast %I2 : vector<2xi32> to vector<4xi16>
%2 = vector.bitcast %1 : vector<4xi16> to vector<2xi32>
@ -679,7 +679,7 @@ func @bitcast_folding(%I1: vector<4x8xf32>, %I2: vector<2xi32>) -> (vector<4x8xf
// bit pattern: 0x00000000
// CHECK-DAG: %[[CST0:.+]] = arith.constant dense<0.000000e+00> : vector<4xf32>
// CHECK: return %[[CST0]], %[[CST1]]
func @bitcast_f16_to_f32() -> (vector<4xf32>, vector<4xf32>) {
func.func @bitcast_f16_to_f32() -> (vector<4xf32>, vector<4xf32>) {
%cst0 = arith.constant dense<0.0> : vector<8xf16> // bit pattern: 0x0000
%cst1 = arith.constant dense<2.0> : vector<8xf16> // bit pattern: 0x4000
%cast0 = vector.bitcast %cst0: vector<8xf16> to vector<4xf32>
@ -693,7 +693,7 @@ func @bitcast_f16_to_f32() -> (vector<4xf32>, vector<4xf32>) {
// CHECK: %[[CST:.*]] = arith.constant dense<42> : vector<4xi32>
// CHECK-NOT: vector.broadcast
// CHECK: return %[[CST]]
func @broadcast_folding1() -> vector<4xi32> {
func.func @broadcast_folding1() -> vector<4xi32> {
%0 = arith.constant 42 : i32
%1 = vector.broadcast %0 : i32 to vector<4xi32>
return %1 : vector<4xi32>
@ -705,7 +705,7 @@ func @broadcast_folding1() -> vector<4xi32> {
// CHECK: %[[CST:.*]] = arith.constant dense<42> : vector<4x16xi32>
// CHECK-NOT: vector.broadcast
// CHECK: return %[[CST]]
func @broadcast_folding2() -> vector<4x16xi32> {
func.func @broadcast_folding2() -> vector<4x16xi32> {
%0 = arith.constant 42 : i32
%1 = vector.broadcast %0 : i32 to vector<16xi32>
%2 = vector.broadcast %1 : vector<16xi32> to vector<4x16xi32>
@ -718,7 +718,7 @@ func @broadcast_folding2() -> vector<4x16xi32> {
// CHECK-SAME: %[[ARG0:.*]]: i32
// CHECK: %[[RESULT:.*]] = vector.broadcast %[[ARG0]] : i32 to vector<4x16xi32>
// CHECK: return %[[RESULT]]
func @fold_consecutive_broadcasts(%a : i32) -> vector<4x16xi32> {
func.func @fold_consecutive_broadcasts(%a : i32) -> vector<4x16xi32> {
%1 = vector.broadcast %a : i32 to vector<16xi32>
%2 = vector.broadcast %1 : vector<16xi32> to vector<4x16xi32>
return %2 : vector<4x16xi32>
@ -730,7 +730,7 @@ func @fold_consecutive_broadcasts(%a : i32) -> vector<4x16xi32> {
// CHECK-DAG: %[[CST1:.*]] = arith.constant dense<1> : vector<3x4x2xi32>
// CHECK-DAG: %[[CST0:.*]] = arith.constant dense<2.000000e+00> : vector<20x2xf32>
// CHECK: return %[[CST0]], %[[CST1]] : vector<20x2xf32>, vector<3x4x2xi32>
func @shape_cast_constant() -> (vector<20x2xf32>, vector<3x4x2xi32>) {
func.func @shape_cast_constant() -> (vector<20x2xf32>, vector<3x4x2xi32>) {
%cst = arith.constant dense<2.000000e+00> : vector<5x4x2xf32>
%cst_1 = arith.constant dense<1> : vector<12x2xi32>
%0 = vector.shape_cast %cst : vector<5x4x2xf32> to vector<20x2xf32>
@ -744,7 +744,7 @@ func @shape_cast_constant() -> (vector<20x2xf32>, vector<3x4x2xi32>) {
// CHECK-DAG: %[[CST1:.*]] = arith.constant dense<1> : vector<2x13x3xi32>
// CHECK-DAG: %[[CST0:.*]] = arith.constant dense<2.000000e+00> : vector<12x2xf32>
// CHECK: return %[[CST0]], %[[CST1]] : vector<12x2xf32>, vector<2x13x3xi32>
func @extract_strided_constant() -> (vector<12x2xf32>, vector<2x13x3xi32>) {
func.func @extract_strided_constant() -> (vector<12x2xf32>, vector<2x13x3xi32>) {
%cst = arith.constant dense<2.000000e+00> : vector<29x7xf32>
%cst_1 = arith.constant dense<1> : vector<4x37x9xi32>
%0 = vector.extract_strided_slice %cst
@ -761,7 +761,7 @@ func @extract_strided_constant() -> (vector<12x2xf32>, vector<2x13x3xi32>) {
// CHECK-LABEL: extract_strided_broadcast
// CHECK: %[[B:.*]] = vector.broadcast %{{.*}} : vector<4xf16> to vector<2x4xf16>
// CHECK-NEXT: return %[[B]] : vector<2x4xf16>
func @extract_strided_broadcast(%arg0: vector<4xf16>) -> vector<2x4xf16> {
func.func @extract_strided_broadcast(%arg0: vector<4xf16>) -> vector<2x4xf16> {
%0 = vector.broadcast %arg0 : vector<4xf16> to vector<16x4xf16>
%1 = vector.extract_strided_slice %0
{offsets = [0, 0], sizes = [2, 4], strides = [1, 1]} :
@ -775,7 +775,7 @@ func @extract_strided_broadcast(%arg0: vector<4xf16>) -> vector<2x4xf16> {
// CHECK: %[[E:.*]] = vector.extract_strided_slice %{{.*}} {offsets = [0], sizes = [2], strides = [1]} : vector<4xf16> to vector<2xf16>
// CHECK-NEXT: %[[B:.*]] = vector.broadcast %[[E]] : vector<2xf16> to vector<2x2xf16>
// CHECK-NEXT: return %[[B]] : vector<2x2xf16>
func @extract_strided_broadcast2(%arg0: vector<4xf16>) -> vector<2x2xf16> {
func.func @extract_strided_broadcast2(%arg0: vector<4xf16>) -> vector<2x2xf16> {
%0 = vector.broadcast %arg0 : vector<4xf16> to vector<16x4xf16>
%1 = vector.extract_strided_slice %0
{offsets = [0, 0], sizes = [2, 2], strides = [1, 1]} :
@ -789,7 +789,7 @@ func @extract_strided_broadcast2(%arg0: vector<4xf16>) -> vector<2x2xf16> {
// CHECK-SAME: (%[[ARG:.+]]: vector<1xf32>)
// CHECK: %[[V:.+]] = vector.broadcast %[[ARG]] : vector<1xf32> to vector<1x4xf32>
// CHECK: return %[[V]]
func @extract_strided_broadcast3(%arg0: vector<1xf32>) -> vector<1x4xf32> {
func.func @extract_strided_broadcast3(%arg0: vector<1xf32>) -> vector<1x4xf32> {
%0 = vector.broadcast %arg0 : vector<1xf32> to vector<1x8xf32>
%1 = vector.extract_strided_slice %0
{offsets = [0, 4], sizes = [1, 4], strides = [1, 1]}
@ -803,7 +803,7 @@ func @extract_strided_broadcast3(%arg0: vector<1xf32>) -> vector<1x4xf32> {
// CHECK-SAME: (%[[ARG:.+]]: f32)
// CHECK: %[[V:.+]] = vector.broadcast %[[ARG]] : f32 to vector<1x4xf32>
// CHECK: return %[[V]]
func @extract_strided_broadcast4(%arg0: f32) -> vector<1x4xf32> {
func.func @extract_strided_broadcast4(%arg0: f32) -> vector<1x4xf32> {
%0 = vector.broadcast %arg0 : f32 to vector<1x8xf32>
%1 = vector.extract_strided_slice %0
{offsets = [0, 4], sizes = [1, 4], strides = [1, 1]}
@ -816,7 +816,7 @@ func @extract_strided_broadcast4(%arg0: f32) -> vector<1x4xf32> {
// CHECK-LABEL: consecutive_shape_cast
// CHECK: %[[C:.*]] = vector.shape_cast %{{.*}} : vector<16xf16> to vector<4x4xf16>
// CHECK-NEXT: return %[[C]] : vector<4x4xf16>
func @consecutive_shape_cast(%arg0: vector<16xf16>) -> vector<4x4xf16> {
func.func @consecutive_shape_cast(%arg0: vector<16xf16>) -> vector<4x4xf16> {
%0 = vector.shape_cast %arg0 : vector<16xf16> to vector<2x8xf16>
%1 = vector.shape_cast %0 : vector<2x8xf16> to vector<4x4xf16>
return %1 : vector<4x4xf16>
@ -828,7 +828,7 @@ func @consecutive_shape_cast(%arg0: vector<16xf16>) -> vector<4x4xf16> {
// CHECK-NOT: vector.transfer_read
// CHECK-NOT: vector.transfer_write
// CHECK: return
func @dead_transfer_op(%arg0 : tensor<4x4xf32>, %arg1 : memref<4x4xf32>,
func.func @dead_transfer_op(%arg0 : tensor<4x4xf32>, %arg1 : memref<4x4xf32>,
%v0 : vector<1x4xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
@ -846,7 +846,7 @@ func @dead_transfer_op(%arg0 : tensor<4x4xf32>, %arg1 : memref<4x4xf32>,
// CHECK-NOT: vector.gather
// CHECK-NOT: vector.expandload
// CHECK: return
func @dead_load(%base: memref<?xf32>, %indices: vector<16xi32>,
func.func @dead_load(%base: memref<?xf32>, %indices: vector<16xi32>,
%mask: vector<16xi1>, %passthru: vector<16xf32>) {
%c0 = arith.constant 0 : index
%0 = vector.maskedload %base[%c0], %mask, %passthru :
@ -877,7 +877,7 @@ func @dead_load(%base: memref<?xf32>, %indices: vector<16xi32>,
// CHECK-SAME: %[[A_I8:[0-9a-zA-Z]+]]: vector<2x3xi8>
// CHECK-SAME: %[[B_I8:[0-9a-zA-Z]+]]: vector<3x4xi8>
// CHECK-SAME: %[[C_I8:[0-9a-zA-Z]+]]: vector<2x4xi8>
func @contractions(%a: vector<2x3xf32>, %b: vector<3x4xf32>, %c: vector<2x4xf32>,
func.func @contractions(%a: vector<2x3xf32>, %b: vector<3x4xf32>, %c: vector<2x4xf32>,
%a_i8: vector<2x3xi8>, %b_i8: vector<3x4xi8>, %c_i8: vector<2x4xi8>)
-> (vector<2x4xf32>, vector<2x4xi8>)
{
@ -908,7 +908,7 @@ func @contractions(%a: vector<2x3xf32>, %b: vector<3x4xf32>, %c: vector<2x4xf32>
// CHECK-LABEL: func @transfer_folding_1
// CHECK-SAME: %[[T0:[0-9a-zA-Z]+]]: tensor<2x3x4xf32>
// CHECK-SAME: %[[T1:[0-9a-zA-Z]+]]: tensor<2x3x4xf32>
func @transfer_folding_1(%t0: tensor<2x3x4xf32>, %t1: tensor<2x3x4xf32>)
func.func @transfer_folding_1(%t0: tensor<2x3x4xf32>, %t1: tensor<2x3x4xf32>)
-> (tensor<2x3x4xf32>, tensor<2x3x4xf32>, tensor<2x3x4xf32>)
{
%c0 = arith.constant 0 : index
@ -940,7 +940,7 @@ func @transfer_folding_1(%t0: tensor<2x3x4xf32>, %t1: tensor<2x3x4xf32>)
// CHECK-NOT: vector.transfer_read
// CHECK-NOT: vector.transfer_write
// CHECK: return %[[ARG]] : tensor<4x4xf32>
func @store_after_load_tensor(%arg0 : tensor<4x4xf32>) -> tensor<4x4xf32> {
func.func @store_after_load_tensor(%arg0 : tensor<4x4xf32>) -> tensor<4x4xf32> {
%c1 = arith.constant 1 : index
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
@ -957,7 +957,7 @@ func @store_after_load_tensor(%arg0 : tensor<4x4xf32>) -> tensor<4x4xf32> {
// CHECK: vector.transfer_read
// CHECK: vector.transfer_write
// CHECK: return
func @store_after_load_tensor_negative(%arg0 : tensor<4x4xf32>) -> tensor<4x4xf32> {
func.func @store_after_load_tensor_negative(%arg0 : tensor<4x4xf32>) -> tensor<4x4xf32> {
%c1 = arith.constant 1 : index
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
@ -975,7 +975,7 @@ func @store_after_load_tensor_negative(%arg0 : tensor<4x4xf32>) -> tensor<4x4xf3
// CHECK-NOT: vector.transfer_write
// CHECK-NOT: vector.transfer_read
// CHECK: return %[[V0]] : vector<1x4xf32>
func @store_to_load_tensor(%arg0 : tensor<4x4xf32>,
func.func @store_to_load_tensor(%arg0 : tensor<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>) -> vector<1x4xf32> {
%c1 = arith.constant 1 : index
%c2 = arith.constant 2 : index
@ -997,7 +997,7 @@ func @store_to_load_tensor(%arg0 : tensor<4x4xf32>,
// CHECK: vector.transfer_write
// CHECK: %[[V:.*]] = vector.transfer_read
// CHECK: return %[[V]] : vector<1x4xf32>
func @store_to_load_negative_tensor(%arg0 : tensor<4x4xf32>,
func.func @store_to_load_negative_tensor(%arg0 : tensor<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) -> vector<1x4xf32> {
%c1 = arith.constant 1 : index
%c2 = arith.constant 2 : index
@ -1023,7 +1023,7 @@ func @store_to_load_negative_tensor(%arg0 : tensor<4x4xf32>,
// CHECK: vector.transfer_write {{.*}}, {{.*}}[%[[C2]], %[[C0]]
// CHECK: %[[VTW:.*]] = vector.transfer_write {{.*}}, {{.*}}[%[[C1]], %[[C0]]
// CHECK: return %[[VTW]] : tensor<4x4xf32>
func @dead_store_tensor(%arg0 : tensor<4x4xf32>,
func.func @dead_store_tensor(%arg0 : tensor<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) -> tensor<4x4xf32> {
%c1 = arith.constant 1 : index
%c2 = arith.constant 2 : index
@ -1048,7 +1048,7 @@ func @dead_store_tensor(%arg0 : tensor<4x4xf32>,
// CHECK: vector.transfer_read
// CHECK: %[[VTW:.*]] = vector.transfer_write {{.*}}, {{.*}}[%[[C1]], %[[C0]]]
// CHECK: return %[[VTW]] : tensor<4x4xf32>
func @dead_store_tensor_negative(%arg0 : tensor<4x4xf32>,
func.func @dead_store_tensor_negative(%arg0 : tensor<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) -> tensor<4x4xf32> {
%c1 = arith.constant 1 : index
%c2 = arith.constant 2 : index
@ -1075,7 +1075,7 @@ func @dead_store_tensor_negative(%arg0 : tensor<4x4xf32>,
// CHECK: %[[add:.*]] = arith.addi %[[s1]], %[[c4]]
// CHECK: %[[r:.*]] = vector.transfer_read %[[t]][%[[c8]], %[[add]]], %{{.*}} {in_bounds = [true, true]} : tensor<?x?xf32>, vector<5x6xf32>
// CHECK: return %[[r]]
func @transfer_read_of_extract_slice(%t : tensor<?x?xf32>, %s1 : index, %s2 : index) -> vector<5x6xf32> {
func.func @transfer_read_of_extract_slice(%t : tensor<?x?xf32>, %s1 : index, %s2 : index) -> vector<5x6xf32> {
%c3 = arith.constant 3 : index
%c4 = arith.constant 4 : index
%cst = arith.constant 0.0 : f32
@ -1094,7 +1094,7 @@ func @transfer_read_of_extract_slice(%t : tensor<?x?xf32>, %s1 : index, %s2 : in
// CHECK: %[[add:.*]] = arith.addi %[[s1]], %[[c3]]
// CHECK: %[[r:.*]] = vector.transfer_read %[[t]][%[[c5]], %[[add]], %[[c10]]], %{{.*}} {in_bounds = [true, true]} : tensor<?x?x?xf32>, vector<5x6xf32>
// CHECK: return %[[r]]
func @transfer_read_of_extract_slice_rank_reducing(%t : tensor<?x?x?xf32>, %s1 : index, %s2 : index) -> vector<5x6xf32> {
func.func @transfer_read_of_extract_slice_rank_reducing(%t : tensor<?x?x?xf32>, %s1 : index, %s2 : index) -> vector<5x6xf32> {
%c3 = arith.constant 3 : index
%c4 = arith.constant 4 : index
%cst = arith.constant 0.0 : f32
@ -1108,7 +1108,7 @@ func @transfer_read_of_extract_slice_rank_reducing(%t : tensor<?x?x?xf32>, %s1 :
// CHECK-LABEL: func @transfer_read_of_extract_slice_illegal_rank_reducing(
// CHECK: extract_slice
// CHECK: vector.transfer_read
func @transfer_read_of_extract_slice_illegal_rank_reducing(%t : tensor<?x?x?xf32>, %s1 : index, %s2 : index) -> vector<5x6xf32> {
func.func @transfer_read_of_extract_slice_illegal_rank_reducing(%t : tensor<?x?x?xf32>, %s1 : index, %s2 : index) -> vector<5x6xf32> {
%c3 = arith.constant 3 : index
%c4 = arith.constant 4 : index
%cst = arith.constant 0.0 : f32
@ -1124,7 +1124,7 @@ func @transfer_read_of_extract_slice_illegal_rank_reducing(%t : tensor<?x?x?xf32
// CHECK: %[[c3:.*]] = arith.constant 3 : index
// CHECK: %[[r:.*]] = vector.transfer_write %[[v]], %[[t1]][%[[c3]], %[[s]]] {in_bounds = [true, true]} : vector<5x6xf32>, tensor<?x12xf32>
// CHECK: return %[[r]]
func @insert_slice_of_transfer_write(%t1 : tensor<?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x12xf32> {
func.func @insert_slice_of_transfer_write(%t1 : tensor<?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x12xf32> {
%c0 = arith.constant 0 : index
%0 = vector.transfer_write %v, %t2[%c0, %c0] {in_bounds = [true, true]} : vector<5x6xf32>, tensor<5x6xf32>
%1 = tensor.insert_slice %0 into %t1[3, %s] [5, 6] [1, 1] : tensor<5x6xf32> into tensor<?x12xf32>
@ -1136,7 +1136,7 @@ func @insert_slice_of_transfer_write(%t1 : tensor<?x12xf32>, %v : vector<5x6xf32
// CHECK-LABEL: func @insert_slice_of_transfer_write_illegal_rank_extending(
// CHECK: vector.transfer_write
// CHECK: insert_slice
func @insert_slice_of_transfer_write_illegal_rank_extending(%t1 : tensor<?x?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x?x12xf32> {
func.func @insert_slice_of_transfer_write_illegal_rank_extending(%t1 : tensor<?x?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x?x12xf32> {
%c0 = arith.constant 0 : index
%0 = vector.transfer_write %v, %t2[%c0, %c0] {in_bounds = [true, true]} : vector<5x6xf32>, tensor<5x6xf32>
%1 = tensor.insert_slice %0 into %t1[4, 3, %s] [5, 1, 6] [1, 1, 1] : tensor<5x6xf32> into tensor<?x?x12xf32>
@ -1151,7 +1151,7 @@ func @insert_slice_of_transfer_write_illegal_rank_extending(%t1 : tensor<?x?x12x
// CHECK-DAG: %[[c4:.*]] = arith.constant 4 : index
// CHECK: %[[r:.*]] = vector.transfer_write %[[v]], %[[t1]][%[[c4]], %[[c3]], %[[s]]] {in_bounds = [true, true]} : vector<5x6xf32>, tensor<?x?x12xf32>
// CHECK: return %[[r]]
func @insert_slice_of_transfer_write_rank_extending(%t1 : tensor<?x?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x?x12xf32> {
func.func @insert_slice_of_transfer_write_rank_extending(%t1 : tensor<?x?x12xf32>, %v : vector<5x6xf32>, %s : index, %t2 : tensor<5x6xf32>) -> tensor<?x?x12xf32> {
%c0 = arith.constant 0 : index
%0 = vector.transfer_write %v, %t2[%c0, %c0] {in_bounds = [true, true]} : vector<5x6xf32>, tensor<5x6xf32>
%1 = tensor.insert_slice %0 into %t1[4, 3, %s] [1, 5, 6] [1, 1, 1] : tensor<5x6xf32> into tensor<?x?x12xf32>
@ -1238,7 +1238,7 @@ func.func @do_not_swap_extract_slice_transfer_write(%arg0 : vector<8xf32>,
// CHECK-LABEL: func @vector_multi_reduction_single_parallel(
// CHECK-SAME: %[[v:.*]]: vector<2xf32>
func @vector_multi_reduction_single_parallel(%arg0: vector<2xf32>) -> vector<2xf32> {
func.func @vector_multi_reduction_single_parallel(%arg0: vector<2xf32>) -> vector<2xf32> {
%0 = vector.multi_reduction <mul>, %arg0 [] : vector<2xf32> to vector<2xf32>
// CHECK: return %[[v]] : vector<2xf32>
@ -1249,7 +1249,7 @@ func @vector_multi_reduction_single_parallel(%arg0: vector<2xf32>) -> vector<2xf
// CHECK-LABEL: func @insert_strided_slice_full_range
// CHECK-SAME: %[[SOURCE:.+]]: vector<16x16xf16>, %{{.+}}: vector<16x16xf16>
func @insert_strided_slice_full_range(%source: vector<16x16xf16>, %dest: vector<16x16xf16>) -> vector<16x16xf16> {
func.func @insert_strided_slice_full_range(%source: vector<16x16xf16>, %dest: vector<16x16xf16>) -> vector<16x16xf16> {
%0 = vector.insert_strided_slice %source, %dest {offsets = [0, 0], strides = [1, 1]} : vector<16x16xf16> into vector<16x16xf16>
// CHECK: return %[[SOURCE]]
return %0: vector<16x16xf16>
@ -1260,7 +1260,7 @@ func @insert_strided_slice_full_range(%source: vector<16x16xf16>, %dest: vector<
// CHECK-LABEL: extract_strided_splat
// CHECK: %[[B:.*]] = vector.splat %{{.*}} : vector<2x4xf16>
// CHECK-NEXT: return %[[B]] : vector<2x4xf16>
func @extract_strided_splat(%arg0: f16) -> vector<2x4xf16> {
func.func @extract_strided_splat(%arg0: f16) -> vector<2x4xf16> {
%0 = vector.splat %arg0 : vector<16x4xf16>
%1 = vector.extract_strided_slice %0
{offsets = [1, 0], sizes = [2, 4], strides = [1, 1]} :
@ -1275,7 +1275,7 @@ func @extract_strided_splat(%arg0: f16) -> vector<2x4xf16> {
// CHECK: %[[V0:.*]] = vector.extract %[[ARG0]][0, 0] : vector<1x1x4xf32>
// CHECK: %[[V1:.*]] = vector.broadcast %[[ARG1]] : vector<4xf32> to vector<1x1x4xf32>
// CHECK: return %[[V0]], %[[V1]] : vector<4xf32>, vector<1x1x4xf32>
func @insert_extract_to_broadcast(%arg0 : vector<1x1x4xf32>,
func.func @insert_extract_to_broadcast(%arg0 : vector<1x1x4xf32>,
%arg1 : vector<4xf32>) -> (vector<4xf32>, vector<1x1x4xf32>) {
%0 = vector.extract %arg0[0, 0] : vector<1x1x4xf32>
%1 = vector.insert %arg1, %arg0 [0, 0] : vector<4xf32> into vector<1x1x4xf32>
@ -1288,7 +1288,7 @@ func @insert_extract_to_broadcast(%arg0 : vector<1x1x4xf32>,
// CHECK-DAG: %[[CST1:.*]] = arith.constant 1 : i32
// CHECK-DAG: %[[CST0:.*]] = arith.constant dense<2.000000e+00> : vector<7xf32>
// CHECK: return %[[CST0]], %[[CST1]] : vector<7xf32>, i32
func @extract_constant() -> (vector<7xf32>, i32) {
func.func @extract_constant() -> (vector<7xf32>, i32) {
%cst = arith.constant dense<2.000000e+00> : vector<29x7xf32>
%cst_1 = arith.constant dense<1> : vector<4x37x9xi32>
%0 = vector.extract %cst[2] : vector<29x7xf32>
@ -1302,7 +1302,7 @@ func @extract_constant() -> (vector<7xf32>, i32) {
// CHECK-SAME: %[[A:.*]]: vector<32x16x4xf16>
// CHECK: %[[V:.*]] = vector.extract %[[A]][9, 7] : vector<32x16x4xf16>
// CHECK: return %[[V]] : vector<4xf16>
func @extract_extract_strided(%arg0: vector<32x16x4xf16>) -> vector<4xf16> {
func.func @extract_extract_strided(%arg0: vector<32x16x4xf16>) -> vector<4xf16> {
%1 = vector.extract_strided_slice %arg0
{offsets = [7, 3], sizes = [10, 8], strides = [1, 1]} :
vector<32x16x4xf16> to vector<10x8x4xf16>
@ -1316,7 +1316,7 @@ func @extract_extract_strided(%arg0: vector<32x16x4xf16>) -> vector<4xf16> {
// CHECK-SAME: %[[A:.*]]: vector<6x4xf32>
// CHECK: %[[V:.*]] = vector.extract %[[A]][0, 2] : vector<6x4xf32>
// CHECK: return %[[V]] : f32
func @extract_insert_strided(%a: vector<6x4xf32>, %b: vector<8x16xf32>)
func.func @extract_insert_strided(%a: vector<6x4xf32>, %b: vector<8x16xf32>)
-> f32 {
%0 = vector.insert_strided_slice %a, %b {offsets = [2, 2], strides = [1, 1]}
: vector<6x4xf32> into vector<8x16xf32>
@ -1330,7 +1330,7 @@ func @extract_insert_strided(%a: vector<6x4xf32>, %b: vector<8x16xf32>)
// CHECK-SAME: %[[A:.*]]: vector<4xf32>
// CHECK: %[[V:.*]] = vector.extract %[[A]][2] : vector<4xf32>
// CHECK: return %[[V]] : f32
func @extract_insert_rank_reduce(%a: vector<4xf32>, %b: vector<8x16xf32>)
func.func @extract_insert_rank_reduce(%a: vector<4xf32>, %b: vector<8x16xf32>)
-> f32 {
%0 = vector.insert_strided_slice %a, %b {offsets = [2, 2], strides = [1]}
: vector<4xf32> into vector<8x16xf32>
@ -1343,7 +1343,7 @@ func @extract_insert_rank_reduce(%a: vector<4xf32>, %b: vector<8x16xf32>)
// CHECK-LABEL: extract_insert_negative
// CHECK: vector.insert_strided_slice
// CHECK: vector.extract
func @extract_insert_negative(%a: vector<2x15xf32>, %b: vector<12x8x16xf32>)
func.func @extract_insert_negative(%a: vector<2x15xf32>, %b: vector<12x8x16xf32>)
-> vector<16xf32> {
%0 = vector.insert_strided_slice %a, %b {offsets = [4, 2, 0], strides = [1, 1]}
: vector<2x15xf32> into vector<12x8x16xf32>
@ -1357,7 +1357,7 @@ func @extract_insert_negative(%a: vector<2x15xf32>, %b: vector<12x8x16xf32>)
// CHECK-SAME: (%[[A:.*]]: vector<2x16xf32>, %[[B:.*]]: vector<12x8x16xf32>, %[[C:.*]]: vector<2x16xf32>)
// CHECK: %[[V:.*]] = vector.extract %[[C]][0] : vector<2x16xf32>
// CHECK: return %[[V]] : vector<16xf32>
func @extract_insert_chain(%a: vector<2x16xf32>, %b: vector<12x8x16xf32>, %c: vector<2x16xf32>)
func.func @extract_insert_chain(%a: vector<2x16xf32>, %b: vector<12x8x16xf32>, %c: vector<2x16xf32>)
-> vector<16xf32> {
%0 = vector.insert_strided_slice %c, %b {offsets = [4, 2, 0], strides = [1, 1]}
: vector<2x16xf32> into vector<12x8x16xf32>
@ -1373,7 +1373,7 @@ func @extract_insert_chain(%a: vector<2x16xf32>, %b: vector<12x8x16xf32>, %c: ve
// CHECK-SAME: %[[A:.*]]: vector<2x4xf32>
// CHECK: %[[V:.*]] = vector.extract %[[A]][1] : vector<2x4xf32>
// CHECK: return %[[V]] : vector<4xf32>
func @extract_extract_strided2(%A: vector<2x4xf32>)
func.func @extract_extract_strided2(%A: vector<2x4xf32>)
-> (vector<4xf32>) {
%0 = vector.extract_strided_slice %A {offsets = [1, 0], sizes = [1, 4], strides = [1, 1]} : vector<2x4xf32> to vector<1x4xf32>
%1 = vector.extract %0[0] : vector<1x4xf32>
@ -1383,7 +1383,7 @@ func @extract_extract_strided2(%A: vector<2x4xf32>)
// -----
// CHECK-LABEL: func @splat_fold
func @splat_fold() -> vector<4xf32> {
func.func @splat_fold() -> vector<4xf32> {
%c = arith.constant 1.0 : f32
%v = vector.splat %c : vector<4xf32>
return %v : vector<4xf32>
@ -1397,7 +1397,7 @@ func @splat_fold() -> vector<4xf32> {
// CHECK-LABEL: func @shuffle_1d
// CHECK: %[[V:.+]] = arith.constant dense<[3, 2, 5, 1]> : vector<4xi32>
// CHECK: return %[[V]]
func @shuffle_1d() -> vector<4xi32> {
func.func @shuffle_1d() -> vector<4xi32> {
%v0 = arith.constant dense<[0, 1, 2]> : vector<3xi32>
%v1 = arith.constant dense<[3, 4, 5]> : vector<3xi32>
%shuffle = vector.shuffle %v0, %v1 [3, 2, 5, 1] : vector<3xi32>, vector<3xi32>
@ -1406,28 +1406,28 @@ func @shuffle_1d() -> vector<4xi32> {
// CHECK-LABEL: func @shuffle_fold1
// CHECK: %arg0 : vector<4xi32>
func @shuffle_fold1(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<4xi32> {
func.func @shuffle_fold1(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<4xi32> {
%shuffle = vector.shuffle %v0, %v1 [0, 1, 2, 3] : vector<4xi32>, vector<2xi32>
return %shuffle : vector<4xi32>
}
// CHECK-LABEL: func @shuffle_fold2
// CHECK: %arg1 : vector<2xi32>
func @shuffle_fold2(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<2xi32> {
func.func @shuffle_fold2(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<2xi32> {
%shuffle = vector.shuffle %v0, %v1 [4, 5] : vector<4xi32>, vector<2xi32>
return %shuffle : vector<2xi32>
}
// CHECK-LABEL: func @shuffle_fold3
// CHECK: return %arg0 : vector<4x5x6xi32>
func @shuffle_fold3(%v0 : vector<4x5x6xi32>, %v1 : vector<2x5x6xi32>) -> vector<4x5x6xi32> {
func.func @shuffle_fold3(%v0 : vector<4x5x6xi32>, %v1 : vector<2x5x6xi32>) -> vector<4x5x6xi32> {
%shuffle = vector.shuffle %v0, %v1 [0, 1, 2, 3] : vector<4x5x6xi32>, vector<2x5x6xi32>
return %shuffle : vector<4x5x6xi32>
}
// CHECK-LABEL: func @shuffle_fold4
// CHECK: return %arg1 : vector<2x5x6xi32>
func @shuffle_fold4(%v0 : vector<4x5x6xi32>, %v1 : vector<2x5x6xi32>) -> vector<2x5x6xi32> {
func.func @shuffle_fold4(%v0 : vector<4x5x6xi32>, %v1 : vector<2x5x6xi32>) -> vector<2x5x6xi32> {
%shuffle = vector.shuffle %v0, %v1 [4, 5] : vector<4x5x6xi32>, vector<2x5x6xi32>
return %shuffle : vector<2x5x6xi32>
}
@ -1435,7 +1435,7 @@ func @shuffle_fold4(%v0 : vector<4x5x6xi32>, %v1 : vector<2x5x6xi32>) -> vector<
// CHECK-LABEL: func @shuffle_nofold1
// CHECK: %[[V:.+]] = vector.shuffle %arg0, %arg1 [0, 1, 2, 3, 4] : vector<4xi32>, vector<2xi32>
// CHECK: return %[[V]]
func @shuffle_nofold1(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<5xi32> {
func.func @shuffle_nofold1(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<5xi32> {
%shuffle = vector.shuffle %v0, %v1 [0, 1, 2, 3, 4] : vector<4xi32>, vector<2xi32>
return %shuffle : vector<5xi32>
}
@ -1443,7 +1443,7 @@ func @shuffle_nofold1(%v0 : vector<4xi32>, %v1 : vector<2xi32>) -> vector<5xi32>
// CHECK-LABEL: func @shuffle_nofold2
// CHECK: %[[V:.+]] = vector.shuffle %arg0, %arg1 [0, 1, 2, 3] : vector<[4]xi32>, vector<[2]xi32>
// CHECK: return %[[V]]
func @shuffle_nofold2(%v0 : vector<[4]xi32>, %v1 : vector<[2]xi32>) -> vector<4xi32> {
func.func @shuffle_nofold2(%v0 : vector<[4]xi32>, %v1 : vector<[2]xi32>) -> vector<4xi32> {
%shuffle = vector.shuffle %v0, %v1 [0, 1, 2, 3] : vector<[4]xi32>, vector<[2]xi32>
return %shuffle : vector<4xi32>
}
@ -1454,7 +1454,7 @@ func @shuffle_nofold2(%v0 : vector<[4]xi32>, %v1 : vector<[2]xi32>) -> vector<4x
// CHECK-SAME: (%[[ARG:.+]]: vector<1xf32>)
// CHECK: %[[V:.+]] = vector.broadcast %[[ARG]] : vector<1xf32> to vector<1x8xf32>
// CHECK: return %[[V]] : vector<1x8xf32>
func @transpose_scalar_broadcast1(%value: vector<1xf32>) -> vector<1x8xf32> {
func.func @transpose_scalar_broadcast1(%value: vector<1xf32>) -> vector<1x8xf32> {
%bcast = vector.broadcast %value : vector<1xf32> to vector<8x1xf32>
%t = vector.transpose %bcast, [1, 0] : vector<8x1xf32> to vector<1x8xf32>
return %t : vector<1x8xf32>
@ -1466,7 +1466,7 @@ func @transpose_scalar_broadcast1(%value: vector<1xf32>) -> vector<1x8xf32> {
// CHECK-SAME: (%[[ARG:.+]]: f32)
// CHECK: %[[V:.+]] = vector.broadcast %[[ARG]] : f32 to vector<1x8xf32>
// CHECK: return %[[V]] : vector<1x8xf32>
func @transpose_scalar_broadcast2(%value: f32) -> vector<1x8xf32> {
func.func @transpose_scalar_broadcast2(%value: f32) -> vector<1x8xf32> {
%bcast = vector.broadcast %value : f32 to vector<8x1xf32>
%t = vector.transpose %bcast, [1, 0] : vector<8x1xf32> to vector<1x8xf32>
return %t : vector<1x8xf32>
@ -1477,7 +1477,7 @@ func @transpose_scalar_broadcast2(%value: f32) -> vector<1x8xf32> {
// CHECK-LABEL: func @transpose_splat_constant
// CHECK: %[[CST:.+]] = arith.constant dense<5.000000e+00> : vector<8x4xf32>
// CHECK: return %[[CST]]
func @transpose_splat_constant() -> vector<8x4xf32> {
func.func @transpose_splat_constant() -> vector<8x4xf32> {
%cst = arith.constant dense<5.0> : vector<4x8xf32>
%0 = vector.transpose %cst, [1, 0] : vector<4x8xf32> to vector<8x4xf32>
return %0 : vector<8x4xf32>
@ -1488,7 +1488,7 @@ func @transpose_splat_constant() -> vector<8x4xf32> {
// CHECK: %[[VAL_1:.*]] = vector.splat %[[VAL_0]] : vector<3x4xf32>
// CHECK: return %[[VAL_1]] : vector<3x4xf32>
// CHECK: }
func @transpose_splat2(%arg : f32) -> vector<3x4xf32> {
func.func @transpose_splat2(%arg : f32) -> vector<3x4xf32> {
%splat = vector.splat %arg : vector<4x3xf32>
%0 = vector.transpose %splat, [1, 0] : vector<4x3xf32> to vector<3x4xf32>
return %0 : vector<3x4xf32>
@ -1499,7 +1499,7 @@ func @transpose_splat2(%arg : f32) -> vector<3x4xf32> {
// CHECK-LABEL: func @insert_element_fold
// CHECK: %[[V:.+]] = arith.constant dense<[0, 1, 7, 3]> : vector<4xi32>
// CHECK: return %[[V]]
func @insert_element_fold() -> vector<4xi32> {
func.func @insert_element_fold() -> vector<4xi32> {
%v = arith.constant dense<[0, 1, 2, 3]> : vector<4xi32>
%s = arith.constant 7 : i32
%i = arith.constant 2 : i32
@ -1512,7 +1512,7 @@ func @insert_element_fold() -> vector<4xi32> {
// CHECK-LABEL: func @extract_element_fold
// CHECK: %[[C:.+]] = arith.constant 5 : i32
// CHECK: return %[[C]]
func @extract_element_fold() -> i32 {
func.func @extract_element_fold() -> i32 {
%v = arith.constant dense<[1, 3, 5, 7]> : vector<4xi32>
%i = arith.constant 2 : i32
%1 = vector.extractelement %v[%i : i32] : vector<4xi32>
@ -1522,7 +1522,7 @@ func @extract_element_fold() -> i32 {
// CHECK-LABEL: func @extract_element_splat_fold
// CHECK-SAME: (%[[ARG:.+]]: i32)
// CHECK: return %[[ARG]]
func @extract_element_splat_fold(%a : i32) -> i32 {
func.func @extract_element_splat_fold(%a : i32) -> i32 {
%v = vector.splat %a : vector<4xi32>
%i = arith.constant 2 : i32
%1 = vector.extractelement %v[%i : i32] : vector<4xi32>

364
mlir/test/Dialect/Vector/invalid.mlir
View File

File diff suppressed because it is too large Load Diff

108
mlir/test/Dialect/Vector/ops.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt %s | mlir-opt | FileCheck %s
// CHECK-LABEL: func @vector_transfer_ops_0d(
func @vector_transfer_ops_0d(%arg0: tensor<f32>, %arg1: memref<f32>)
func.func @vector_transfer_ops_0d(%arg0: tensor<f32>, %arg1: memref<f32>)
-> tensor<f32> {
%f0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[], %f0 {permutation_map = affine_map<()->()>} :
@ -16,7 +16,7 @@ func @vector_transfer_ops_0d(%arg0: tensor<f32>, %arg1: memref<f32>)
}
// CHECK-LABEL: func @vector_transfer_ops_0d_from_higher_d(
func @vector_transfer_ops_0d_from_higher_d(%arg0: tensor<?xf32>, %arg1: memref<?x?xf32>)
func.func @vector_transfer_ops_0d_from_higher_d(%arg0: tensor<?xf32>, %arg1: memref<?x?xf32>)
-> tensor<?xf32> {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
@ -32,7 +32,7 @@ func @vector_transfer_ops_0d_from_higher_d(%arg0: tensor<?xf32>, %arg1: memref<?
}
// CHECK-LABEL: func @vector_transfer_ops(
func @vector_transfer_ops(%arg0: memref<?x?xf32>,
func.func @vector_transfer_ops(%arg0: memref<?x?xf32>,
%arg1 : memref<?x?xvector<4x3xf32>>,
%arg2 : memref<?x?xvector<4x3xi32>>,
%arg3 : memref<?x?xvector<4x3xindex>>,
@ -92,7 +92,7 @@ func @vector_transfer_ops(%arg0: memref<?x?xf32>,
// CHECK-LABEL: func @vector_transfer_ops_tensor(
func @vector_transfer_ops_tensor(%arg0: tensor<?x?xf32>,
func.func @vector_transfer_ops_tensor(%arg0: tensor<?x?xf32>,
%arg1 : tensor<?x?xvector<4x3xf32>>,
%arg2 : tensor<?x?xvector<4x3xi32>>,
%arg3 : tensor<?x?xvector<4x3xindex>>) ->
@ -149,7 +149,7 @@ func @vector_transfer_ops_tensor(%arg0: tensor<?x?xf32>,
}
// CHECK-LABEL: @vector_broadcast
func @vector_broadcast(%a: f32, %b: vector<f32>, %c: vector<16xf32>, %d: vector<1x16xf32>, %e: vector<8x1xf32>) -> vector<8x16xf32> {
func.func @vector_broadcast(%a: f32, %b: vector<f32>, %c: vector<16xf32>, %d: vector<1x16xf32>, %e: vector<8x1xf32>) -> vector<8x16xf32> {
// CHECK: vector.broadcast %{{.*}} : f32 to vector<f32>
%0 = vector.broadcast %a : f32 to vector<f32>
// CHECK: vector.broadcast %{{.*}} : vector<f32> to vector<4xf32>
@ -166,7 +166,7 @@ func @vector_broadcast(%a: f32, %b: vector<f32>, %c: vector<16xf32>, %d: vector<
}
// CHECK-LABEL: @shuffle1D
func @shuffle1D(%a: vector<2xf32>, %b: vector<4xf32>) -> vector<2xf32> {
func.func @shuffle1D(%a: vector<2xf32>, %b: vector<4xf32>) -> vector<2xf32> {
// CHECK: vector.shuffle %{{.*}}, %{{.*}}[0, 1, 2, 3] : vector<2xf32>, vector<2xf32>
%1 = vector.shuffle %a, %a[0, 1, 2, 3] : vector<2xf32>, vector<2xf32>
// CHECK-NEXT: vector.shuffle %{{.*}}, %{{.*}}[0, 1, 2] : vector<4xf32>, vector<4xf32>
@ -177,21 +177,21 @@ func @shuffle1D(%a: vector<2xf32>, %b: vector<4xf32>) -> vector<2xf32> {
}
// CHECK-LABEL: @shuffle2D
func @shuffle2D(%a: vector<1x4xf32>, %b: vector<2x4xf32>) -> vector<3x4xf32> {
func.func @shuffle2D(%a: vector<1x4xf32>, %b: vector<2x4xf32>) -> vector<3x4xf32> {
// CHECK: vector.shuffle %{{.*}}, %{{.*}}[0, 1, 2] : vector<1x4xf32>, vector<2x4xf32>
%1 = vector.shuffle %a, %b[0, 1, 2] : vector<1x4xf32>, vector<2x4xf32>
return %1 : vector<3x4xf32>
}
// CHECK-LABEL: @extract_element_0d
func @extract_element_0d(%a: vector<f32>) -> f32 {
func.func @extract_element_0d(%a: vector<f32>) -> f32 {
// CHECK-NEXT: vector.extractelement %{{.*}}[] : vector<f32>
%1 = vector.extractelement %a[] : vector<f32>
return %1 : f32
}
// CHECK-LABEL: @extract_element
func @extract_element(%a: vector<16xf32>) -> f32 {
func.func @extract_element(%a: vector<16xf32>) -> f32 {
// CHECK: %[[C15:.*]] = arith.constant 15 : i32
%c = arith.constant 15 : i32
// CHECK-NEXT: vector.extractelement %{{.*}}[%[[C15]] : i32] : vector<16xf32>
@ -200,7 +200,7 @@ func @extract_element(%a: vector<16xf32>) -> f32 {
}
// CHECK-LABEL: @extract
func @extract(%arg0: vector<4x8x16xf32>) -> (vector<4x8x16xf32>, vector<8x16xf32>, vector<16xf32>, f32) {
func.func @extract(%arg0: vector<4x8x16xf32>) -> (vector<4x8x16xf32>, vector<8x16xf32>, vector<16xf32>, f32) {
// CHECK: vector.extract {{.*}}[] : vector<4x8x16xf32>
%0 = vector.extract %arg0[] : vector<4x8x16xf32>
// CHECK: vector.extract {{.*}}[3] : vector<4x8x16xf32>
@ -213,14 +213,14 @@ func @extract(%arg0: vector<4x8x16xf32>) -> (vector<4x8x16xf32>, vector<8x16xf32
}
// CHECK-LABEL: @insert_element_0d
func @insert_element_0d(%a: f32, %b: vector<f32>) -> vector<f32> {
func.func @insert_element_0d(%a: f32, %b: vector<f32>) -> vector<f32> {
// CHECK-NEXT: vector.insertelement %{{.*}}, %{{.*}}[] : vector<f32>
%1 = vector.insertelement %a, %b[] : vector<f32>
return %1 : vector<f32>
}
// CHECK-LABEL: @insert_element
func @insert_element(%a: f32, %b: vector<16xf32>) -> vector<16xf32> {
func.func @insert_element(%a: f32, %b: vector<16xf32>) -> vector<16xf32> {
// CHECK: %[[C15:.*]] = arith.constant 15 : i32
%c = arith.constant 15 : i32
// CHECK-NEXT: vector.insertelement %{{.*}}, %{{.*}}[%[[C15]] : i32] : vector<16xf32>
@ -229,7 +229,7 @@ func @insert_element(%a: f32, %b: vector<16xf32>) -> vector<16xf32> {
}
// CHECK-LABEL: @insert
func @insert(%a: f32, %b: vector<16xf32>, %c: vector<8x16xf32>, %res: vector<4x8x16xf32>) -> vector<4x8x16xf32> {
func.func @insert(%a: f32, %b: vector<16xf32>, %c: vector<8x16xf32>, %res: vector<4x8x16xf32>) -> vector<4x8x16xf32> {
// CHECK: vector.insert %{{.*}}, %{{.*}}[3] : vector<8x16xf32> into vector<4x8x16xf32>
%1 = vector.insert %c, %res[3] : vector<8x16xf32> into vector<4x8x16xf32>
// CHECK: vector.insert %{{.*}}, %{{.*}}[3, 3] : vector<16xf32> into vector<4x8x16xf32>
@ -242,7 +242,7 @@ func @insert(%a: f32, %b: vector<16xf32>, %c: vector<8x16xf32>, %res: vector<4x8
}
// CHECK-LABEL: @outerproduct
func @outerproduct(%arg0: vector<4xf32>, %arg1: vector<8xf32>, %arg2: vector<4x8xf32>) -> vector<4x8xf32> {
func.func @outerproduct(%arg0: vector<4xf32>, %arg1: vector<8xf32>, %arg2: vector<4x8xf32>) -> vector<4x8xf32> {
// CHECK: vector.outerproduct {{.*}} : vector<4xf32>, vector<8xf32>
%0 = vector.outerproduct %arg0, %arg1 : vector<4xf32>, vector<8xf32>
// CHECK: vector.outerproduct {{.*}}, {{.*}}, {{.*}} : vector<4xf32>, vector<8xf32>
@ -251,14 +251,14 @@ func @outerproduct(%arg0: vector<4xf32>, %arg1: vector<8xf32>, %arg2: vector<4x8
}
// CHECK-LABEL: @insert_strided_slice
func @insert_strided_slice(%a: vector<4x4xf32>, %b: vector<4x8x16xf32>) {
func.func @insert_strided_slice(%a: vector<4x4xf32>, %b: vector<4x8x16xf32>) {
// CHECK: vector.insert_strided_slice %{{.*}}, %{{.*}} {offsets = [2, 2, 2], strides = [1, 1]} : vector<4x4xf32> into vector<4x8x16xf32>
%1 = vector.insert_strided_slice %a, %b {offsets = [2, 2, 2], strides = [1, 1]} : vector<4x4xf32> into vector<4x8x16xf32>
return
}
// CHECK-LABEL: @extract_strided_slice
func @extract_strided_slice(%arg0: vector<4x8x16xf32>) -> vector<2x2x16xf32> {
func.func @extract_strided_slice(%arg0: vector<4x8x16xf32>) -> vector<2x2x16xf32> {
// CHECK: vector.extract_strided_slice %{{.*}} {offsets = [2, 2], sizes = [2, 2], strides = [1, 1]} : vector<4x8x16xf32>
%1 = vector.extract_strided_slice %arg0 {offsets = [2, 2], sizes = [2, 2], strides = [1, 1]} : vector<4x8x16xf32> to vector<2x2x16xf32>
return %1: vector<2x2x16xf32>
@ -274,7 +274,7 @@ func @extract_strided_slice(%arg0: vector<4x8x16xf32>) -> vector<2x2x16xf32> {
iterator_types = ["reduction"]
}
// CHECK-LABEL: @contraction_to_scalar
func @contraction_to_scalar(%arg0: vector<10xf32>, %arg1: vector<10xf32>) -> f32 {
func.func @contraction_to_scalar(%arg0: vector<10xf32>, %arg1: vector<10xf32>) -> f32 {
// CHECK: %[[C0:.*]] = arith.constant 0.000000e+00 : f32
%f0 = arith.constant 0.0: f32
// CHECK: %[[X:.*]] = vector.contract {indexing_maps = [#{{.*}}, #{{.*}}, #{{.*}}], iterator_types = ["reduction"], kind = #vector.kind<add>} %{{.*}}, %{{.*}}, %[[C0]] : vector<10xf32>, vector<10xf32> into f32
@ -295,7 +295,7 @@ func @contraction_to_scalar(%arg0: vector<10xf32>, %arg1: vector<10xf32>) -> f32
kind = #vector.kind<maxf>
}
// CHECK-LABEL: @contraction_to_scalar_with_max
func @contraction_to_scalar_with_max(%arg0: vector<10xf32>, %arg1: vector<10xf32>) -> f32 {
func.func @contraction_to_scalar_with_max(%arg0: vector<10xf32>, %arg1: vector<10xf32>) -> f32 {
// CHECK: %[[C0:.*]] = arith.constant 0.000000e+00 : f32
%f0 = arith.constant 0.0: f32
// CHECK: %[[X:.*]] = vector.contract {indexing_maps = [#{{.*}}, #{{.*}}, #{{.*}}], iterator_types = ["reduction"], kind = #vector.kind<maxf>} %{{.*}}, %{{.*}}, %[[C0]] : vector<10xf32>, vector<10xf32> into f32
@ -333,7 +333,7 @@ func @contraction_to_scalar_with_max(%arg0: vector<10xf32>, %arg1: vector<10xf32
kind = #vector.kind<maxf>
}
// CHECK-LABEL: @contraction
func @contraction(%arg0 : vector<7x8x16x15xf32>, %arg1 : vector<8x16x7x5xf32>,
func.func @contraction(%arg0 : vector<7x8x16x15xf32>, %arg1 : vector<8x16x7x5xf32>,
%arg2 : vector<8x15x5xf32>, %arg3 : vector<8x8x15x5xf32>,
%arg4 : vector<7x8x16x15xf16>, %arg5 : vector<8x16x7x5xf16>) {
// Test contraction with batch and contracting dims.
@ -365,7 +365,7 @@ func @contraction(%arg0 : vector<7x8x16x15xf32>, %arg1 : vector<8x16x7x5xf32>,
}
// CHECK-LABEL: @create_vector_mask
func @create_vector_mask() {
func.func @create_vector_mask() {
// CHECK: %[[C2:.*]] = arith.constant 2 : index
%c2 = arith.constant 2 : index
// CHECK-NEXT: %[[C3:.*]] = arith.constant 3 : index
@ -377,7 +377,7 @@ func @create_vector_mask() {
}
// CHECK-LABEL: @constant_vector_mask_0d
func @constant_vector_mask_0d() {
func.func @constant_vector_mask_0d() {
// CHECK: vector.constant_mask [0] : vector<i1>
%0 = vector.constant_mask [0] : vector<i1>
// CHECK: vector.constant_mask [1] : vector<i1>
@ -386,7 +386,7 @@ func @constant_vector_mask_0d() {
}
// CHECK-LABEL: @constant_vector_mask
func @constant_vector_mask() {
func.func @constant_vector_mask() {
// CHECK: vector.constant_mask [3, 2] : vector<4x3xi1>
%0 = vector.constant_mask [3, 2] : vector<4x3xi1>
// CHECK: vector.constant_mask [0] : vector<[4]xi1>
@ -395,14 +395,14 @@ func @constant_vector_mask() {
}
// CHECK-LABEL: @vector_print
func @vector_print(%arg0: vector<8x4xf32>) {
func.func @vector_print(%arg0: vector<8x4xf32>) {
// CHECK: vector.print %{{.*}} : vector<8x4xf32>
vector.print %arg0 : vector<8x4xf32>
return
}
// CHECK-LABEL: @reshape
func @reshape(%arg0 : vector<3x2x4xf32>) -> (vector<2x3x4xf32>) {
func.func @reshape(%arg0 : vector<3x2x4xf32>) -> (vector<2x3x4xf32>) {
// CHECK: %[[C2:.*]] = arith.constant 2 : index
%c2 = arith.constant 2 : index
// CHECK: %[[C3:.*]] = arith.constant 3 : index
@ -419,7 +419,7 @@ func @reshape(%arg0 : vector<3x2x4xf32>) -> (vector<2x3x4xf32>) {
}
// CHECK-LABEL: @shape_cast
func @shape_cast(%arg0 : vector<5x1x3x2xf32>,
func.func @shape_cast(%arg0 : vector<5x1x3x2xf32>,
%arg1 : vector<8x1xf32>,
%arg2 : vector<16x1x1xf32>)
-> (vector<15x2xf32>, vector<8xf32>, vector<16xf32>, vector<16x1xf32>) {
@ -440,7 +440,7 @@ func @shape_cast(%arg0 : vector<5x1x3x2xf32>,
}
// CHECK-LABEL: @bitcast
func @bitcast(%arg0 : vector<5x1x3x2xf32>,
func.func @bitcast(%arg0 : vector<5x1x3x2xf32>,
%arg1 : vector<8x1xi32>,
%arg2 : vector<16x1x8xi8>,
%arg3 : vector<8x2x1xindex>,
@ -478,7 +478,7 @@ func @bitcast(%arg0 : vector<5x1x3x2xf32>,
}
// CHECK-LABEL: @vector_fma
func @vector_fma(%a: vector<8xf32>, %b: vector<8x4xf32>) {
func.func @vector_fma(%a: vector<8xf32>, %b: vector<8x4xf32>) {
// CHECK: vector.fma %{{.*}} : vector<8xf32>
vector.fma %a, %a, %a : vector<8xf32>
// CHECK: vector.fma %{{.*}} : vector<8x4xf32>
@ -487,7 +487,7 @@ func @vector_fma(%a: vector<8xf32>, %b: vector<8x4xf32>) {
}
// CHECK-LABEL: @reduce_fp
func @reduce_fp(%arg0: vector<16xf32>, %arg1: f32) -> f32 {
func.func @reduce_fp(%arg0: vector<16xf32>, %arg1: f32) -> f32 {
// CHECK: vector.reduction <add>, %{{.*}} : vector<16xf32> into f32
vector.reduction <add>, %arg0 : vector<16xf32> into f32
// CHECK: vector.reduction <add>, %{{.*}}, %{{.*}} : vector<16xf32> into f32
@ -505,7 +505,7 @@ func @reduce_fp(%arg0: vector<16xf32>, %arg1: f32) -> f32 {
}
// CHECK-LABEL: @reduce_int
func @reduce_int(%arg0: vector<16xi32>) -> i32 {
func.func @reduce_int(%arg0: vector<16xi32>) -> i32 {
// CHECK: vector.reduction <add>, %{{.*}} : vector<16xi32> into i32
vector.reduction <add>, %arg0 : vector<16xi32> into i32
// CHECK: vector.reduction <mul>, %{{.*}} : vector<16xi32> into i32
@ -529,7 +529,7 @@ func @reduce_int(%arg0: vector<16xi32>) -> i32 {
}
// CHECK-LABEL: @transpose_fp
func @transpose_fp(%arg0: vector<3x7xf32>) -> vector<7x3xf32> {
func.func @transpose_fp(%arg0: vector<3x7xf32>) -> vector<7x3xf32> {
// CHECK: %[[X:.*]] = vector.transpose %{{.*}}, [1, 0] : vector<3x7xf32> to vector<7x3xf32>
%0 = vector.transpose %arg0, [1, 0] : vector<3x7xf32> to vector<7x3xf32>
// CHECK: return %[[X]] : vector<7x3xf32>
@ -537,7 +537,7 @@ func @transpose_fp(%arg0: vector<3x7xf32>) -> vector<7x3xf32> {
}
// CHECK-LABEL: @transpose_int
func @transpose_int(%arg0: vector<11x7x3x2xi32>) -> vector<2x11x7x3xi32> {
func.func @transpose_int(%arg0: vector<11x7x3x2xi32>) -> vector<2x11x7x3xi32> {
// CHECK: %[[X:.*]] = vector.transpose %{{.*}}, [3, 0, 1, 2] : vector<11x7x3x2xi32> to vector<2x11x7x3xi32>
%0 = vector.transpose %arg0, [3, 0, 1, 2] : vector<11x7x3x2xi32> to vector<2x11x7x3xi32>
// CHECK: return %[[X]] : vector<2x11x7x3xi32>
@ -545,7 +545,7 @@ func @transpose_int(%arg0: vector<11x7x3x2xi32>) -> vector<2x11x7x3xi32> {
}
// CHECK-LABEL: @flat_transpose_fp
func @flat_transpose_fp(%arg0: vector<16xf32>) -> vector<16xf32> {
func.func @flat_transpose_fp(%arg0: vector<16xf32>) -> vector<16xf32> {
// CHECK: %[[X:.*]] = vector.flat_transpose %{{.*}} {columns = 4 : i32, rows = 4 : i32} : vector<16xf32> -> vector<16xf32>
%0 = vector.flat_transpose %arg0 { rows = 4: i32, columns = 4: i32 } : vector<16xf32> -> vector<16xf32>
// CHECK: return %[[X]] : vector<16xf32>
@ -553,7 +553,7 @@ func @flat_transpose_fp(%arg0: vector<16xf32>) -> vector<16xf32> {
}
// CHECK-LABEL: @flat_transpose_int
func @flat_transpose_int(%arg0: vector<16xi32>) -> vector<16xi32> {
func.func @flat_transpose_int(%arg0: vector<16xi32>) -> vector<16xi32> {
// CHECK: %[[X:.*]] = vector.flat_transpose %{{.*}} {columns = 8 : i32, rows = 2 : i32} : vector<16xi32> -> vector<16xi32>
%0 = vector.flat_transpose %arg0 { rows = 2: i32, columns = 8: i32 } : vector<16xi32> -> vector<16xi32>
// CHECK: return %[[X]] : vector<16xi32>
@ -561,7 +561,7 @@ func @flat_transpose_int(%arg0: vector<16xi32>) -> vector<16xi32> {
}
// CHECK-LABEL: @vector_load_and_store_1d_scalar_memref
func @vector_load_and_store_1d_scalar_memref(%memref : memref<200x100xf32>,
func.func @vector_load_and_store_1d_scalar_memref(%memref : memref<200x100xf32>,
%i : index, %j : index) {
// CHECK: %[[ld:.*]] = vector.load %{{.*}}[%{{.*}}] : memref<200x100xf32>, vector<8xf32>
%0 = vector.load %memref[%i, %j] : memref<200x100xf32>, vector<8xf32>
@ -571,7 +571,7 @@ func @vector_load_and_store_1d_scalar_memref(%memref : memref<200x100xf32>,
}
// CHECK-LABEL: @vector_load_and_store_1d_vector_memref
func @vector_load_and_store_1d_vector_memref(%memref : memref<200x100xvector<8xf32>>,
func.func @vector_load_and_store_1d_vector_memref(%memref : memref<200x100xvector<8xf32>>,
%i : index, %j : index) {
// CHECK: %[[ld:.*]] = vector.load %{{.*}}[%{{.*}}] : memref<200x100xvector<8xf32>>, vector<8xf32>
%0 = vector.load %memref[%i, %j] : memref<200x100xvector<8xf32>>, vector<8xf32>
@ -581,7 +581,7 @@ func @vector_load_and_store_1d_vector_memref(%memref : memref<200x100xvector<8xf
}
// CHECK-LABEL: @vector_load_and_store_scalable_vector_memref
func @vector_load_and_store_scalable_vector_memref(%v: vector<[4]xi32>, %m: memref<?xi32>) -> vector<[4]xi32> {
func.func @vector_load_and_store_scalable_vector_memref(%v: vector<[4]xi32>, %m: memref<?xi32>) -> vector<[4]xi32> {
%c0 = arith.constant 0 : index
// CHECK: vector.load {{.*}}: memref<?xi32>, vector<[4]xi32>
%0 = vector.load %m[%c0] : memref<?xi32>, vector<[4]xi32>
@ -590,7 +590,7 @@ func @vector_load_and_store_scalable_vector_memref(%v: vector<[4]xi32>, %m: memr
return %0 : vector<[4]xi32>
}
func @vector_load_and_store_1d_scalable_vector_memref(%memref : memref<200x100xvector<8xf32>>,
func.func @vector_load_and_store_1d_scalable_vector_memref(%memref : memref<200x100xvector<8xf32>>,
%i : index, %j : index) {
// CHECK: %[[ld:.*]] = vector.load %{{.*}}[%{{.*}}] : memref<200x100xvector<8xf32>>, vector<8xf32>
%0 = vector.load %memref[%i, %j] : memref<200x100xvector<8xf32>>, vector<8xf32>
@ -600,7 +600,7 @@ func @vector_load_and_store_1d_scalable_vector_memref(%memref : memref<200x100xv
}
// CHECK-LABEL: @vector_load_and_store_out_of_bounds
func @vector_load_and_store_out_of_bounds(%memref : memref<7xf32>) {
func.func @vector_load_and_store_out_of_bounds(%memref : memref<7xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[ld:.*]] = vector.load %{{.*}}[%{{.*}}] : memref<7xf32>, vector<8xf32>
%0 = vector.load %memref[%c0] : memref<7xf32>, vector<8xf32>
@ -610,7 +610,7 @@ func @vector_load_and_store_out_of_bounds(%memref : memref<7xf32>) {
}
// CHECK-LABEL: @vector_load_and_store_2d_scalar_memref
func @vector_load_and_store_2d_scalar_memref(%memref : memref<200x100xf32>,
func.func @vector_load_and_store_2d_scalar_memref(%memref : memref<200x100xf32>,
%i : index, %j : index) {
// CHECK: %[[ld:.*]] = vector.load %{{.*}}[%{{.*}}] : memref<200x100xf32>, vector<4x8xf32>
%0 = vector.load %memref[%i, %j] : memref<200x100xf32>, vector<4x8xf32>
@ -620,7 +620,7 @@ func @vector_load_and_store_2d_scalar_memref(%memref : memref<200x100xf32>,
}
// CHECK-LABEL: @vector_load_and_store_2d_vector_memref
func @vector_load_and_store_2d_vector_memref(%memref : memref<200x100xvector<4x8xf32>>,
func.func @vector_load_and_store_2d_vector_memref(%memref : memref<200x100xvector<4x8xf32>>,
%i : index, %j : index) {
// CHECK: %[[ld:.*]] = vector.load %{{.*}}[%{{.*}}] : memref<200x100xvector<4x8xf32>>, vector<4x8xf32>
%0 = vector.load %memref[%i, %j] : memref<200x100xvector<4x8xf32>>, vector<4x8xf32>
@ -630,7 +630,7 @@ func @vector_load_and_store_2d_vector_memref(%memref : memref<200x100xvector<4x8
}
// CHECK-LABEL: @masked_load_and_store
func @masked_load_and_store(%base: memref<?xf32>, %mask: vector<16xi1>, %passthru: vector<16xf32>) {
func.func @masked_load_and_store(%base: memref<?xf32>, %mask: vector<16xi1>, %passthru: vector<16xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[X:.*]] = vector.maskedload %{{.*}}[%{{.*}}], %{{.*}}, %{{.*}} : memref<?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
%0 = vector.maskedload %base[%c0], %mask, %passthru : memref<?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
@ -640,7 +640,7 @@ func @masked_load_and_store(%base: memref<?xf32>, %mask: vector<16xi1>, %passthr
}
// CHECK-LABEL: @masked_load_and_store2d
func @masked_load_and_store2d(%base: memref<?x?xf32>, %mask: vector<16xi1>, %passthru: vector<16xf32>) {
func.func @masked_load_and_store2d(%base: memref<?x?xf32>, %mask: vector<16xi1>, %passthru: vector<16xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[X:.*]] = vector.maskedload %{{.*}}[%{{.*}}, %{{.*}}], %{{.*}}, %{{.*}} : memref<?x?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
%0 = vector.maskedload %base[%c0, %c0], %mask, %passthru : memref<?x?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
@ -650,7 +650,7 @@ func @masked_load_and_store2d(%base: memref<?x?xf32>, %mask: vector<16xi1>, %pas
}
// CHECK-LABEL: @gather_and_scatter
func @gather_and_scatter(%base: memref<?xf32>, %v: vector<16xi32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
func.func @gather_and_scatter(%base: memref<?xf32>, %v: vector<16xi32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[X:.*]] = vector.gather %{{.*}}[%{{.*}}] [%{{.*}}], %{{.*}}, %{{.*}} : memref<?xf32>, vector<16xi32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
%0 = vector.gather %base[%c0][%v], %mask, %pass_thru : memref<?xf32>, vector<16xi32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
@ -660,7 +660,7 @@ func @gather_and_scatter(%base: memref<?xf32>, %v: vector<16xi32>, %mask: vector
}
// CHECK-LABEL: @gather_and_scatter2d
func @gather_and_scatter2d(%base: memref<?x?xf32>, %v: vector<16xi32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
func.func @gather_and_scatter2d(%base: memref<?x?xf32>, %v: vector<16xi32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[X:.*]] = vector.gather %{{.*}}[%{{.*}}, %{{.*}}] [%{{.*}}], %{{.*}}, %{{.*}} : memref<?x?xf32>, vector<16xi32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
%0 = vector.gather %base[%c0, %c0][%v], %mask, %pass_thru : memref<?x?xf32>, vector<16xi32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
@ -670,7 +670,7 @@ func @gather_and_scatter2d(%base: memref<?x?xf32>, %v: vector<16xi32>, %mask: ve
}
// CHECK-LABEL: @expand_and_compress
func @expand_and_compress(%base: memref<?xf32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
func.func @expand_and_compress(%base: memref<?xf32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[X:.*]] = vector.expandload %{{.*}}[%{{.*}}], %{{.*}}, %{{.*}} : memref<?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
%0 = vector.expandload %base[%c0], %mask, %pass_thru : memref<?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
@ -680,7 +680,7 @@ func @expand_and_compress(%base: memref<?xf32>, %mask: vector<16xi1>, %pass_thru
}
// CHECK-LABEL: @expand_and_compress2d
func @expand_and_compress2d(%base: memref<?x?xf32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
func.func @expand_and_compress2d(%base: memref<?x?xf32>, %mask: vector<16xi1>, %pass_thru: vector<16xf32>) {
%c0 = arith.constant 0 : index
// CHECK: %[[X:.*]] = vector.expandload %{{.*}}[%{{.*}}, %{{.*}}], %{{.*}}, %{{.*}} : memref<?x?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
%0 = vector.expandload %base[%c0, %c0], %mask, %pass_thru : memref<?x?xf32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
@ -690,7 +690,7 @@ func @expand_and_compress2d(%base: memref<?x?xf32>, %mask: vector<16xi1>, %pass_
}
// CHECK-LABEL: @extract_insert_map
func @extract_insert_map(%v: vector<32xf32>, %v2: vector<16x32xf32>,
func.func @extract_insert_map(%v: vector<32xf32>, %v2: vector<16x32xf32>,
%id0 : index, %id1 : index) -> (vector<32xf32>, vector<16x32xf32>) {
// CHECK: %[[V:.*]] = vector.extract_map %{{.*}}[%{{.*}}] : vector<32xf32> to vector<2xf32>
%vd = vector.extract_map %v[%id0] : vector<32xf32> to vector<2xf32>
@ -705,7 +705,7 @@ func @extract_insert_map(%v: vector<32xf32>, %v2: vector<16x32xf32>,
}
// CHECK-LABEL: @multi_reduction
func @multi_reduction(%0: vector<4x8x16x32xf32>) -> f32 {
func.func @multi_reduction(%0: vector<4x8x16x32xf32>) -> f32 {
%1 = vector.multi_reduction <add>, %0 [1, 3] :
vector<4x8x16x32xf32> to vector<4x16xf32>
%2 = vector.multi_reduction <add>, %1 [0, 1] :
@ -714,14 +714,14 @@ func @multi_reduction(%0: vector<4x8x16x32xf32>) -> f32 {
}
// CHECK-LABEL: @get_vector_scale
func @get_vector_scale() -> index {
func.func @get_vector_scale() -> index {
// CHECK: vector.vscale
%0 = vector.vscale
return %0 : index
}
// CHECK-LABEL: @vector_scan
func @vector_scan(%0: vector<4x8x16x32xf32>) -> vector<4x8x16x32xf32> {
func.func @vector_scan(%0: vector<4x8x16x32xf32>) -> vector<4x8x16x32xf32> {
%1 = arith.constant dense<0.0> : vector<4x16x32xf32>
%2:2 = vector.scan <add>, %0, %1 {reduction_dim = 1 : i64, inclusive = true} :
vector<4x8x16x32xf32>, vector<4x16x32xf32>
@ -730,7 +730,7 @@ func @vector_scan(%0: vector<4x8x16x32xf32>) -> vector<4x8x16x32xf32> {
// CHECK-LABEL: func @test_splat_op
// CHECK-SAME: [[S:%arg[0-9]+]]: f32
func @test_splat_op(%s : f32) {
func.func @test_splat_op(%s : f32) {
// CHECK: vector.splat [[S]] : vector<8xf32>
%v = vector.splat %s : vector<8xf32>
@ -740,14 +740,14 @@ func @test_splat_op(%s : f32) {
}
// CHECK-LABEL: func @vector_splat_0d(
func @vector_splat_0d(%a: f32) -> vector<f32> {
func.func @vector_splat_0d(%a: f32) -> vector<f32> {
// CHECK: vector.splat %{{.*}} : vector<f32>
%0 = vector.splat %a : vector<f32>
return %0 : vector<f32>
}
// CHECK-LABEL: func @warp_execute_on_lane_0(
func @warp_execute_on_lane_0(%laneid: index) {
func.func @warp_execute_on_lane_0(%laneid: index) {
// CHECK-NEXT: vector.warp_execute_on_lane_0(%{{.*}})[32] {
vector.warp_execute_on_lane_0(%laneid)[32] {
// CHECK-NEXT: }
@ -757,7 +757,7 @@ func @warp_execute_on_lane_0(%laneid: index) {
}
// CHECK-LABEL: func @warp_operand_result(
func @warp_operand_result(%laneid: index, %v0 : vector<4xi32>) -> (vector<4xi32>) {
func.func @warp_operand_result(%laneid: index, %v0 : vector<4xi32>) -> (vector<4xi32>) {
// CHECK-NEXT: %{{.*}} = vector.warp_execute_on_lane_0(%{{.*}})[32] args(%{{.*}} : vector<4xi32>) -> (vector<4xi32>) {
%2 = vector.warp_execute_on_lane_0(%laneid)[32]
args(%v0 : vector<4xi32>) -> (vector<4xi32>) {

12
mlir/test/Dialect/Vector/vector-contract-matvec-transforms.mlir
View File

@ -71,7 +71,7 @@
// CHECK: %[[T9:.*]] = vector.outerproduct %[[T7]], %[[T8]], %[[T6]] {kind = #vector.kind<add>} : vector<2xf32>, f32
// CHECK: memref.store %[[T9]], %[[C]][] : memref<vector<2xf32>>
// CHECK: return
func @matvec2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
func.func @matvec2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
%arg2: memref<vector<2xf32>>) {
%A = memref.load %arg0[] : memref<vector<2x2xf32>>
%x = memref.load %arg1[] : memref<vector<2xf32>>
@ -97,7 +97,7 @@ func @matvec2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
// CHECK: %[[T9:.*]] = vector.outerproduct %[[T7]], %[[T8]], %[[T6]] {kind = #vector.kind<maxf>} : vector<2xf32>, f32
// CHECK: memref.store %[[T9]], %[[C]][] : memref<vector<2xf32>>
// CHECK: return
func @matvecmax2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
func.func @matvecmax2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
%arg2: memref<vector<2xf32>>) {
%A = memref.load %arg0[] : memref<vector<2x2xf32>>
%x = memref.load %arg1[] : memref<vector<2xf32>>
@ -122,7 +122,7 @@ func @matvecmax2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
// CHECK: %[[T8:.*]] = vector.outerproduct %[[T6]], %[[T7]], %[[T5]] {kind = #vector.kind<add>} : vector<2xf32>, f32
// CHECK: memref.store %[[T8]], %[[C]][] : memref<vector<2xf32>>
// CHECK: return
func @mattransvec2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
func.func @mattransvec2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
%arg2: memref<vector<2xf32>>) {
%A = memref.load %arg0[] : memref<vector<2x2xf32>>
%x = memref.load %arg1[] : memref<vector<2xf32>>
@ -148,7 +148,7 @@ func @mattransvec2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>
// CHECK: %[[T9:.*]] = vector.outerproduct %[[T7]], %[[T8]], %[[T6]] {kind = #vector.kind<add>} : vector<2xf32>, f32
// CHECK: memref.store %[[T9]], %[[C]][] : memref<vector<2xf32>>
// CHECK: return
func @vecmat2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
func.func @vecmat2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
%arg2: memref<vector<2xf32>>) {
%A = memref.load %arg0[] : memref<vector<2x2xf32>>
%x = memref.load %arg1[] : memref<vector<2xf32>>
@ -173,7 +173,7 @@ func @vecmat2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
// CHECK: %[[T8:.*]] = vector.outerproduct %[[T6]], %[[T7]], %[[T5]] {kind = #vector.kind<add>} : vector<2xf32>, f32
// CHECK: memref.store %[[T8]], %[[C]][] : memref<vector<2xf32>>
// CHECK: return
func @vecmattrans2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
func.func @vecmattrans2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
%arg2: memref<vector<2xf32>>) {
%A = memref.load %arg0[] : memref<vector<2x2xf32>>
%x = memref.load %arg1[] : memref<vector<2xf32>>
@ -198,7 +198,7 @@ func @vecmattrans2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>
// CHECK: %[[T8:.*]] = vector.outerproduct %[[T6]], %[[T7]], %[[T5]] {kind = #vector.kind<add>} : vector<2xf32>, f32
// CHECK: memref.store %[[T8]], %[[C]][] : memref<vector<2xf32>>
// CHECK: return
func @redpar_vecmattrans2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
func.func @redpar_vecmattrans2x2(%arg0: memref<vector<2x2xf32>>, %arg1: memref<vector<2xf32>>,
%arg2: memref<vector<2xf32>>) {
%A = memref.load %arg0[] : memref<vector<2x2xf32>>
%x = memref.load %arg1[] : memref<vector<2xf32>>

100
mlir/test/Dialect/Vector/vector-contract-transforms.mlir
View File

@ -22,7 +22,7 @@
// CHECK: %[[ACC:.*]] = arith.addf %[[R]], %[[C]] : f32
// CHECK: return %[[ACC]] : f32
func @extract_contract1(%arg0: vector<4xf32>, %arg1: vector<4xf32>, %arg2: f32) -> f32 {
func.func @extract_contract1(%arg0: vector<4xf32>, %arg1: vector<4xf32>, %arg2: f32) -> f32 {
%0 = vector.contract #dotp_trait %arg0, %arg1, %arg2
: vector<4xf32>, vector<4xf32> into f32
return %0 : f32
@ -37,7 +37,7 @@ func @extract_contract1(%arg0: vector<4xf32>, %arg1: vector<4xf32>, %arg2: f32)
// CHECK: %[[ACC:.*]] = arith.addi %[[R]], %[[C]] : i32
// CHECK: return %[[ACC]] : i32
func @extract_contract1_int(%arg0: vector<4xi32>, %arg1: vector<4xi32>, %arg2: i32) -> i32 {
func.func @extract_contract1_int(%arg0: vector<4xi32>, %arg1: vector<4xi32>, %arg2: i32) -> i32 {
%0 = vector.contract #dotp_trait %arg0, %arg1, %arg2
: vector<4xi32>, vector<4xi32> into i32
return %0 : i32
@ -69,7 +69,7 @@ func @extract_contract1_int(%arg0: vector<4xi32>, %arg1: vector<4xi32>, %arg2: i
// CHECK: %[[T10:.*]] = arith.addf %[[T9]], %[[C]] : vector<2xf32>
// CHECK: return %[[T10]] : vector<2xf32>
func @extract_contract2(%arg0: vector<2x3xf32>,
func.func @extract_contract2(%arg0: vector<2x3xf32>,
%arg1: vector<3xf32>,
%arg2: vector<2xf32>) -> vector<2xf32> {
%0 = vector.contract #matvec_trait %arg0, %arg1, %arg2
@ -92,7 +92,7 @@ func @extract_contract2(%arg0: vector<2x3xf32>,
// CHECK: %[[T9:.*]] = vector.insert %[[T8]], %[[T4]] [1] : i32 into vector<2xi32>
// CHECK: %[[T10:.*]] = arith.addi %[[T9]], %[[C]] : vector<2xi32>
// CHECK: return %[[T10]] : vector<2xi32>
func @extract_contract2_int(%arg0: vector<2x3xi32>,
func.func @extract_contract2_int(%arg0: vector<2x3xi32>,
%arg1: vector<3xi32>,
%arg2: vector<2xi32>) -> vector<2xi32> {
%0 = vector.contract #matvec_trait %arg0, %arg1, %arg2
@ -126,7 +126,7 @@ func @extract_contract2_int(%arg0: vector<2x3xi32>,
// CHECK: %[[T10:.*]] = arith.addf %[[T9]], %[[C]] : vector<2xf32>
// CHECK: return %[[T10]] : vector<2xf32>
func @extract_contract3(%arg0: vector<3xf32>,
func.func @extract_contract3(%arg0: vector<3xf32>,
%arg1: vector<2x3xf32>,
%arg2: vector<2xf32>) -> vector<2xf32> {
%0 = vector.contract #vecmat_trait %arg0, %arg1, %arg2
@ -175,7 +175,7 @@ func @extract_contract3(%arg0: vector<3xf32>,
// CHECK: %[[T52:.*]] = arith.addf %[[T43]], %[[C]] : vector<2x2xf32>
// CHECK: return %[[T52]] : vector<2x2xf32>
func @extract_contract4(%arg0: vector<2x2xf32>,
func.func @extract_contract4(%arg0: vector<2x2xf32>,
%arg1: vector<2x2xf32>,
%arg2: vector<2x2xf32>) -> vector<2x2xf32> {
%0 = vector.contract #matmat_trait %arg0, %arg1, %arg2
@ -209,7 +209,7 @@ func @extract_contract4(%arg0: vector<2x2xf32>,
// CHECK: %[[T9:.*]] = arith.addf %[[T8]], %[[T4]] : f32
// CHECK: return %[[T9]] : f32
func @full_contract1(%arg0: vector<2x3xf32>,
func.func @full_contract1(%arg0: vector<2x3xf32>,
%arg1: vector<2x3xf32>,
%arg2: f32) -> f32 {
%0 = vector.contract #contraction2d_trait %arg0, %arg1, %arg2
@ -255,7 +255,7 @@ func @full_contract1(%arg0: vector<2x3xf32>,
// CHECK: %[[ACC1:.*]] = arith.addf %[[T23]], %[[ACC0]] : f32
// CHECK: return %[[ACC1]] : f32
func @full_contract2(%arg0: vector<2x3xf32>,
func.func @full_contract2(%arg0: vector<2x3xf32>,
%arg1: vector<3x2xf32>,
%arg2: f32) -> f32 {
%0 = vector.contract #contraction2d_trans_trait %arg0, %arg1, %arg2
@ -277,7 +277,7 @@ func @full_contract2(%arg0: vector<2x3xf32>,
// CHECK: %[[T7:.*]] = vector.insert %[[T6]], %[[T3]] [1] : vector<3xf32> into vector<2x3xf32>
// CHECK: return %[[T7]] : vector<2x3xf32>
func @outerproduct_noacc(%arg0: vector<2xf32>,
func.func @outerproduct_noacc(%arg0: vector<2xf32>,
%arg1: vector<3xf32>) -> vector<2x3xf32> {
%0 = vector.outerproduct %arg0, %arg1 : vector<2xf32>, vector<3xf32>
return %0: vector<2x3xf32>
@ -300,7 +300,7 @@ func @outerproduct_noacc(%arg0: vector<2xf32>,
// CHECK: %[[T9:.*]] = vector.insert %[[T8]], %[[T4]] [1] : vector<3xf32> into vector<2x3xf32>
// CHECK: return %[[T9]] : vector<2x3xf32>
func @outerproduct_acc(%arg0: vector<2xf32>,
func.func @outerproduct_acc(%arg0: vector<2xf32>,
%arg1: vector<3xf32>,
%arg2: vector<2x3xf32>) -> vector<2x3xf32> {
%0 = vector.outerproduct %arg0, %arg1, %arg2 : vector<2xf32>, vector<3xf32>
@ -320,7 +320,7 @@ func @outerproduct_acc(%arg0: vector<2xf32>,
// CHECK: %[[T6:.*]] = arith.muli %[[T5]], %[[B]] : vector<3xi32>
// CHECK: %[[T7:.*]] = vector.insert %[[T6]], %[[T3]] [1] : vector<3xi32> into vector<2x3xi32>
// CHECK: return %[[T7]] : vector<2x3xi32>
func @outerproduct_noacc_int(%arg0: vector<2xi32>,
func.func @outerproduct_noacc_int(%arg0: vector<2xi32>,
%arg1: vector<3xi32>) -> vector<2x3xi32> {
%0 = vector.outerproduct %arg0, %arg1 : vector<2xi32>, vector<3xi32>
return %0: vector<2x3xi32>
@ -344,7 +344,7 @@ func @outerproduct_noacc_int(%arg0: vector<2xi32>,
// CHECK: %[[T10:.*]] = arith.addi %[[T9]], %[[T8]] : vector<3xi32>
// CHECK: %[[T11:.*]] = vector.insert %[[T10]], %[[T5]] [1] : vector<3xi32> into vector<2x3xi32>
// CHECK: return %[[T11]] : vector<2x3xi32>
func @outerproduct_acc_int(%arg0: vector<2xi32>,
func.func @outerproduct_acc_int(%arg0: vector<2xi32>,
%arg1: vector<3xi32>,
%arg2: vector<2x3xi32>) -> vector<2x3xi32> {
%0 = vector.outerproduct %arg0, %arg1, %arg2 : vector<2xi32>, vector<3xi32>
@ -357,7 +357,7 @@ func @outerproduct_acc_int(%arg0: vector<2xi32>,
// CHECK: %[[T0:.*]] = vector.splat %[[B]] : vector<16xf32>
// CHECK: %[[T1:.*]] = arith.mulf %[[A]], %[[T0]] : vector<16xf32>
// CHECK: return %[[T1]] : vector<16xf32>
func @axpy_fp(%arg0: vector<16xf32>, %arg1: f32) -> vector<16xf32> {
func.func @axpy_fp(%arg0: vector<16xf32>, %arg1: f32) -> vector<16xf32> {
%0 = vector.outerproduct %arg0, %arg1: vector<16xf32>, f32
return %0: vector<16xf32>
}
@ -369,7 +369,7 @@ func @axpy_fp(%arg0: vector<16xf32>, %arg1: f32) -> vector<16xf32> {
// CHECK: %[[T0:.*]] = vector.splat %[[B]] : vector<16xf32>
// CHECK: %[[T1:.*]] = vector.fma %[[A]], %[[T0]], %[[C]] : vector<16xf32>
// CHECK: return %[[T1]] : vector<16xf32>
func @axpy_fp_add(%arg0: vector<16xf32>, %arg1: f32, %arg2 : vector<16xf32>) -> vector<16xf32> {
func.func @axpy_fp_add(%arg0: vector<16xf32>, %arg1: f32, %arg2 : vector<16xf32>) -> vector<16xf32> {
%0 = vector.outerproduct %arg0, %arg1, %arg2: vector<16xf32>, f32
return %0: vector<16xf32>
}
@ -380,7 +380,7 @@ func @axpy_fp_add(%arg0: vector<16xf32>, %arg1: f32, %arg2 : vector<16xf32>) ->
// CHECK: %[[T0:.*]] = vector.splat %[[B]] : vector<16xi32>
// CHECK: %[[T1:.*]] = arith.muli %[[A]], %[[T0]] : vector<16xi32>
// CHECK: return %[[T1]] : vector<16xi32>
func @axpy_int(%arg0: vector<16xi32>, %arg1: i32) -> vector<16xi32> {
func.func @axpy_int(%arg0: vector<16xi32>, %arg1: i32) -> vector<16xi32> {
%0 = vector.outerproduct %arg0, %arg1: vector<16xi32>, i32
return %0: vector<16xi32>
}
@ -393,7 +393,7 @@ func @axpy_int(%arg0: vector<16xi32>, %arg1: i32) -> vector<16xi32> {
// CHECK: %[[T1:.*]] = arith.muli %[[A]], %[[T0]] : vector<16xi32>
// CHECK: %[[T2:.*]] = arith.addi %[[T1]], %[[C]] : vector<16xi32>
// CHECK: return %[[T2]] : vector<16xi32>
func @axpy_int_add(%arg0: vector<16xi32>, %arg1: i32, %arg2: vector<16xi32>) -> vector<16xi32> {
func.func @axpy_int_add(%arg0: vector<16xi32>, %arg1: i32, %arg2: vector<16xi32>) -> vector<16xi32> {
%0 = vector.outerproduct %arg0, %arg1, %arg2: vector<16xi32>, i32
return %0: vector<16xi32>
}
@ -402,7 +402,7 @@ func @axpy_int_add(%arg0: vector<16xi32>, %arg1: i32, %arg2: vector<16xi32>) ->
// CHECK-SAME: %[[A:.*]]: vector<16xf32>
// CHECK: return %[[A]] : vector<16xf32>
func @nop_shape_cast(%arg0: vector<16xf32>) -> vector<16xf32> {
func.func @nop_shape_cast(%arg0: vector<16xf32>) -> vector<16xf32> {
%0 = vector.shape_cast %arg0 : vector<16xf32> to vector<16xf32>
return %0 : vector<16xf32>
}
@ -412,7 +412,7 @@ func @nop_shape_cast(%arg0: vector<16xf32>) -> vector<16xf32> {
// HECK-SAME: %[[A:.*]]: vector<16xf32>
// HECK: return %[[A]] : vector<16xf32>
func @cancel_shape_cast(%arg0: vector<16xf32>) -> vector<16xf32> {
func.func @cancel_shape_cast(%arg0: vector<16xf32>) -> vector<16xf32> {
%0 = vector.shape_cast %arg0 : vector<16xf32> to vector<4x4xf32>
%1 = vector.shape_cast %0 : vector<4x4xf32> to vector<16xf32>
return %1 : vector<16xf32>
@ -421,7 +421,7 @@ func @cancel_shape_cast(%arg0: vector<16xf32>) -> vector<16xf32> {
// Shape up and downcasts for 2-D vectors, for supporting conversion to
// llvm.matrix operations
// CHECK-LABEL: func @shape_casts
func @shape_casts(%a: vector<2x2xf32>) -> (vector<4xf32>, vector<2x2xf32>) {
func.func @shape_casts(%a: vector<2x2xf32>) -> (vector<4xf32>, vector<2x2xf32>) {
// CHECK-DAG: %[[cst22:.*]] = arith.constant dense<0.000000e+00> : vector<2x2xf32>
// CHECK-DAG: %[[cst:.*]] = arith.constant dense<0.000000e+00> : vector<4xf32>
// CHECK: %[[ex0:.*]] = vector.extract %{{.*}}[0] : vector<2x2xf32>
@ -474,7 +474,7 @@ func @shape_casts(%a: vector<2x2xf32>) -> (vector<4xf32>, vector<2x2xf32>) {
// CHECK: %[[T11:.*]] = vector.insert %[[T10]], %[[T9]] [1, 2] : f32 into vector<2x3xf32>
// CHECK: return %[[T11]] : vector<2x3xf32>
func @shape_cast_2d2d(%arg0 : vector<3x2xf32>) -> vector<2x3xf32> {
func.func @shape_cast_2d2d(%arg0 : vector<3x2xf32>) -> vector<2x3xf32> {
%s = vector.shape_cast %arg0: vector<3x2xf32> to vector<2x3xf32>
return %s : vector<2x3xf32>
}
@ -496,7 +496,7 @@ func @shape_cast_2d2d(%arg0 : vector<3x2xf32>) -> vector<2x3xf32> {
// CHECK: %[[T11:.*]] = vector.insert %[[T10]], %[[T9]] [5] : f32 into vector<6xf32>
// CHECK: return %[[T11]] : vector<6xf32>
func @shape_cast_3d1d(%arg0 : vector<1x3x2xf32>) -> vector<6xf32> {
func.func @shape_cast_3d1d(%arg0 : vector<1x3x2xf32>) -> vector<6xf32> {
%s = vector.shape_cast %arg0 : vector<1x3x2xf32> to vector<6xf32>
return %s : vector<6xf32>
}
@ -518,7 +518,7 @@ func @shape_cast_3d1d(%arg0 : vector<1x3x2xf32>) -> vector<6xf32> {
// CHECK: %[[T11:.*]] = vector.insert %[[T10]], %[[T9]] [1, 0, 2] : f32 into vector<2x1x3xf32>
// CHECK: return %[[T11]] : vector<2x1x3xf32>
func @shape_cast_1d3d(%arg0 : vector<6xf32>) -> vector<2x1x3xf32> {
func.func @shape_cast_1d3d(%arg0 : vector<6xf32>) -> vector<2x1x3xf32> {
%s = vector.shape_cast %arg0 : vector<6xf32> to vector<2x1x3xf32>
return %s : vector<2x1x3xf32>
}
@ -602,7 +602,7 @@ func @shape_cast_1d3d(%arg0 : vector<6xf32>) -> vector<2x1x3xf32> {
// REDUCE-NEXT: %[[r12:.*]] = vector.insert %[[s12]], %{{.*}} [1, 2] : f32 into vector<2x3xf32>
//
// REDUCE: return %[[c3]] : vector<2x3xf32>
func @matmul(%arg0: vector<2x4xf32>,
func.func @matmul(%arg0: vector<2x4xf32>,
%arg1: vector<4x3xf32>,
%arg2: vector<2x3xf32>) -> vector<2x3xf32> {
%0 = vector.contract #matmat_trait %arg0, %arg1, %arg2
@ -615,7 +615,7 @@ func @matmul(%arg0: vector<2x4xf32>,
// CHECK: %[[T0:.*]] = vector.splat %[[A]] : vector<2xf32>
// CHECK: return %[[T0]] : vector<2xf32>
func @broadcast_vec1d_from_scalar(%arg0: f32) -> vector<2xf32> {
func.func @broadcast_vec1d_from_scalar(%arg0: f32) -> vector<2xf32> {
%0 = vector.broadcast %arg0 : f32 to vector<2xf32>
return %0 : vector<2xf32>
}
@ -625,7 +625,7 @@ func @broadcast_vec1d_from_scalar(%arg0: f32) -> vector<2xf32> {
// CHECK: %[[T0:.*]] = vector.splat %[[A]] : vector<2x3xf32>
// CHECK: return %[[T0]] : vector<2x3xf32>
func @broadcast_vec2d_from_scalar(%arg0: f32) -> vector<2x3xf32> {
func.func @broadcast_vec2d_from_scalar(%arg0: f32) -> vector<2x3xf32> {
%0 = vector.broadcast %arg0 : f32 to vector<2x3xf32>
return %0 : vector<2x3xf32>
}
@ -635,7 +635,7 @@ func @broadcast_vec2d_from_scalar(%arg0: f32) -> vector<2x3xf32> {
// CHECK: %[[T0:.*]] = vector.splat %[[A]] : vector<2x3x4xf32>
// CHECK: return %[[T0]] : vector<2x3x4xf32>
func @broadcast_vec3d_from_scalar(%arg0: f32) -> vector<2x3x4xf32> {
func.func @broadcast_vec3d_from_scalar(%arg0: f32) -> vector<2x3x4xf32> {
%0 = vector.broadcast %arg0 : f32 to vector<2x3x4xf32>
return %0 : vector<2x3x4xf32>
}
@ -644,7 +644,7 @@ func @broadcast_vec3d_from_scalar(%arg0: f32) -> vector<2x3x4xf32> {
// CHECK-SAME: %[[A:.*0]]: vector<2xf32>
// CHECK: return %[[A]] : vector<2xf32>
func @broadcast_vec1d_from_vec1d(%arg0: vector<2xf32>) -> vector<2xf32> {
func.func @broadcast_vec1d_from_vec1d(%arg0: vector<2xf32>) -> vector<2xf32> {
%0 = vector.broadcast %arg0 : vector<2xf32> to vector<2xf32>
return %0 : vector<2xf32>
}
@ -657,7 +657,7 @@ func @broadcast_vec1d_from_vec1d(%arg0: vector<2xf32>) -> vector<2xf32> {
// CHECK: %[[T2:.*]] = vector.insert %[[A]], %[[T1]] [2] : vector<2xf32> into vector<3x2xf32>
// CHECK: return %[[T2]] : vector<3x2xf32>
func @broadcast_vec2d_from_vec1d(%arg0: vector<2xf32>) -> vector<3x2xf32> {
func.func @broadcast_vec2d_from_vec1d(%arg0: vector<2xf32>) -> vector<3x2xf32> {
%0 = vector.broadcast %arg0 : vector<2xf32> to vector<3x2xf32>
return %0 : vector<3x2xf32>
}
@ -675,7 +675,7 @@ func @broadcast_vec2d_from_vec1d(%arg0: vector<2xf32>) -> vector<3x2xf32> {
// CHECK: %[[T6:.*]] = vector.insert %[[T2]], %[[T5]] [3] : vector<3x2xf32> into vector<4x3x2xf32>
// CHECK: return %[[T6]] : vector<4x3x2xf32>
func @broadcast_vec3d_from_vec1d(%arg0: vector<2xf32>) -> vector<4x3x2xf32> {
func.func @broadcast_vec3d_from_vec1d(%arg0: vector<2xf32>) -> vector<4x3x2xf32> {
%0 = vector.broadcast %arg0 : vector<2xf32> to vector<4x3x2xf32>
return %0 : vector<4x3x2xf32>
}
@ -689,7 +689,7 @@ func @broadcast_vec3d_from_vec1d(%arg0: vector<2xf32>) -> vector<4x3x2xf32> {
// CHECK: %[[T3:.*]] = vector.insert %[[A]], %[[T2]] [3] : vector<3x2xf32> into vector<4x3x2xf32>
// CHECK: return %[[T3]] : vector<4x3x2xf32>
func @broadcast_vec3d_from_vec2d(%arg0: vector<3x2xf32>) -> vector<4x3x2xf32> {
func.func @broadcast_vec3d_from_vec2d(%arg0: vector<3x2xf32>) -> vector<4x3x2xf32> {
%0 = vector.broadcast %arg0 : vector<3x2xf32> to vector<4x3x2xf32>
return %0 : vector<4x3x2xf32>
}
@ -700,7 +700,7 @@ func @broadcast_vec3d_from_vec2d(%arg0: vector<3x2xf32>) -> vector<4x3x2xf32> {
// CHECK: %[[T1:.*]] = vector.splat %[[T0]] : vector<4xf32>
// CHECK: return %[[T1]] : vector<4xf32>
func @broadcast_stretch(%arg0: vector<1xf32>) -> vector<4xf32> {
func.func @broadcast_stretch(%arg0: vector<1xf32>) -> vector<4xf32> {
%0 = vector.broadcast %arg0 : vector<1xf32> to vector<4xf32>
return %0 : vector<4xf32>
}
@ -714,7 +714,7 @@ func @broadcast_stretch(%arg0: vector<1xf32>) -> vector<4xf32> {
// CHECK: %[[T3:.*]] = vector.insert %[[T0]], %[[T2]] [2] : vector<4xf32> into vector<3x4xf32>
// CHECK: return %[[T3]] : vector<3x4xf32>
func @broadcast_stretch_at_start(%arg0: vector<1x4xf32>) -> vector<3x4xf32> {
func.func @broadcast_stretch_at_start(%arg0: vector<1x4xf32>) -> vector<3x4xf32> {
%0 = vector.broadcast %arg0 : vector<1x4xf32> to vector<3x4xf32>
return %0 : vector<3x4xf32>
}
@ -736,7 +736,7 @@ func @broadcast_stretch_at_start(%arg0: vector<1x4xf32>) -> vector<3x4xf32> {
// CHECK: %[[T15:.*]] = vector.insert %[[T14]], %[[T11]] [3] : vector<3xf32> into vector<4x3xf32>
// CHECK: return %[[T15]] : vector<4x3xf32>
func @broadcast_stretch_at_end(%arg0: vector<4x1xf32>) -> vector<4x3xf32> {
func.func @broadcast_stretch_at_end(%arg0: vector<4x1xf32>) -> vector<4x3xf32> {
%0 = vector.broadcast %arg0 : vector<4x1xf32> to vector<4x3xf32>
return %0 : vector<4x3xf32>
}
@ -767,7 +767,7 @@ func @broadcast_stretch_at_end(%arg0: vector<4x1xf32>) -> vector<4x3xf32> {
// CHECK: %[[T23:.*]] = vector.insert %[[T22]], %[[T17]] [3] : vector<3x2xf32> into vector<4x3x2xf32>
// CHECK: return %[[T23]] : vector<4x3x2xf32>
func @broadcast_stretch_in_middle(%arg0: vector<4x1x2xf32>) -> vector<4x3x2xf32> {
func.func @broadcast_stretch_in_middle(%arg0: vector<4x1x2xf32>) -> vector<4x3x2xf32> {
%0 = vector.broadcast %arg0 : vector<4x1x2xf32> to vector<4x3x2xf32>
return %0 : vector<4x3x2xf32>
}
@ -776,7 +776,7 @@ func @broadcast_stretch_in_middle(%arg0: vector<4x1x2xf32>) -> vector<4x3x2xf32>
// CHECK: %[[T0:.*]] = arith.constant dense<[true, true, true, true, false, false, false, false]> : vector<8xi1>
// CHECK: return %[[T0]] : vector<8xi1>
func @genbool_1d() -> vector<8xi1> {
func.func @genbool_1d() -> vector<8xi1> {
%0 = vector.constant_mask [4] : vector<8xi1>
return %0 : vector<8xi1>
}
@ -788,7 +788,7 @@ func @genbool_1d() -> vector<8xi1> {
// CHECK: %[[T1:.*]] = vector.insert %[[C1]], %[[T0]] [1] : vector<4xi1> into vector<4x4xi1>
// CHECK: return %[[T1]] : vector<4x4xi1>
func @genbool_2d() -> vector<4x4xi1> {
func.func @genbool_2d() -> vector<4x4xi1> {
%v = vector.constant_mask [2, 2] : vector<4x4xi1>
return %v: vector<4x4xi1>
}
@ -801,7 +801,7 @@ func @genbool_2d() -> vector<4x4xi1> {
// CHECK: %[[T1:.*]] = vector.insert %[[T0]], %[[C3]] [0] : vector<3x4xi1> into vector<2x3x4xi1>
// CHECK: return %[[T1]] : vector<2x3x4xi1>
func @genbool_3d() -> vector<2x3x4xi1> {
func.func @genbool_3d() -> vector<2x3x4xi1> {
%v = vector.constant_mask [1, 1, 3] : vector<2x3x4xi1>
return %v: vector<2x3x4xi1>
}
@ -811,7 +811,7 @@ func @genbool_3d() -> vector<2x3x4xi1> {
// CHECK: %[[T0:.*]] = vector.create_mask %[[A]] : vector<3xi1>
// CHECK: return %[[T0]] : vector<3xi1>
func @genbool_var_1d(%arg0: index) -> vector<3xi1> {
func.func @genbool_var_1d(%arg0: index) -> vector<3xi1> {
%0 = vector.create_mask %arg0 : vector<3xi1>
return %0 : vector<3xi1>
}
@ -832,7 +832,7 @@ func @genbool_var_1d(%arg0: index) -> vector<3xi1> {
// CHECK: %[[T6:.*]] = vector.insert %[[T5]], %[[T3]] [1] : vector<3xi1> into vector<2x3xi1>
// CHECK: return %[[T6]] : vector<2x3xi1>
func @genbool_var_2d(%arg0: index, %arg1: index) -> vector<2x3xi1> {
func.func @genbool_var_2d(%arg0: index, %arg1: index) -> vector<2x3xi1> {
%0 = vector.create_mask %arg0, %arg1 : vector<2x3xi1>
return %0 : vector<2x3xi1>
}
@ -858,7 +858,7 @@ func @genbool_var_2d(%arg0: index, %arg1: index) -> vector<2x3xi1> {
// CHECK: %[[T9:.*]] = vector.insert %[[T8]], %[[T6]] [1] : vector<1x7xi1> into vector<2x1x7xi1>
// CHECK: return %[[T9]] : vector<2x1x7xi1>
func @genbool_var_3d(%arg0: index, %arg1: index, %arg2: index) -> vector<2x1x7xi1> {
func.func @genbool_var_3d(%arg0: index, %arg1: index, %arg2: index) -> vector<2x1x7xi1> {
%0 = vector.create_mask %arg0, %arg1, %arg2 : vector<2x1x7xi1>
return %0 : vector<2x1x7xi1>
}
@ -882,7 +882,7 @@ func @genbool_var_3d(%arg0: index, %arg1: index, %arg2: index) -> vector<2x1x7xi
// OUTERPRODUCT: %[[b0:.*]] = vector.extract %[[B]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[a0]], %[[b0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<2x3xf32>
func @matmul_0(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<2x3xf32>)
func.func @matmul_0(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<2x3xf32>)
-> vector<2x3xf32>
{
%0 = vector.contract #matmat_trait_0 %arg0, %arg1, %arg2
@ -910,7 +910,7 @@ func @matmul_0(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<2x3
// OUTERPRODUCT: %[[b0:.*]] = vector.extract %[[Bt]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[a0]], %[[b0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<2x3xf32>
func @matmul_1(%arg0: vector<2x1xf32>, %arg1: vector<3x1xf32>, %arg2: vector<2x3xf32>)
func.func @matmul_1(%arg0: vector<2x1xf32>, %arg1: vector<3x1xf32>, %arg2: vector<2x3xf32>)
-> vector<2x3xf32>
{
%0 = vector.contract #matmat_trait_1 %arg0, %arg1, %arg2
@ -936,7 +936,7 @@ func @matmul_1(%arg0: vector<2x1xf32>, %arg1: vector<3x1xf32>, %arg2: vector<2x3
// OUTERPRODUCT: %[[b0:.*]] = vector.extract %[[B]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[a0]], %[[b0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<2x3xf32>
func @matmul_2(%arg0: vector<1x2xf32>, %arg1: vector<1x3xf32>, %arg2: vector<2x3xf32>)
func.func @matmul_2(%arg0: vector<1x2xf32>, %arg1: vector<1x3xf32>, %arg2: vector<2x3xf32>)
-> vector<2x3xf32>
{
%0 = vector.contract #matmat_trait_2 %arg0, %arg1, %arg2
@ -963,7 +963,7 @@ func @matmul_2(%arg0: vector<1x2xf32>, %arg1: vector<1x3xf32>, %arg2: vector<2x3
// OUTERPRODUCT: %[[b0:.*]] = vector.extract %[[Bt]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[a0]], %[[b0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<2x3xf32>
func @matmul_3(%arg0: vector<1x2xf32>, %arg1: vector<3x1xf32>, %arg2: vector<2x3xf32>)
func.func @matmul_3(%arg0: vector<1x2xf32>, %arg1: vector<3x1xf32>, %arg2: vector<2x3xf32>)
-> vector<2x3xf32>
{
%0 = vector.contract #matmat_trait_3 %arg0, %arg1, %arg2
@ -990,7 +990,7 @@ func @matmul_3(%arg0: vector<1x2xf32>, %arg1: vector<3x1xf32>, %arg2: vector<2x3
// OUTERPRODUCT: %[[a0:.*]] = vector.extract %[[At]][0] : vector<1x2xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[b0]], %[[a0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<3x2xf32>
func @matmul_4(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
func.func @matmul_4(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
-> vector<3x2xf32>
{
%0 = vector.contract #matmat_trait_4 %arg0, %arg1, %arg2
@ -1017,7 +1017,7 @@ func @matmul_4(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2
// OUTERPRODUCT-DAG: %[[b0:.*]] = vector.extract %[[B]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[b0]], %[[a0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<3x2xf32>
func @matmul_5(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
func.func @matmul_5(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
-> vector<3x2xf32>
{
%0 = vector.contract #matmat_trait_5 %arg0, %arg1, %arg2
@ -1044,7 +1044,7 @@ func @matmul_5(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2
// OUTERPRODUCT-DAG: %[[b0:.*]] = vector.extract %[[B]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[b0]], %[[a0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<3x2xf32>
func @matmul_6(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
func.func @matmul_6(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
-> vector<3x2xf32>
{
%0 = vector.contract #matmat_trait_6 %arg0, %arg1, %arg2
@ -1071,7 +1071,7 @@ func @matmul_6(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2
// OUTERPRODUCT-DAG: %[[b0:.*]] = vector.extract %[[B]][0] : vector<1x3xf32>
// OUTERPRODUCT: %[[c0:.*]] = vector.outerproduct %[[b0]], %[[a0]], %[[C]]
// OUTERPRODUCT: return %[[c0]] : vector<3x2xf32>
func @matmul_7(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
func.func @matmul_7(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2xf32>)
-> vector<3x2xf32>
{
%0 = vector.contract #matmat_trait_7 %arg0, %arg1, %arg2
@ -1084,7 +1084,7 @@ func @matmul_7(%arg0: vector<2x1xf32>, %arg1: vector<1x3xf32>, %arg2: vector<3x2
// FILTEROUTERPRODUCT-SAME: %[[B:[a-zA-Z0-9]*]]: vector<4x4xf32>,
// FILTEROUTERPRODUCT-SAME: %[[C:[a-zA-Z0-9]*]]: vector<4x4xf32>
// FILTEROUTERPRODUCT: %[[c0:.*]] = vector.contract {{{.*}}} %[[A]], %[[B]], %[[C]]
func @matmul_4_filtered(%arg0: vector<4x4xf32>, %arg1: vector<4x4xf32>, %arg2: vector<4x4xf32>)
func.func @matmul_4_filtered(%arg0: vector<4x4xf32>, %arg1: vector<4x4xf32>, %arg2: vector<4x4xf32>)
-> vector<4x4xf32>
{
%0 = vector.contract #matmat_trait_0 %arg0, %arg1, %arg2
@ -1097,7 +1097,7 @@ func @matmul_4_filtered(%arg0: vector<4x4xf32>, %arg1: vector<4x4xf32>, %arg2: v
// FILTEROUTERPRODUCT-SAME: %[[B:[a-zA-Z0-9]*]]: vector<4x4xf32>,
// FILTEROUTERPRODUCT-SAME: %[[C:[a-zA-Z0-9]*]]: vector<3x4xf32>
// FILTEROUTERPRODUCT: %[[c0:.*]] = vector.contract {{{.*}}} %[[A]], %[[B]], %[[C]]
func @matmul_4_not_filtered(%arg0: vector<3x4xf32>, %arg1: vector<4x4xf32>, %arg2: vector<3x4xf32>)
func.func @matmul_4_not_filtered(%arg0: vector<3x4xf32>, %arg1: vector<4x4xf32>, %arg2: vector<3x4xf32>)
-> vector<3x4xf32>
{
%0 = vector.contract #matmat_trait_0 %arg0, %arg1, %arg2

18
mlir/test/Dialect/Vector/vector-distribution.mlir
View File

@ -9,7 +9,7 @@
// CHECK-NEXT: %[[ADD:.*]] = arith.addf %[[EXA]], %[[EXB]] : vector<1xf32>
// CHECK-NEXT: %[[INS:.*]] = vector.insert_map %[[ADD]], %[[ADDV]][%[[ID]]] : vector<1xf32> into vector<32xf32>
// CHECK-NEXT: return %[[INS]] : vector<32xf32>
func @distribute_vector_add(%id : index, %A: vector<32xf32>, %B: vector<32xf32>) -> vector<32xf32> {
func.func @distribute_vector_add(%id : index, %A: vector<32xf32>, %B: vector<32xf32>) -> vector<32xf32> {
%0 = arith.addf %A, %B : vector<32xf32>
return %0: vector<32xf32>
}
@ -26,7 +26,7 @@ func @distribute_vector_add(%id : index, %A: vector<32xf32>, %B: vector<32xf32>)
// CHECK-NEXT: %[[ADD:.*]] = arith.addf %[[EXC]], %[[EXB]] : vector<1xf32>
// CHECK-NEXT: %[[INS:.*]] = vector.insert_map %[[ADD]], %[[ADDV]][%[[ID]]] : vector<1xf32> into vector<32xf32>
// CHECK-NEXT: return %[[INS]] : vector<32xf32>
func @distribute_vector_add_exp(%id : index, %A: vector<32xf32>, %B: vector<32xf32>) -> vector<32xf32> {
func.func @distribute_vector_add_exp(%id : index, %A: vector<32xf32>, %B: vector<32xf32>) -> vector<32xf32> {
%C = math.exp %A : vector<32xf32>
%0 = arith.addf %C, %B : vector<32xf32>
return %0: vector<32xf32>
@ -43,7 +43,7 @@ func @distribute_vector_add_exp(%id : index, %A: vector<32xf32>, %B: vector<32xf
// CHECK-NEXT: %[[ADD2:.*]] = arith.addf %[[ADD1]], %[[EXC]] : vector<1xf32>
// CHECK-NEXT: vector.transfer_write %[[ADD2]], %{{.*}}[%[[ID]]] {{.*}} : vector<1xf32>, memref<32xf32>
// CHECK-NEXT: return
func @vector_add_read_write(%id : index, %A: memref<32xf32>, %B: memref<32xf32>, %C: memref<32xf32>, %D: memref<32xf32>) {
func.func @vector_add_read_write(%id : index, %A: memref<32xf32>, %B: memref<32xf32>, %C: memref<32xf32>, %D: memref<32xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%a = vector.transfer_read %A[%c0], %cf0: memref<32xf32>, vector<32xf32>
@ -69,7 +69,7 @@ func @vector_add_read_write(%id : index, %A: memref<32xf32>, %B: memref<32xf32>,
// CHECK-NEXT: %[[ID3:.*]] = affine.apply #[[MAP0]]()[%[[ID]]]
// CHECK-NEXT: vector.transfer_write %[[ADD]], %{{.*}}[%[[ID3]]] {{.*}} : vector<2xf32>, memref<64xf32>
// CHECK-NEXT: return
func @vector_add_cycle(%id : index, %A: memref<64xf32>, %B: memref<64xf32>, %C: memref<64xf32>) {
func.func @vector_add_cycle(%id : index, %A: memref<64xf32>, %B: memref<64xf32>, %C: memref<64xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%a = vector.transfer_read %A[%c0], %cf0: memref<64xf32>, vector<64xf32>
@ -90,7 +90,7 @@ func @vector_add_cycle(%id : index, %A: memref<64xf32>, %B: memref<64xf32>, %C:
// CHECK-NEXT: %[[ADD:.*]] = arith.addf %[[EXA]], %[[EXB]] : vector<16xf32>
// CHECK-NEXT: vector.transfer_write %[[ADD]], %{{.*}}[%[[C0]]] {{.*}} : vector<16xf32>, memref<64xf32>
// CHECK-NEXT: return
func @vector_negative_test(%id : index, %A: memref<64xf32>, %B: memref<64xf32>, %C: memref<64xf32>) {
func.func @vector_negative_test(%id : index, %A: memref<64xf32>, %B: memref<64xf32>, %C: memref<64xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%a = vector.transfer_read %A[%c0], %cf0: memref<64xf32>, vector<16xf32>
@ -110,7 +110,7 @@ func @vector_negative_test(%id : index, %A: memref<64xf32>, %B: memref<64xf32>,
// CHECK-NEXT: %[[ADD:.*]] = arith.addf %[[EXA]], %[[EXB]] : vector<2x4x1xf32>
// CHECK-NEXT: %[[INS:.*]] = vector.insert_map %[[ADD]], %[[ADDV]][%[[ID0]], %[[ID1]]] : vector<2x4x1xf32> into vector<64x4x32xf32>
// CHECK-NEXT: return %[[INS]] : vector<64x4x32xf32>
func @distribute_vector_add_3d(%id0 : index, %id1 : index,
func.func @distribute_vector_add_3d(%id0 : index, %id1 : index,
%A: vector<64x4x32xf32>, %B: vector<64x4x32xf32>) -> vector<64x4x32xf32> {
%0 = arith.addf %A, %B : vector<64x4x32xf32>
return %0: vector<64x4x32xf32>
@ -131,7 +131,7 @@ func @distribute_vector_add_3d(%id0 : index, %id1 : index,
// CHECK-NEXT: %[[ID3:.*]] = affine.apply #[[MAP0]]()[%[[ID_0]]]
// CHECK-NEXT: vector.transfer_write %[[ADD]], %{{.*}}[%[[ID3]], %[[C0]], %[[ID_1]]] {{.*}} : vector<2x4x1xf32>, memref<64x64x64xf32>
// CHECK-NEXT: return
func @vector_add_transfer_3d(%id0 : index, %id1 : index, %A: memref<64x64x64xf32>,
func.func @vector_add_transfer_3d(%id0 : index, %id1 : index, %A: memref<64x64x64xf32>,
%B: memref<64x64x64xf32>, %C: memref<64x64x64xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
@ -163,7 +163,7 @@ func @vector_add_transfer_3d(%id0 : index, %id1 : index, %A: memref<64x64x64xf32
// CHECK-NEXT: %[[ID3:.*]] = affine.apply #[[MAP0]]()[%[[ID_0]]]
// CHECK-NEXT: vector.transfer_write %[[ADD]], %{{.*}}[%[[C0]], %[[ID_1]], %[[C0]], %[[ID3]]] {permutation_map = #[[MAP3]]} : vector<2x4x1xf32>, memref<?x?x?x?xf32>
// CHECK-NEXT: return
func @vector_add_transfer_permutation(%id0 : index, %id1 : index, %A: memref<?x?x?x?xf32>,
func.func @vector_add_transfer_permutation(%id0 : index, %id1 : index, %A: memref<?x?x?x?xf32>,
%B: memref<?x?x?x?xf32>, %C: memref<?x?x?x?xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
@ -184,7 +184,7 @@ func @vector_add_transfer_permutation(%id0 : index, %id1 : index, %A: memref<?x?
// CHECK2D: %[[D:.+]] = vector.contract {{.*}} %[[A]], %[[B]], %[[C]] : vector<2x4xf32>, vector<16x4xf32> into vector<2x16xf32>
// CHECK2D: %[[R:.+]] = arith.addf %[[D]], %[[E]] : vector<2x16xf32>
// CHECK2D: vector.transfer_write %[[R]], {{.*}} : vector<2x16xf32>, memref<?x?xf32>
func @vector_add_contract(%id0 : index, %id1 : index, %A: memref<?x?xf32>,
func.func @vector_add_contract(%id0 : index, %id1 : index, %A: memref<?x?xf32>,
%B: memref<?x?xf32>, %C: memref<?x?xf32>, %D: memref<?x?xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32

34
mlir/test/Dialect/Vector/vector-dropleadunitdim-transforms.mlir
View File

@ -24,7 +24,7 @@
iterator_types = ["parallel", "parallel", "parallel", "reduction"]
}
func @cast_away_contraction_leading_one_dims(%arg0: vector<1x16x8xf32>, %arg1: vector<1x8x16xf32>, %arg2: vector<1x16x16xf32>) -> vector<1x16x16xf32> {
func.func @cast_away_contraction_leading_one_dims(%arg0: vector<1x16x8xf32>, %arg1: vector<1x8x16xf32>, %arg2: vector<1x16x16xf32>) -> vector<1x16x16xf32> {
%0 = vector.contract #contraction_trait0 %arg0, %arg1, %arg2 : vector<1x16x8xf32>, vector<1x8x16xf32> into vector<1x16x16xf32>
return %0: vector<1x16x16xf32>
}
@ -56,7 +56,7 @@ func @cast_away_contraction_leading_one_dims(%arg0: vector<1x16x8xf32>, %arg1: v
kind = #vector.kind<mul>
}
func @cast_away_contraction_leading_one_dims_transposeneeded(%arg0: vector<1x1x8xf32>, %arg1: vector<1x8x16xf32>, %arg2: vector<1x1x16xf32>) -> vector<1x1x16xf32> {
func.func @cast_away_contraction_leading_one_dims_transposeneeded(%arg0: vector<1x1x8xf32>, %arg1: vector<1x8x16xf32>, %arg2: vector<1x1x16xf32>) -> vector<1x1x16xf32> {
%0 = vector.contract #contraction_trait1 %arg0, %arg1, %arg2 : vector<1x1x8xf32>, vector<1x8x16xf32> into vector<1x1x16xf32>
return %0: vector<1x1x16xf32>
}
@ -89,7 +89,7 @@ func @cast_away_contraction_leading_one_dims_transposeneeded(%arg0: vector<1x1x8
}
func @cast_away_contraction_leading_one_dims_transposeneeded2(%arg0: vector<8x1x16xf32>, %arg1: vector<2x8x1xf32>, %arg2: vector<1x2x16xf32>) -> vector<1x2x16xf32> {
func.func @cast_away_contraction_leading_one_dims_transposeneeded2(%arg0: vector<8x1x16xf32>, %arg1: vector<2x8x1xf32>, %arg2: vector<1x2x16xf32>) -> vector<1x2x16xf32> {
%0 = vector.contract #contraction_trait2 %arg0, %arg1, %arg2 : vector<8x1x16xf32>, vector<2x8x1xf32> into vector<1x2x16xf32>
return %0: vector<1x2x16xf32>
}
@ -126,7 +126,7 @@ func @cast_away_contraction_leading_one_dims_transposeneeded2(%arg0: vector<8x1x
}
func @cast_away_contraction_leading_one_dims_nonleadingunitdim_rank4(%arg0: vector<1x8x1x16xf32>, %arg1: vector<1x2x8x1xf32>, %arg2: vector<1x1x2x16xf32>) -> vector<1x1x2x16xf32> {
func.func @cast_away_contraction_leading_one_dims_nonleadingunitdim_rank4(%arg0: vector<1x8x1x16xf32>, %arg1: vector<1x2x8x1xf32>, %arg2: vector<1x1x2x16xf32>) -> vector<1x1x2x16xf32> {
%0 = vector.contract #contraction_trait2 %arg0, %arg1, %arg2 : vector<1x8x1x16xf32>, vector<1x2x8x1xf32> into vector<1x1x2x16xf32>
return %0: vector<1x1x2x16xf32>
}
@ -159,14 +159,14 @@ func @cast_away_contraction_leading_one_dims_nonleadingunitdim_rank4(%arg0: vect
iterator_types = ["parallel","parallel", "parallel", "parallel", "reduction"]
}
func @cast_away_contraction_leading_one_dims_nonleadingunitdim_rank4_acctranspose(%arg0: vector<1x8x1x16xf32>, %arg1: vector<1x2x8x1xf32>, %arg2: vector<1x1x2x16xf32>) -> vector<1x1x2x16xf32> {
func.func @cast_away_contraction_leading_one_dims_nonleadingunitdim_rank4_acctranspose(%arg0: vector<1x8x1x16xf32>, %arg1: vector<1x2x8x1xf32>, %arg2: vector<1x1x2x16xf32>) -> vector<1x1x2x16xf32> {
%0 = vector.contract #contraction_trait3 %arg0, %arg1, %arg2 : vector<1x8x1x16xf32>, vector<1x2x8x1xf32> into vector<1x1x2x16xf32>
return %0: vector<1x1x2x16xf32>
}
// -----
// CHECK-LABEL: func @cast_away_extract_strided_slice_leading_one_dims
func @cast_away_extract_strided_slice_leading_one_dims(%arg0: vector<1x8x8xf16>) -> vector<1x1x8xf16> {
func.func @cast_away_extract_strided_slice_leading_one_dims(%arg0: vector<1x8x8xf16>) -> vector<1x1x8xf16> {
// CHECK: %[[SRC:.+]] = vector.extract %{{.*}}[0] : vector<1x8x8xf16>
// CHECK: %[[EXTRACT:.+]] = vector.extract_strided_slice %[[SRC]] {offsets = [4], sizes = [1], strides = [1]} : vector<8x8xf16> to vector<1x8xf16>
%0 = vector.extract_strided_slice %arg0 {offsets = [0, 4], sizes = [1, 1], strides = [1, 1]} : vector<1x8x8xf16> to vector<1x1x8xf16>
@ -176,7 +176,7 @@ func @cast_away_extract_strided_slice_leading_one_dims(%arg0: vector<1x8x8xf16>)
}
// CHECK-LABEL: func @cast_away_insert_strided_slice_leading_one_dims
func @cast_away_insert_strided_slice_leading_one_dims(%arg0: vector<1x8xf16>, %arg1: vector<1x8x8xf16>) -> vector<1x8x8xf16> {
func.func @cast_away_insert_strided_slice_leading_one_dims(%arg0: vector<1x8xf16>, %arg1: vector<1x8x8xf16>) -> vector<1x8x8xf16> {
// CHECK: %[[SRC:.+]] = vector.extract %{{.*}}[0] : vector<1x8xf16>
// CHECK: %[[DST:.+]] = vector.extract %{{.*}}[0] : vector<1x8x8xf16>
// CHECK: %[[INSERT:.+]] = vector.insert_strided_slice %[[SRC]], %[[DST]] {offsets = [0, 0], strides = [1]} : vector<8xf16> into vector<8x8xf16>
@ -188,7 +188,7 @@ func @cast_away_insert_strided_slice_leading_one_dims(%arg0: vector<1x8xf16>, %a
// CHECK-LABEL: func @cast_away_insert_strided_slice_leading_one_dims_one_element
// CHECK-SAME: %[[ARG0:.+]]: vector<1x1xf16>, %{{.+}}: vector<1x1x1xf16>
func @cast_away_insert_strided_slice_leading_one_dims_one_element(%arg0: vector<1x1xf16>, %arg1: vector<1x1x1xf16>) -> vector<1x1x1xf16> {
func.func @cast_away_insert_strided_slice_leading_one_dims_one_element(%arg0: vector<1x1xf16>, %arg1: vector<1x1x1xf16>) -> vector<1x1x1xf16> {
// CHECK: %[[EXT:.+]] = vector.extract %{{.*}}[0] : vector<1x1xf16>
// CHECK: %[[B:.+]] = vector.broadcast %[[EXT]] : vector<1xf16> to vector<1x1x1xf16>
%0 = vector.insert_strided_slice %arg0, %arg1 {offsets = [0, 0, 0], strides = [1, 1]} : vector<1x1xf16> into vector<1x1x1xf16>
@ -197,7 +197,7 @@ func @cast_away_insert_strided_slice_leading_one_dims_one_element(%arg0: vector<
}
// CHECK-LABEL: func @cast_away_transfer_read_leading_one_dims
func @cast_away_transfer_read_leading_one_dims(%arg0: memref<1x4x8x16xf16>) -> vector<1x4xf16> {
func.func @cast_away_transfer_read_leading_one_dims(%arg0: memref<1x4x8x16xf16>) -> vector<1x4xf16> {
// CHECK: %[[C0:.+]] = arith.constant 0 : index
%c0 = arith.constant 0 : index
// CHECK: %[[F0:.+]] = arith.constant 0.000000e+00 : f16
@ -210,7 +210,7 @@ func @cast_away_transfer_read_leading_one_dims(%arg0: memref<1x4x8x16xf16>) -> v
}
// CHECK-LABEL: func @cast_away_transfer_read_leading_one_dims_one_element
func @cast_away_transfer_read_leading_one_dims_one_element(%arg0: memref<1x1x1x1xf16>) -> vector<1x1xf16> {
func.func @cast_away_transfer_read_leading_one_dims_one_element(%arg0: memref<1x1x1x1xf16>) -> vector<1x1xf16> {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0. : f16
// CHECK: vector.broadcast %{{.+}} : vector<1xf16> to vector<1x1xf16>
@ -219,7 +219,7 @@ func @cast_away_transfer_read_leading_one_dims_one_element(%arg0: memref<1x1x1x1
}
// CHECK-LABEL: func @cast_away_transfer_write_leading_one_dims
func @cast_away_transfer_write_leading_one_dims(%arg0: memref<1x4x8x16xf16>, %arg1: vector<1x4xf16>) {
func.func @cast_away_transfer_write_leading_one_dims(%arg0: memref<1x4x8x16xf16>, %arg1: vector<1x4xf16>) {
// CHECK: %[[C0:.+]] = arith.constant 0 : index
%c0 = arith.constant 0 : index
// CHECK: %[[CAST:.+]] = vector.extract %{{.*}}[0] : vector<1x4xf16>
@ -230,7 +230,7 @@ func @cast_away_transfer_write_leading_one_dims(%arg0: memref<1x4x8x16xf16>, %ar
}
// CHECK-LABEL: func @cast_away_transfer_write_leading_one_dims_one_element
func @cast_away_transfer_write_leading_one_dims_one_element(%arg0: memref<1x1x1x1xf16>, %arg1: vector<1x1xf16>) {
func.func @cast_away_transfer_write_leading_one_dims_one_element(%arg0: memref<1x1x1x1xf16>, %arg1: vector<1x1xf16>) {
%c0 = arith.constant 0 : index
// CHECK: vector.extract %{{.+}}[0] : vector<1x1xf16>
vector.transfer_write %arg1, %arg0[%c0, %c0, %c0, %c0] {in_bounds = [true, true]} : vector<1x1xf16>, memref<1x1x1x1xf16>
@ -238,7 +238,7 @@ func @cast_away_transfer_write_leading_one_dims_one_element(%arg0: memref<1x1x1x
}
// CHECK-LABEL: func @cast_away_elementwise_leading_one_dims
func @cast_away_elementwise_leading_one_dims(
func.func @cast_away_elementwise_leading_one_dims(
%arg0: vector<1x1x8xf32>, %arg1: f32, %arg2: vector<1x4xf32>,
%arg3: vector<1x4xf32>, %arg4: i1) ->
(vector<1x1x8xf32>, vector<1x4xi1>, vector<1x4xf32>, vector<1x4xf32>) {
@ -271,7 +271,7 @@ func @cast_away_elementwise_leading_one_dims(
// CHECK: %[[INSERT:.+]] = vector.insert %[[S]], %[[EXTRACT]] [0] : f32 into vector<4xf32>
// CHECK: %[[BCAST:.+]] = vector.broadcast %[[INSERT]] : vector<4xf32> to vector<1x1x4xf32>
// CHECK: return %[[BCAST]]
func @cast_away_insert_leading_one_dims_scalar(%s: f32, %v: vector<1x1x4xf32>) -> vector<1x1x4xf32> {
func.func @cast_away_insert_leading_one_dims_scalar(%s: f32, %v: vector<1x1x4xf32>) -> vector<1x1x4xf32> {
%0 = vector.insert %s, %v [0, 0, 0] : f32 into vector<1x1x4xf32>
return %0: vector<1x1x4xf32>
}
@ -280,7 +280,7 @@ func @cast_away_insert_leading_one_dims_scalar(%s: f32, %v: vector<1x1x4xf32>) -
// CHECK-SAME: (%[[S:.+]]: vector<4xf32>, %[[V:.+]]: vector<1x1x4xf32>)
// CHECK: %[[BCAST:.+]] = vector.broadcast %[[S]] : vector<4xf32> to vector<1x1x4xf32>
// CHECK: return %[[BCAST]]
func @cast_away_insert_leading_one_dims_rank1(%s: vector<4xf32>, %v: vector<1x1x4xf32>) -> vector<1x1x4xf32> {
func.func @cast_away_insert_leading_one_dims_rank1(%s: vector<4xf32>, %v: vector<1x1x4xf32>) -> vector<1x1x4xf32> {
%0 = vector.insert %s, %v [0, 0] : vector<4xf32> into vector<1x1x4xf32>
return %0: vector<1x1x4xf32>
}
@ -290,7 +290,7 @@ func @cast_away_insert_leading_one_dims_rank1(%s: vector<4xf32>, %v: vector<1x1x
// CHECK: %[[EXTRACT:.+]] = vector.extract %[[S]][0] : vector<1x4xf32>
// CHECK: %[[BCAST:.+]] = vector.broadcast %[[EXTRACT]] : vector<4xf32> to vector<1x1x4xf32>
// CHECK: return %[[BCAST]]
func @cast_away_insert_leading_one_dims_rank2(%s: vector<1x4xf32>, %v: vector<1x1x4xf32>) -> vector<1x1x4xf32> {
func.func @cast_away_insert_leading_one_dims_rank2(%s: vector<1x4xf32>, %v: vector<1x1x4xf32>) -> vector<1x1x4xf32> {
%0 = vector.insert %s, %v [0] : vector<1x4xf32> into vector<1x1x4xf32>
return %0: vector<1x1x4xf32>
}
@ -300,7 +300,7 @@ func @cast_away_insert_leading_one_dims_rank2(%s: vector<1x4xf32>, %v: vector<1x
// CHECK: %[[EXTRACT:.+]] = vector.extract %[[S]][0] : vector<1x4xf32>
// CHECK: %[[INSERT:.+]] = vector.insert %[[EXTRACT]], %[[V]] [5, 0] : vector<4xf32> into vector<8x1x4xf32>
// CHECK: return %[[INSERT]]
func @cast_away_insert_leading_one_dims_non_one_dest(%s: vector<1x4xf32>, %v: vector<8x1x4xf32>) -> vector<8x1x4xf32> {
func.func @cast_away_insert_leading_one_dims_non_one_dest(%s: vector<1x4xf32>, %v: vector<8x1x4xf32>) -> vector<8x1x4xf32> {
%0 = vector.insert %s, %v [5] : vector<1x4xf32> into vector<8x1x4xf32>
return %0: vector<8x1x4xf32>
}

28
mlir/test/Dialect/Vector/vector-mem-transforms.mlir
View File

@ -6,7 +6,7 @@
// CHECK-DAG: %[[C:.*]] = arith.constant 0 : index
// CHECK-NEXT: %[[T:.*]] = vector.load %[[A0]][%[[C]]] : memref<?xf32>, vector<16xf32>
// CHECK-NEXT: return %[[T]] : vector<16xf32>
func @maskedload0(%base: memref<?xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @maskedload0(%base: memref<?xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
%ld = vector.maskedload %base[%c0], %mask, %pass_thru
@ -20,7 +20,7 @@ func @maskedload0(%base: memref<?xf32>, %pass_thru: vector<16xf32>) -> vector<16
// CHECK-DAG: %[[C:.*]] = arith.constant 0 : index
// CHECK-NEXT: %[[T:.*]] = vector.load %[[A0]][%[[C]]] : memref<16xf32>, vector<16xf32>
// CHECK-NEXT: return %[[T]] : vector<16xf32>
func @maskedload1(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @maskedload1(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
%ld = vector.maskedload %base[%c0], %mask, %pass_thru
@ -32,7 +32,7 @@ func @maskedload1(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<1
// CHECK-SAME: %[[A0:.*]]: memref<16xf32>,
// CHECK-SAME: %[[A1:.*]]: vector<16xf32>) -> vector<16xf32> {
// CHECK-NEXT: return %[[A1]] : vector<16xf32>
func @maskedload2(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @maskedload2(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [0] : vector<16xi1>
%ld = vector.maskedload %base[%c0], %mask, %pass_thru
@ -46,7 +46,7 @@ func @maskedload2(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<1
// CHECK-DAG: %[[C:.*]] = arith.constant 8 : index
// CHECK-NEXT: %[[T:.*]] = vector.load %[[A0]][%[[C]]] : memref<?xf32>, vector<16xf32>
// CHECK-NEXT: return %[[T]] : vector<16xf32>
func @maskedload3(%base: memref<?xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @maskedload3(%base: memref<?xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c8 = arith.constant 8 : index
%mask = vector.constant_mask [16] : vector<16xi1>
%ld = vector.maskedload %base[%c8], %mask, %pass_thru
@ -60,7 +60,7 @@ func @maskedload3(%base: memref<?xf32>, %pass_thru: vector<16xf32>) -> vector<16
// CHECK-NEXT: %[[C:.*]] = arith.constant 0 : index
// CHECK-NEXT: vector.store %[[A1]], %[[A0]][%[[C]]] : memref<16xf32>, vector<16xf32>
// CHECK-NEXT: return
func @maskedstore1(%base: memref<16xf32>, %value: vector<16xf32>) {
func.func @maskedstore1(%base: memref<16xf32>, %value: vector<16xf32>) {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
vector.maskedstore %base[%c0], %mask, %value : memref<16xf32>, vector<16xi1>, vector<16xf32>
@ -71,7 +71,7 @@ func @maskedstore1(%base: memref<16xf32>, %value: vector<16xf32>) {
// CHECK-SAME: %[[A0:.*]]: memref<16xf32>,
// CHECK-SAME: %[[A1:.*]]: vector<16xf32>) {
// CHECK-NEXT: return
func @maskedstore2(%base: memref<16xf32>, %value: vector<16xf32>) {
func.func @maskedstore2(%base: memref<16xf32>, %value: vector<16xf32>) {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [0] : vector<16xi1>
vector.maskedstore %base[%c0], %mask, %value : memref<16xf32>, vector<16xi1>, vector<16xf32>
@ -86,7 +86,7 @@ func @maskedstore2(%base: memref<16xf32>, %value: vector<16xf32>) {
// CHECK-NEXT: %[[M:.*]] = vector.constant_mask [16] : vector<16xi1>
// CHECK-NEXT: %[[G:.*]] = vector.gather %[[A0]][%[[C]]] [%[[A1]]], %[[M]], %[[A2]] : memref<16xf32>, vector<16xi32>, vector<16xi1>, vector<16xf32> into vector<16xf32>
// CHECK-NEXT: return %[[G]] : vector<16xf32>
func @gather1(%base: memref<16xf32>, %indices: vector<16xi32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @gather1(%base: memref<16xf32>, %indices: vector<16xi32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
%ld = vector.gather %base[%c0][%indices], %mask, %pass_thru
@ -99,7 +99,7 @@ func @gather1(%base: memref<16xf32>, %indices: vector<16xi32>, %pass_thru: vecto
// CHECK-SAME: %[[A1:.*]]: vector<16xi32>,
// CHECK-SAME: %[[A2:.*]]: vector<16xf32>) -> vector<16xf32> {
// CHECK-NEXT: return %[[A2]] : vector<16xf32>
func @gather2(%base: memref<16xf32>, %indices: vector<16xi32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @gather2(%base: memref<16xf32>, %indices: vector<16xi32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [0] : vector<16xi1>
%ld = vector.gather %base[%c0][%indices], %mask, %pass_thru
@ -115,7 +115,7 @@ func @gather2(%base: memref<16xf32>, %indices: vector<16xi32>, %pass_thru: vecto
// CHECK-NEXT: %[[M:.*]] = vector.constant_mask [16] : vector<16xi1>
// CHECK-NEXT: vector.scatter %[[A0]][%[[C]]] [%[[A1]]], %[[M]], %[[A2]] : memref<16xf32>, vector<16xi32>, vector<16xi1>, vector<16xf32>
// CHECK-NEXT: return
func @scatter1(%base: memref<16xf32>, %indices: vector<16xi32>, %value: vector<16xf32>) {
func.func @scatter1(%base: memref<16xf32>, %indices: vector<16xi32>, %value: vector<16xf32>) {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
vector.scatter %base[%c0][%indices], %mask, %value
@ -128,7 +128,7 @@ func @scatter1(%base: memref<16xf32>, %indices: vector<16xi32>, %value: vector<1
// CHECK-SAME: %[[A1:.*]]: vector<16xi32>,
// CHECK-SAME: %[[A2:.*]]: vector<16xf32>) {
// CHECK-NEXT: return
func @scatter2(%base: memref<16xf32>, %indices: vector<16xi32>, %value: vector<16xf32>) {
func.func @scatter2(%base: memref<16xf32>, %indices: vector<16xi32>, %value: vector<16xf32>) {
%c0 = arith.constant 0 : index
%0 = vector.type_cast %base : memref<16xf32> to memref<vector<16xf32>>
%mask = vector.constant_mask [0] : vector<16xi1>
@ -143,7 +143,7 @@ func @scatter2(%base: memref<16xf32>, %indices: vector<16xi32>, %value: vector<1
// CHECK-DAG: %[[C:.*]] = arith.constant 0 : index
// CHECK-NEXT: %[[T:.*]] = vector.load %[[A0]][%[[C]]] : memref<16xf32>, vector<16xf32>
// CHECK-NEXT: return %[[T]] : vector<16xf32>
func @expand1(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @expand1(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
%ld = vector.expandload %base[%c0], %mask, %pass_thru
@ -155,7 +155,7 @@ func @expand1(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf3
// CHECK-SAME: %[[A0:.*]]: memref<16xf32>,
// CHECK-SAME: %[[A1:.*]]: vector<16xf32>) -> vector<16xf32> {
// CHECK-NEXT: return %[[A1]] : vector<16xf32>
func @expand2(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
func.func @expand2(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf32> {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [0] : vector<16xi1>
%ld = vector.expandload %base[%c0], %mask, %pass_thru
@ -169,7 +169,7 @@ func @expand2(%base: memref<16xf32>, %pass_thru: vector<16xf32>) -> vector<16xf3
// CHECK-NEXT: %[[C:.*]] = arith.constant 0 : index
// CHECK-NEXT: vector.store %[[A1]], %[[A0]][%[[C]]] : memref<16xf32>, vector<16xf32>
// CHECK-NEXT: return
func @compress1(%base: memref<16xf32>, %value: vector<16xf32>) {
func.func @compress1(%base: memref<16xf32>, %value: vector<16xf32>) {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [16] : vector<16xi1>
vector.compressstore %base[%c0], %mask, %value : memref<16xf32>, vector<16xi1>, vector<16xf32>
@ -180,7 +180,7 @@ func @compress1(%base: memref<16xf32>, %value: vector<16xf32>) {
// CHECK-SAME: %[[A0:.*]]: memref<16xf32>,
// CHECK-SAME: %[[A1:.*]]: vector<16xf32>) {
// CHECK-NEXT: return
func @compress2(%base: memref<16xf32>, %value: vector<16xf32>) {
func.func @compress2(%base: memref<16xf32>, %value: vector<16xf32>) {
%c0 = arith.constant 0 : index
%mask = vector.constant_mask [0] : vector<16xi1>
vector.compressstore %base[%c0], %mask, %value : memref<16xf32>, vector<16xi1>, vector<16xf32>

10
mlir/test/Dialect/Vector/vector-multi-reduction-lowering.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt %s -test-vector-multi-reduction-lowering-patterns | FileCheck %s
func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
func.func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
%0 = vector.multi_reduction <mul>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
return %0 : vector<2xf32>
}
@ -17,7 +17,7 @@ func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
// CHECK: %[[RESULT_VEC:.+]] = vector.insertelement %[[RV1:.+]], %[[RESULT_VEC_1]][%[[C1]] : index] : vector<2xf32>
// CHECK: return %[[RESULT_VEC]]
func @vector_multi_reduction_to_scalar(%arg0: vector<2x4xf32>) -> f32 {
func.func @vector_multi_reduction_to_scalar(%arg0: vector<2x4xf32>) -> f32 {
%0 = vector.multi_reduction <mul>, %arg0 [0, 1] : vector<2x4xf32> to f32
return %0 : f32
}
@ -29,7 +29,7 @@ func @vector_multi_reduction_to_scalar(%arg0: vector<2x4xf32>) -> f32 {
// CHECK: %[[RES:.*]] = vector.extract %[[INSERTED]][0] : vector<1xf32>
// CHECK: return %[[RES]]
func @vector_reduction_inner(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
func.func @vector_reduction_inner(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
%0 = vector.multi_reduction <add>, %arg0 [2, 3] : vector<2x3x4x5xi32> to vector<2x3xi32>
return %0 : vector<2x3xi32>
}
@ -65,7 +65,7 @@ func @vector_reduction_inner(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
// CHECK: return %[[RESULT]]
func @vector_multi_reduction_transposed(%arg0: vector<2x3x4x5xf32>) -> vector<2x5xf32> {
func.func @vector_multi_reduction_transposed(%arg0: vector<2x3x4x5xf32>) -> vector<2x5xf32> {
%0 = vector.multi_reduction <add>, %arg0 [1, 2] : vector<2x3x4x5xf32> to vector<2x5xf32>
return %0 : vector<2x5xf32>
}
@ -77,7 +77,7 @@ func @vector_multi_reduction_transposed(%arg0: vector<2x3x4x5xf32>) -> vector<2x
// CHECK: %[[RESULT:.+]] = vector.shape_cast %{{.*}} : vector<10xf32> to vector<2x5xf32>
// CHECK: return %[[RESULT]]
func @vector_multi_reduction_ordering(%arg0: vector<3x2x4xf32>) -> vector<2x4xf32> {
func.func @vector_multi_reduction_ordering(%arg0: vector<3x2x4xf32>) -> vector<2x4xf32> {
%0 = vector.multi_reduction <mul>, %arg0 [0] : vector<3x2x4xf32> to vector<2x4xf32>
return %0 : vector<2x4xf32>
}

18
mlir/test/Dialect/Vector/vector-multi-reduction-outer-lowering.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt %s -test-vector-multi-reduction-lowering-patterns="use-outer-reductions" | FileCheck %s
func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
func.func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
%0 = vector.multi_reduction <mul>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
return %0 : vector<2xf32>
}
@ -17,7 +17,7 @@ func @vector_multi_reduction(%arg0: vector<2x4xf32>) -> vector<2xf32> {
// CHECK: %[[RESULT_VEC:.+]] = arith.mulf %[[V3]], %[[RV012]] : vector<2xf32>
// CHECK: return %[[RESULT_VEC]] : vector<2xf32>
func @vector_multi_reduction_min(%arg0: vector<2x4xf32>) -> vector<2xf32> {
func.func @vector_multi_reduction_min(%arg0: vector<2x4xf32>) -> vector<2xf32> {
%0 = vector.multi_reduction <minf>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
return %0 : vector<2xf32>
}
@ -34,7 +34,7 @@ func @vector_multi_reduction_min(%arg0: vector<2x4xf32>) -> vector<2xf32> {
// CHECK: %[[RESULT_VEC:.+]] = arith.minf %[[V3]], %[[RV012]] : vector<2xf32>
// CHECK: return %[[RESULT_VEC]] : vector<2xf32>
func @vector_multi_reduction_max(%arg0: vector<2x4xf32>) -> vector<2xf32> {
func.func @vector_multi_reduction_max(%arg0: vector<2x4xf32>) -> vector<2xf32> {
%0 = vector.multi_reduction <maxf>, %arg0 [1] : vector<2x4xf32> to vector<2xf32>
return %0 : vector<2xf32>
}
@ -51,7 +51,7 @@ func @vector_multi_reduction_max(%arg0: vector<2x4xf32>) -> vector<2xf32> {
// CHECK: %[[RESULT_VEC:.+]] = arith.maxf %[[V3]], %[[RV012]] : vector<2xf32>
// CHECK: return %[[RESULT_VEC]] : vector<2xf32>
func @vector_multi_reduction_and(%arg0: vector<2x4xi32>) -> vector<2xi32> {
func.func @vector_multi_reduction_and(%arg0: vector<2x4xi32>) -> vector<2xi32> {
%0 = vector.multi_reduction <and>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
return %0 : vector<2xi32>
}
@ -68,7 +68,7 @@ func @vector_multi_reduction_and(%arg0: vector<2x4xi32>) -> vector<2xi32> {
// CHECK: %[[RESULT_VEC:.+]] = arith.andi %[[V3]], %[[RV012]] : vector<2xi32>
// CHECK: return %[[RESULT_VEC]] : vector<2xi32>
func @vector_multi_reduction_or(%arg0: vector<2x4xi32>) -> vector<2xi32> {
func.func @vector_multi_reduction_or(%arg0: vector<2x4xi32>) -> vector<2xi32> {
%0 = vector.multi_reduction <or>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
return %0 : vector<2xi32>
}
@ -85,7 +85,7 @@ func @vector_multi_reduction_or(%arg0: vector<2x4xi32>) -> vector<2xi32> {
// CHECK: %[[RESULT_VEC:.+]] = arith.ori %[[V3]], %[[RV012]] : vector<2xi32>
// CHECK: return %[[RESULT_VEC]] : vector<2xi32>
func @vector_multi_reduction_xor(%arg0: vector<2x4xi32>) -> vector<2xi32> {
func.func @vector_multi_reduction_xor(%arg0: vector<2x4xi32>) -> vector<2xi32> {
%0 = vector.multi_reduction <xor>, %arg0 [1] : vector<2x4xi32> to vector<2xi32>
return %0 : vector<2xi32>
}
@ -103,7 +103,7 @@ func @vector_multi_reduction_xor(%arg0: vector<2x4xi32>) -> vector<2xi32> {
// CHECK: return %[[RESULT_VEC]] : vector<2xi32>
func @vector_reduction_outer(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
func.func @vector_reduction_outer(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
%0 = vector.multi_reduction <add>, %arg0 [2, 3] : vector<2x3x4x5xi32> to vector<2x3xi32>
return %0 : vector<2x3xi32>
}
@ -157,14 +157,14 @@ func @vector_reduction_outer(%arg0: vector<2x3x4x5xi32>) -> vector<2x3xi32> {
// This test is mainly to catch a bug that running
// `InnerOuterDimReductionConversion` on this function results in an
// infinite loop. So just check that some value is returned.
func @vector_reduction_1D(%arg0 : vector<2xf32>) -> f32 {
func.func @vector_reduction_1D(%arg0 : vector<2xf32>) -> f32 {
%0 = vector.multi_reduction #vector.kind<maxf>, %arg0 [0] : vector<2xf32> to f32
return %0 : f32
}
// CHECK-LABEL: func @vector_reduction_1D
// CHECK: return %{{.+}}
func @vector_multi_reduction_to_scalar(%arg0: vector<2x3xf32>) -> f32 {
func.func @vector_multi_reduction_to_scalar(%arg0: vector<2x3xf32>) -> f32 {
%0 = vector.multi_reduction <add>, %arg0 [0, 1] : vector<2x3xf32> to f32
return %0 : f32
}

24
mlir/test/Dialect/Vector/vector-reduce-to-contract.mlir
View File

@ -9,7 +9,7 @@
// CHECK-SAME: iterator_types = ["parallel", "reduction", "parallel"], kind = #vector.kind<add>}
// CHECK-SAME: %{{.*}}, %{{.*}}, %[[C0]] : vector<8x32x16xf32>, vector<8x32x16xf32> into vector<8x16xf32>
// CHECK-NEXT: return %[[R]] : vector<8x16xf32>
func @multidimreduction_contract(
func.func @multidimreduction_contract(
%arg0: vector<8x32x16xf32>,%arg1: vector<8x32x16xf32>) -> vector<8x16xf32> {
%0 = arith.mulf %arg0, %arg1 : vector<8x32x16xf32>
%1 = vector.multi_reduction <add>, %0 [1] : vector<8x32x16xf32> to vector<8x16xf32>
@ -27,7 +27,7 @@ func @multidimreduction_contract(
// CHECK-SAME: iterator_types = ["parallel", "reduction", "parallel"], kind = #vector.kind<add>}
// CHECK-SAME: %{{.*}}, %{{.*}}, %[[C0]] : vector<8x32x16xi32>, vector<8x32x16xi32> into vector<8x16xi32>
// CHECK-NEXT: return %[[R]] : vector<8x16xi32>
func @multidimreduction_contract_int(
func.func @multidimreduction_contract_int(
%arg0: vector<8x32x16xi32>,%arg1: vector<8x32x16xi32>) -> vector<8x16xi32> {
%0 = arith.muli %arg0, %arg1 : vector<8x32x16xi32>
%1 = vector.multi_reduction <add>, %0 [1] : vector<8x32x16xi32> to vector<8x16xi32>
@ -50,7 +50,7 @@ func @multidimreduction_contract_int(
// CHECK-SAME: iterator_types = ["parallel", "parallel", "reduction"], kind = #vector.kind<add>}
// CHECK-SAME: %[[ARG0]], %{{.*}}, %[[C0]] : vector<32x16x8xf32>, vector<8x32x16xf32> into vector<8x32xf32>
// CHECK-NEXT: return %[[R]] : vector<8x32xf32>
func @contract_transpose(
func.func @contract_transpose(
%arg0: vector<32x16x8xf32>, %arg1: vector<8x32x16xf32>) -> vector<8x32xf32> {
%cst = arith.constant dense<0.000000e+00> : vector<8x32xf32>
%0 = vector.transpose %arg0, [2, 0, 1] : vector<32x16x8xf32> to vector<8x32x16xf32>
@ -76,7 +76,7 @@ func @contract_transpose(
// CHECK-SAME: iterator_types = ["parallel", "parallel", "reduction"], kind = #vector.kind<add>}
// CHECK-SAME: %[[ARG0]], %{{.*}}, %[[C0]] : vector<32x16xf32>, vector<8x32x16xf32> into vector<8x32xf32>
// CHECK-NEXT: return %[[R]] : vector<8x32xf32>
func @contract_broadcast(
func.func @contract_broadcast(
%arg0: vector<32x16xf32>, %arg1: vector<8x32x16xf32>) -> vector<8x32xf32> {
%cst = arith.constant dense<0.000000e+00> : vector<8x32xf32>
%0 = vector.broadcast %arg0 : vector<32x16xf32> to vector<8x32x16xf32>
@ -93,7 +93,7 @@ func @contract_broadcast(
// -----
func @broadcast_vector_extsi(%a : vector<4xi8>) -> vector<2x4xi32> {
func.func @broadcast_vector_extsi(%a : vector<4xi8>) -> vector<2x4xi32> {
// CHECK: %[[EXT:.+]] = arith.extsi %{{.+}} : vector<4xi8> to vector<4xi32>
// CHECK: vector.broadcast %[[EXT:.+]] : vector<4xi32> to vector<2x4xi32>
%b = vector.broadcast %a : vector<4xi8> to vector<2x4xi8>
@ -103,7 +103,7 @@ func @broadcast_vector_extsi(%a : vector<4xi8>) -> vector<2x4xi32> {
// -----
func @broadcast_scalar_extsi(%a : i8) -> vector<2x4xi32> {
func.func @broadcast_scalar_extsi(%a : i8) -> vector<2x4xi32> {
// CHECK: %[[EXT:.+]] = arith.extsi %{{.+}} : i8 to i32
// CHECK: vector.broadcast %[[EXT]] : i32 to vector<2x4xi32>
%b = vector.broadcast %a : i8 to vector<2x4xi8>
@ -113,7 +113,7 @@ func @broadcast_scalar_extsi(%a : i8) -> vector<2x4xi32> {
// -----
func @transpose_extsi(%a : vector<4x2xi8>) -> vector<2x4xi32> {
func.func @transpose_extsi(%a : vector<4x2xi8>) -> vector<2x4xi32> {
// CHECK: %[[EXT:.+]] = arith.extsi %{{.+}} : vector<4x2xi8> to vector<4x2xi32>
// CHECK: vector.transpose %[[EXT]], [1, 0] : vector<4x2xi32> to vector<2x4xi32>
%b = vector.transpose %a, [1, 0]: vector<4x2xi8> to vector<2x4xi8>
@ -133,7 +133,7 @@ func @transpose_extsi(%a : vector<4x2xi8>) -> vector<2x4xi32> {
// CHECK: %[[T:.+]] = vector.transpose %[[ADD]], [1, 0]
// CHECK: return %[[T]]
func @transpose_elementwise_same_type(%a : vector<4x2xf32>, %b : vector<4x2xf32>) -> vector<2x4xf32> {
func.func @transpose_elementwise_same_type(%a : vector<4x2xf32>, %b : vector<4x2xf32>) -> vector<2x4xf32> {
%at = vector.transpose %a, [1, 0]: vector<4x2xf32> to vector<2x4xf32>
%bt = vector.transpose %b, [1, 0]: vector<4x2xf32> to vector<2x4xf32>
%r = arith.addf %at, %bt : vector<2x4xf32>
@ -147,7 +147,7 @@ func @transpose_elementwise_same_type(%a : vector<4x2xf32>, %b : vector<4x2xf32>
// CHECK: %[[S:.+]] = arith.select %[[COND]], %[[A]], %[[B]] : vector<4x2xi1>, vector<4x2xf32>
// CHECK: %[[T:.+]] = vector.transpose %[[S]], [1, 0] : vector<4x2xf32> to vector<2x4xf32>
// CHECK: return %[[T]]
func @transpose_elementwise_diff_operand_types(%cond: vector<4x2xi1>, %a : vector<4x2xf32>, %b : vector<4x2xf32>) -> vector<2x4xf32> {
func.func @transpose_elementwise_diff_operand_types(%cond: vector<4x2xi1>, %a : vector<4x2xf32>, %b : vector<4x2xf32>) -> vector<2x4xf32> {
%condt = vector.transpose %cond, [1, 0]: vector<4x2xi1> to vector<2x4xi1>
%at = vector.transpose %a, [1, 0]: vector<4x2xf32> to vector<2x4xf32>
%bt = vector.transpose %b, [1, 0]: vector<4x2xf32> to vector<2x4xf32>
@ -162,7 +162,7 @@ func @transpose_elementwise_diff_operand_types(%cond: vector<4x2xi1>, %a : vecto
// CHECK: %[[CMP:.+]] = arith.cmpf olt, %[[A]], %[[B]] : vector<4x2xf32>
// CHECK: %[[T:.+]] = vector.transpose %[[CMP]], [1, 0] : vector<4x2xi1> to vector<2x4xi1>
// CHECK: return %[[T]]
func @transpose_elementwise_diff_operand_result_type(%a : vector<4x2xf32>, %b : vector<4x2xf32>) -> vector<2x4xi1> {
func.func @transpose_elementwise_diff_operand_result_type(%a : vector<4x2xf32>, %b : vector<4x2xf32>) -> vector<2x4xi1> {
%at = vector.transpose %a, [1, 0]: vector<4x2xf32> to vector<2x4xf32>
%bt = vector.transpose %b, [1, 0]: vector<4x2xf32> to vector<2x4xf32>
%r = arith.cmpf olt, %at, %bt : vector<2x4xf32>
@ -178,7 +178,7 @@ func @transpose_elementwise_diff_operand_result_type(%a : vector<4x2xf32>, %b :
// CHECK: %[[T:.+]] = vector.transpose %[[ADD]], [1, 0, 3, 2] : vector<4x6x3x2xf32> to vector<6x4x2x3xf32>
// CHECK: return %[[T:.+]] : vector<6x4x2x3xf32>
func @transpose_elementwise_splat_constant(%a : vector<4x6x3x2xf32>) -> vector<6x4x2x3xf32> {
func.func @transpose_elementwise_splat_constant(%a : vector<4x6x3x2xf32>) -> vector<6x4x2x3xf32> {
%b = arith.constant dense<5.0> : vector<6x4x2x3xf32>
%at = vector.transpose %a, [1, 0, 3, 2]: vector<4x6x3x2xf32> to vector<6x4x2x3xf32>
%r = arith.addf %at, %b : vector<6x4x2x3xf32>
@ -191,7 +191,7 @@ func @transpose_elementwise_splat_constant(%a : vector<4x6x3x2xf32>) -> vector<6
// CHECK: vector.transpose
// CHECK: vector.transpose
// CHECK: arith.addf
func @transpose_elementwise_diff_map(%a : vector<4x6x3x2xf32>, %b: vector<6x2x4x3xf32>) -> vector<6x4x2x3xf32> {
func.func @transpose_elementwise_diff_map(%a : vector<4x6x3x2xf32>, %b: vector<6x2x4x3xf32>) -> vector<6x4x2x3xf32> {
%at = vector.transpose %a, [1, 0, 3, 2]: vector<4x6x3x2xf32> to vector<6x4x2x3xf32>
%bt = vector.transpose %b, [0, 2, 1, 3]: vector<6x2x4x3xf32> to vector<6x4x2x3xf32>
%r = arith.addf %at, %bt : vector<6x4x2x3xf32>

2
mlir/test/Dialect/Vector/vector-scalable-memcpy.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt %s -convert-vector-to-llvm | mlir-opt | FileCheck %s
// CHECK: vector_scalable_memcopy([[SRC:%arg[0-9]+]]: memref<?xf32>, [[DST:%arg[0-9]+]]
func @vector_scalable_memcopy(%src : memref<?xf32>, %dst : memref<?xf32>, %size : index) {
func.func @vector_scalable_memcopy(%src : memref<?xf32>, %dst : memref<?xf32>, %size : index) {
%c0 = arith.constant 0 : index
%c4 = arith.constant 4 : index
%vs = vector.vscale

10
mlir/test/Dialect/Vector/vector-scan-transforms.mlir
View File

@ -12,7 +12,7 @@
// CHECK: %[[G:.*]] = vector.extract %[[E]][0] : vector<1xi32>
// CHECK: %[[H:.*]] = vector.broadcast %[[G]] : i32 to vector<i32>
// CHECK: return %[[F]], %[[H]] : vector<2xi32>, vector<i32>
func @scan1d_inc(%arg0 : vector<2xi32>, %arg1 : vector<i32>) -> (vector<2xi32>, vector<i32>) {
func.func @scan1d_inc(%arg0 : vector<2xi32>, %arg1 : vector<i32>) -> (vector<2xi32>, vector<i32>) {
%0:2 = vector.scan <add>, %arg0, %arg1 {inclusive = true, reduction_dim = 0} :
vector<2xi32>, vector<i32>
return %0#0, %0#1 : vector<2xi32>, vector<i32>
@ -30,7 +30,7 @@ func @scan1d_inc(%arg0 : vector<2xi32>, %arg1 : vector<i32>) -> (vector<2xi32>,
// CHECK: %[[G:.*]] = vector.extract %[[E]][0] : vector<1xi32>
// CHECK: %[[H:.*]] = vector.broadcast %[[G]] : i32 to vector<i32>
// CHECK: return %[[F]], %[[H]] : vector<2xi32>, vector<i32>
func @scan1d_exc(%arg0 : vector<2xi32>, %arg1 : vector<i32>) -> (vector<2xi32>, vector<i32>) {
func.func @scan1d_exc(%arg0 : vector<2xi32>, %arg1 : vector<i32>) -> (vector<2xi32>, vector<i32>) {
%0:2 = vector.scan <add>, %arg0, %arg1 {inclusive = false, reduction_dim = 0} :
vector<2xi32>, vector<i32>
return %0#0, %0#1 : vector<2xi32>, vector<i32>
@ -47,7 +47,7 @@ func @scan1d_exc(%arg0 : vector<2xi32>, %arg1 : vector<i32>) -> (vector<2xi32>,
// CHECK: %[[F:.*]] = vector.insert_strided_slice %[[E]], %[[C]] {offsets = [1, 0], strides = [1, 1]} : vector<1x3xi32> into vector<2x3xi32>
// CHECK: %[[G:.*]] = vector.shape_cast %[[E]] : vector<1x3xi32> to vector<3xi32>
// CHECK: return %[[F]], %[[G]] : vector<2x3xi32>, vector<3xi32>
func @scan2d_mul_dim0(%arg0 : vector<2x3xi32>, %arg1 : vector<3xi32>) -> (vector<2x3xi32>, vector<3xi32>) {
func.func @scan2d_mul_dim0(%arg0 : vector<2x3xi32>, %arg1 : vector<3xi32>) -> (vector<2x3xi32>, vector<3xi32>) {
%0:2 = vector.scan <mul>, %arg0, %arg1 {inclusive = true, reduction_dim = 0} :
vector<2x3xi32>, vector<3xi32>
return %0#0, %0#1 : vector<2x3xi32>, vector<3xi32>
@ -67,7 +67,7 @@ func @scan2d_mul_dim0(%arg0 : vector<2x3xi32>, %arg1 : vector<3xi32>) -> (vector
// CHECK: %[[I:.*]] = vector.insert_strided_slice %[[H]], %[[F]] {offsets = [0, 2], strides = [1, 1]} : vector<2x1xi32> into vector<2x3xi32>
// CHECK: %[[J:.*]] = vector.shape_cast %[[H]] : vector<2x1xi32> to vector<2xi32>
// CHECK: return %[[I]], %[[J]] : vector<2x3xi32>, vector<2xi32>
func @scan2d_mul_dim1(%arg0 : vector<2x3xi32>, %arg1 : vector<2xi32>) -> (vector<2x3xi32>, vector<2xi32>) {
func.func @scan2d_mul_dim1(%arg0 : vector<2x3xi32>, %arg1 : vector<2xi32>) -> (vector<2x3xi32>, vector<2xi32>) {
%0:2 = vector.scan <mul>, %arg0, %arg1 {inclusive = true, reduction_dim = 1} :
vector<2x3xi32>, vector<2xi32>
return %0#0, %0#1 : vector<2x3xi32>, vector<2xi32>
@ -84,7 +84,7 @@ func @scan2d_mul_dim1(%arg0 : vector<2x3xi32>, %arg1 : vector<2xi32>) -> (vector
// CHECK: %[[F:.*]] = vector.insert_strided_slice %[[E]], %[[D]] {offsets = [0, 1, 0], strides = [1, 1, 1]} : vector<4x1x3xf32> into vector<4x2x3xf32>
// CHECK: %[[G:.*]] = vector.shape_cast %[[E]] : vector<4x1x3xf32> to vector<4x3xf32>
// CHECK: return %[[F]], %[[G]] : vector<4x2x3xf32>, vector<4x3xf32>
func @scan3d_mul_dim1(%arg0 : vector<4x2x3xf32>, %arg1 : vector<4x3xf32>) -> (vector<4x2x3xf32>, vector<4x3xf32>) {
func.func @scan3d_mul_dim1(%arg0 : vector<4x2x3xf32>, %arg1 : vector<4x3xf32>) -> (vector<4x2x3xf32>, vector<4x3xf32>) {
%0:2 = vector.scan <mul>, %arg0, %arg1 {inclusive = false, reduction_dim = 1} :
vector<4x2x3xf32>, vector<4x3xf32>
return %0#0, %0#1 : vector<4x2x3xf32>, vector<4x3xf32>

6
mlir/test/Dialect/Vector/vector-transfer-collapse-inner-most-dims.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt %s -test-vector-transfer-collapse-inner-most-dims -split-input-file | FileCheck %s
#map1 = affine_map<(d0, d1, d2, d3)[s0] -> (d0 * 3072 + s0 + d1 * 8 + d2 + d3)>
func @contiguous_inner_most_view(%in: memref<1x1x8x1xf32, #map1>) -> vector<1x8x1xf32>{
func.func @contiguous_inner_most_view(%in: memref<1x1x8x1xf32, #map1>) -> vector<1x8x1xf32>{
%c0 = arith.constant 0 : index
%cst = arith.constant 0.0 : f32
%0 = vector.transfer_read %in[%c0, %c0, %c0, %c0], %cst {in_bounds = [true, true, true]} : memref<1x1x8x1xf32, #map1>, vector<1x8x1xf32>
@ -19,7 +19,7 @@ func @contiguous_inner_most_view(%in: memref<1x1x8x1xf32, #map1>) -> vector<1x8x
// -----
func @contiguous_inner_most_dim(%A: memref<16x1xf32>, %i:index, %j:index) -> (vector<8x1xf32>) {
func.func @contiguous_inner_most_dim(%A: memref<16x1xf32>, %i:index, %j:index) -> (vector<8x1xf32>) {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
%1 = vector.transfer_read %A[%i, %j], %f0 : memref<16x1xf32>, vector<8x1xf32>
@ -34,7 +34,7 @@ func @contiguous_inner_most_dim(%A: memref<16x1xf32>, %i:index, %j:index) -> (ve
// -----
func @contiguous_inner_most_dim_bounds(%A: memref<1000x1xf32>, %i:index, %ii:index) -> (vector<4x1xf32>) {
func.func @contiguous_inner_most_dim_bounds(%A: memref<1000x1xf32>, %i:index, %ii:index) -> (vector<4x1xf32>) {
%c0 = arith.constant 0 : index
%cst = arith.constant 0.0 : f32
%0 = memref.subview %A[%i, 0] [40, 1] [1, 1] : memref<1000x1xf32> to memref<40x1xf32, affine_map<(d0, d1)[s0] -> (d0 + s0 + d1)>>

4
mlir/test/Dialect/Vector/vector-transfer-drop-unit-dims-patterns.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt %s -test-vector-transfer-drop-unit-dims-patterns -split-input-file | FileCheck %s
func @transfer_read_rank_reducing(
func.func @transfer_read_rank_reducing(
%arg : memref<1x1x3x2xi8, offset:?, strides:[6, 6, 2, 1]>) -> vector<3x2xi8> {
%c0 = arith.constant 0 : index
%cst = arith.constant 0 : i8
@ -17,7 +17,7 @@ func @transfer_read_rank_reducing(
// -----
func @transfer_write_rank_reducing(%arg : memref<1x1x3x2xi8, offset:?, strides:[6, 6, 2, 1]>, %vec : vector<3x2xi8>) {
func.func @transfer_write_rank_reducing(%arg : memref<1x1x3x2xi8, offset:?, strides:[6, 6, 2, 1]>, %vec : vector<3x2xi8>) {
%c0 = arith.constant 0 : index
vector.transfer_write %vec, %arg [%c0, %c0, %c0, %c0] :
vector<3x2xi8>, memref<1x1x3x2xi8, offset:?, strides:[6, 6, 2, 1]>

8
mlir/test/Dialect/Vector/vector-transfer-flatten.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt %s -test-vector-transfer-flatten-patterns -split-input-file | FileCheck %s
func @transfer_read_flattenable_with_offset(
func.func @transfer_read_flattenable_with_offset(
%arg : memref<5x4x3x2xi8, offset:?, strides:[24, 6, 2, 1]>) -> vector<5x4x3x2xi8> {
%c0 = arith.constant 0 : index
%cst = arith.constant 0 : i8
@ -18,7 +18,7 @@ func @transfer_read_flattenable_with_offset(
// -----
func @transfer_write_flattenable_with_offset(
func.func @transfer_write_flattenable_with_offset(
%arg : memref<5x4x3x2xi8, offset:?, strides:[24, 6, 2, 1]>, %vec : vector<5x4x3x2xi8>) {
%c0 = arith.constant 0 : index
vector.transfer_write %vec, %arg [%c0, %c0, %c0, %c0] :
@ -35,7 +35,7 @@ func @transfer_write_flattenable_with_offset(
// -----
func @transfer_write_0d(%arg : memref<i8>, %vec : vector<i8>) {
func.func @transfer_write_0d(%arg : memref<i8>, %vec : vector<i8>) {
vector.transfer_write %vec, %arg[] : vector<i8>, memref<i8>
return
}
@ -48,7 +48,7 @@ func @transfer_write_0d(%arg : memref<i8>, %vec : vector<i8>) {
// -----
func @transfer_read_0d(%arg : memref<i8>) -> vector<i8> {
func.func @transfer_read_0d(%arg : memref<i8>) -> vector<i8> {
%cst = arith.constant 0 : i8
%0 = vector.transfer_read %arg[], %cst : memref<i8>, vector<i8>
return %0 : vector<i8>

12
mlir/test/Dialect/Vector/vector-transfer-full-partial-split.mlir
View File

@ -21,7 +21,7 @@
// LINALG-SAME: %[[A:[a-zA-Z0-9]*]]: memref
// LINALG-SAME: %[[i:[a-zA-Z0-9]*]]: index
// LINALG-SAME: %[[j:[a-zA-Z0-9]*]]: index
func @split_vector_transfer_read_2d(%A: memref<?x8xf32>, %i: index, %j: index) -> vector<4x8xf32> {
func.func @split_vector_transfer_read_2d(%A: memref<?x8xf32>, %i: index, %j: index) -> vector<4x8xf32> {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
@ -105,7 +105,7 @@ func @split_vector_transfer_read_2d(%A: memref<?x8xf32>, %i: index, %j: index) -
// LINALG-SAME: %[[A:[a-zA-Z0-9]*]]: memref
// LINALG-SAME: %[[i:[a-zA-Z0-9]*]]: index
// LINALG-SAME: %[[j:[a-zA-Z0-9]*]]: index
func @split_vector_transfer_read_strided_2d(
func.func @split_vector_transfer_read_strided_2d(
%A: memref<7x8xf32, offset:?, strides:[?, 1]>,
%i: index, %j: index) -> vector<4x8xf32> {
%c0 = arith.constant 0 : index
@ -191,7 +191,7 @@ func @split_vector_transfer_read_strided_2d(
// -----
func @split_vector_transfer_write_2d(%V: vector<4x8xf32>, %A: memref<?x8xf32>, %i: index, %j: index) {
func.func @split_vector_transfer_write_2d(%V: vector<4x8xf32>, %A: memref<?x8xf32>, %i: index, %j: index) {
vector.transfer_write %V, %A[%i, %j] :
vector<4x8xf32>, memref<?x8xf32>
return
@ -287,7 +287,7 @@ func @split_vector_transfer_write_2d(%V: vector<4x8xf32>, %A: memref<?x8xf32>, %
// -----
func @split_vector_transfer_write_strided_2d(
func.func @split_vector_transfer_write_strided_2d(
%V: vector<4x8xf32>, %A: memref<7x8xf32, offset:?, strides:[?, 1]>,
%i: index, %j: index) {
vector.transfer_write %V, %A[%i, %j] :
@ -396,10 +396,10 @@ func @split_vector_transfer_write_strided_2d(
// -----
func private @fake_side_effecting_fun(%0: vector<2x2xf32>) -> ()
func.func private @fake_side_effecting_fun(%0: vector<2x2xf32>) -> ()
// CHECK-LABEL: transfer_read_within_async_execute
func @transfer_read_within_async_execute(%A : memref<?x?xf32>) -> !async.token {
func.func @transfer_read_within_async_execute(%A : memref<?x?xf32>) -> !async.token {
%c0 = arith.constant 0 : index
%f0 = arith.constant 0.0 : f32
// CHECK-NOT: alloca

32
mlir/test/Dialect/Vector/vector-transfer-to-vector-load-store.mlir
View File

@ -3,7 +3,7 @@
// CHECK-LABEL: func @vector_transfer_ops_0d_memref(
// CHECK-SAME: %[[MEM:.*]]: memref<f32>
// CHECK-SAME: %[[VV:.*]]: vector<1x1x1xf32>
func @vector_transfer_ops_0d_memref(%M: memref<f32>, %v: vector<1x1x1xf32>) {
func.func @vector_transfer_ops_0d_memref(%M: memref<f32>, %v: vector<1x1x1xf32>) {
%f0 = arith.constant 0.0 : f32
// CHECK-NEXT: %[[s:.*]] = memref.load %[[MEM]][] : memref<f32>
@ -25,7 +25,7 @@ func @vector_transfer_ops_0d_memref(%M: memref<f32>, %v: vector<1x1x1xf32>) {
// CHECK-LABEL: func @vector_transfer_ops_0d_tensor(
// CHECK-SAME: %[[SOURCE:.*]]: tensor<f32>
func @vector_transfer_ops_0d_tensor(%M: tensor<f32>) -> vector<1xf32> {
func.func @vector_transfer_ops_0d_tensor(%M: tensor<f32>) -> vector<1xf32> {
%f0 = arith.constant 0.0 : f32
// CHECK-NEXT: %[[S:.*]] = tensor.extract %[[SOURCE]][] : tensor<f32>
@ -48,7 +48,7 @@ func @vector_transfer_ops_0d_tensor(%M: tensor<f32>) -> vector<1xf32> {
// CHECK-NEXT: return %[[RES]] : vector<4xf32>
// CHECK-NEXT: }
func @transfer_to_load(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
func.func @transfer_to_load(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true]} : memref<8x8xf32>, vector<4xf32>
vector.transfer_write %res, %mem[%i, %i] {in_bounds = [true]} : vector<4xf32>, memref<8x8xf32>
@ -66,7 +66,7 @@ func @transfer_to_load(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
// CHECK-NEXT: return %[[RES]] : vector<2x4xf32>
// CHECK-NEXT: }
func @transfer_2D(%mem : memref<8x8xf32>, %i : index) -> vector<2x4xf32> {
func.func @transfer_2D(%mem : memref<8x8xf32>, %i : index) -> vector<2x4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true, true]} : memref<8x8xf32>, vector<2x4xf32>
vector.transfer_write %res, %mem[%i, %i] {in_bounds = [true, true]} : vector<2x4xf32>, memref<8x8xf32>
@ -84,7 +84,7 @@ func @transfer_2D(%mem : memref<8x8xf32>, %i : index) -> vector<2x4xf32> {
// CHECK-NEXT: return %[[RES]] : vector<2x4xf32>
// CHECK-NEXT: }
func @transfer_vector_element(%mem : memref<8x8xvector<2x4xf32>>, %i : index) -> vector<2x4xf32> {
func.func @transfer_vector_element(%mem : memref<8x8xvector<2x4xf32>>, %i : index) -> vector<2x4xf32> {
%cf0 = arith.constant dense<0.0> : vector<2x4xf32>
%res = vector.transfer_read %mem[%i, %i], %cf0 : memref<8x8xvector<2x4xf32>>, vector<2x4xf32>
vector.transfer_write %res, %mem[%i, %i] : vector<2x4xf32>, memref<8x8xvector<2x4xf32>>
@ -104,7 +104,7 @@ func @transfer_vector_element(%mem : memref<8x8xvector<2x4xf32>>, %i : index) ->
// CHECK-NEXT: return %[[RES]] : vector<1x2x4xf32>
// CHECK-NEXT: }
func @transfer_vector_element_different_types(%mem : memref<8x8xvector<2x4xf32>>, %i : index) -> vector<1x2x4xf32> {
func.func @transfer_vector_element_different_types(%mem : memref<8x8xvector<2x4xf32>>, %i : index) -> vector<1x2x4xf32> {
%cf0 = arith.constant dense<0.0> : vector<2x4xf32>
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true]} : memref<8x8xvector<2x4xf32>>, vector<1x2x4xf32>
vector.transfer_write %res, %mem[%i, %i] {in_bounds = [true]} : vector<1x2x4xf32>, memref<8x8xvector<2x4xf32>>
@ -124,7 +124,7 @@ func @transfer_vector_element_different_types(%mem : memref<8x8xvector<2x4xf32>>
// CHECK-NEXT: return %[[RES]] : vector<2x4xf32>
// CHECK-NEXT: }
func @transfer_2D_not_inbounds(%mem : memref<8x8xf32>, %i : index) -> vector<2x4xf32> {
func.func @transfer_2D_not_inbounds(%mem : memref<8x8xf32>, %i : index) -> vector<2x4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true, false]} : memref<8x8xf32>, vector<2x4xf32>
vector.transfer_write %res, %mem[%i, %i] {in_bounds = [false, true]} : vector<2x4xf32>, memref<8x8xf32>
@ -144,7 +144,7 @@ func @transfer_2D_not_inbounds(%mem : memref<8x8xf32>, %i : index) -> vector<2x4
// CHECK-NEXT: return %[[RES]] : vector<4xf32>
// CHECK-NEXT: }
func @transfer_not_inbounds(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
func.func @transfer_not_inbounds(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 : memref<8x8xf32>, vector<4xf32>
vector.transfer_write %res, %mem[%i, %i] : vector<4xf32>, memref<8x8xf32>
@ -162,7 +162,7 @@ func @transfer_not_inbounds(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32>
// CHECK-NEXT: }
#layout = affine_map<(d0, d1) -> (d0*16 + d1)>
func @transfer_nondefault_layout(%mem : memref<8x8xf32, #layout>, %i : index) -> vector<4xf32> {
func.func @transfer_nondefault_layout(%mem : memref<8x8xf32, #layout>, %i : index) -> vector<4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true]} : memref<8x8xf32, #layout>, vector<4xf32>
vector.transfer_write %res, %mem[%i, %i] {in_bounds = [true]} : vector<4xf32>, memref<8x8xf32, #layout>
@ -182,7 +182,7 @@ func @transfer_nondefault_layout(%mem : memref<8x8xf32, #layout>, %i : index) ->
// CHECK-NEXT: return %[[RES]] : vector<4xf32>
// CHECK-NEXT: }
func @transfer_perm_map(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
func.func @transfer_perm_map(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true], permutation_map = affine_map<(d0, d1) -> (d0)>} : memref<8x8xf32>, vector<4xf32>
vector.transfer_write %res, %mem[%i, %i] {in_bounds = [true], permutation_map = affine_map<(d0, d1) -> (d0)>} : vector<4xf32>, memref<8x8xf32>
@ -202,7 +202,7 @@ func @transfer_perm_map(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
// CHECK-NEXT: }
#broadcast = affine_map<(d0, d1) -> (0)>
func @transfer_broadcasting(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
func.func @transfer_broadcasting(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true], permutation_map = #broadcast} : memref<8x8xf32>, vector<4xf32>
return %res : vector<4xf32>
@ -217,7 +217,7 @@ func @transfer_broadcasting(%mem : memref<8x8xf32>, %i : index) -> vector<4xf32>
// CHECK-NEXT: %[[RES:.*]] = vector.broadcast %[[LOAD]] : f32 to vector<1xf32>
// CHECK-NEXT: return %[[RES]] : vector<1xf32>
// CHECK-NEXT: }
func @transfer_scalar(%mem : memref<?x?xf32>, %i : index) -> vector<1xf32> {
func.func @transfer_scalar(%mem : memref<?x?xf32>, %i : index) -> vector<1xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true]} : memref<?x?xf32>, vector<1xf32>
return %res : vector<1xf32>
@ -235,7 +235,7 @@ func @transfer_scalar(%mem : memref<?x?xf32>, %i : index) -> vector<1xf32> {
// CHECK-NEXT: }
#broadcast = affine_map<(d0, d1) -> (0, 0)>
func @transfer_broadcasting_2D(%mem : memref<8x8xf32>, %i : index) -> vector<4x4xf32> {
func.func @transfer_broadcasting_2D(%mem : memref<8x8xf32>, %i : index) -> vector<4x4xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i], %cf0 {in_bounds = [true, true], permutation_map = #broadcast} : memref<8x8xf32>, vector<4x4xf32>
return %res : vector<4x4xf32>
@ -253,7 +253,7 @@ func @transfer_broadcasting_2D(%mem : memref<8x8xf32>, %i : index) -> vector<4x4
// CHECK-NEXT: }
#broadcast = affine_map<(d0, d1, d2, d3, d4) -> (d1, 0, 0, d4)>
func @transfer_broadcasting_complex(%mem : memref<10x20x30x8x8xf32>, %i : index) -> vector<3x2x4x5xf32> {
func.func @transfer_broadcasting_complex(%mem : memref<10x20x30x8x8xf32>, %i : index) -> vector<3x2x4x5xf32> {
%cf0 = arith.constant 0.0 : f32
%res = vector.transfer_read %mem[%i, %i, %i, %i, %i], %cf0 {in_bounds = [true, true, true, true], permutation_map = #broadcast} : memref<10x20x30x8x8xf32>, vector<3x2x4x5xf32>
return %res : vector<3x2x4x5xf32>
@ -273,7 +273,7 @@ func @transfer_broadcasting_complex(%mem : memref<10x20x30x8x8xf32>, %i : index)
// CHECK-DAG: #[[$MAP1:.*]] = affine_map<(d0, d1, d2, d3) -> (d1, 0, d3)>
// CHECK-LABEL: func @transfer_read_permutations
func @transfer_read_permutations(%arg0 : memref<?x?xf32>, %arg1 : memref<?x?x?x?xf32>, %m: i1)
func.func @transfer_read_permutations(%arg0 : memref<?x?xf32>, %arg1 : memref<?x?x?x?xf32>, %m: i1)
-> (vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<7x14x8x16xf32>,
vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<7x14x8x16xf32>, vector<8xf32>) {
// CHECK-DAG: %[[CF0:.*]] = arith.constant 0.000000e+00 : f32
@ -331,7 +331,7 @@ func @transfer_read_permutations(%arg0 : memref<?x?xf32>, %arg1 : memref<?x?x?x?
// CHECK-SAME: %[[ARG2:.*]]: vector<7x14x8x16xf32>
// CHECK-SAME: %[[ARG3:.*]]: vector<8x16xf32>
// CHECK-SAME: %[[M:.*]]: i1
func @transfer_write_permutations(
func.func @transfer_write_permutations(
%arg0 : memref<?x?x?x?xf32>, %arg1 : tensor<?x?x?x?xf32>,
%v1 : vector<7x14x8x16xf32>, %v2 : vector<8x16xf32>, %m: i1) -> tensor<?x?x?x?xf32> {
// CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index

20
mlir/test/Dialect/Vector/vector-transfer-unroll.mlir
View File

@ -13,7 +13,7 @@
// CHECK-NEXT: %[[VEC3:.*]] = vector.insert_strided_slice %[[VTR3]], %[[VEC2]] {offsets = [2, 2], strides = [1, 1]} : vector<2x2xf32> into vector<4x4xf32>
// CHECK-NEXT: return %[[VEC3]] : vector<4x4xf32>
func @transfer_read_unroll(%arg0 : memref<4x4xf32>) -> vector<4x4xf32> {
func.func @transfer_read_unroll(%arg0 : memref<4x4xf32>) -> vector<4x4xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 : memref<4x4xf32>, vector<4x4xf32>
@ -33,7 +33,7 @@ func @transfer_read_unroll(%arg0 : memref<4x4xf32>) -> vector<4x4xf32> {
// CHECK-NEXT: vector.transfer_write %[[S3]], {{.*}}[%[[C2]], %[[C2]]] {{.*}} : vector<2x2xf32>, memref<4x4xf32>
// CHECK-NEXT: return
func @transfer_write_unroll(%arg0 : memref<4x4xf32>, %arg1 : vector<4x4xf32>) {
func.func @transfer_write_unroll(%arg0 : memref<4x4xf32>, %arg1 : vector<4x4xf32>) {
%c0 = arith.constant 0 : index
vector.transfer_write %arg1, %arg0[%c0, %c0] : vector<4x4xf32>, memref<4x4xf32>
return
@ -52,7 +52,7 @@ func @transfer_write_unroll(%arg0 : memref<4x4xf32>, %arg1 : vector<4x4xf32>) {
// CHECK-NEXT: vector.transfer_write %[[VTR3]], {{.*}}[%[[C2]], %[[C2]]] {{.*}} : vector<2x2xf32>, memref<4x4xf32>
// CHECK-NEXT: return
func @transfer_readwrite_unroll(%arg0 : memref<4x4xf32>) {
func.func @transfer_readwrite_unroll(%arg0 : memref<4x4xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 : memref<4x4xf32>, vector<4x4xf32>
@ -73,7 +73,7 @@ func @transfer_readwrite_unroll(%arg0 : memref<4x4xf32>) {
// CHECK-NEXT: %[[VEC3:.*]] = vector.insert_strided_slice %[[VTR3]], %[[VEC2]] {offsets = [2, 2], strides = [1, 1]} : vector<2x2xf32> into vector<4x4xf32>
// CHECK-NEXT: return %[[VEC3]] : vector<4x4xf32>
func @transfer_read_unroll_tensor(%arg0 : tensor<4x4xf32>) -> vector<4x4xf32> {
func.func @transfer_read_unroll_tensor(%arg0 : tensor<4x4xf32>) -> vector<4x4xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 : tensor<4x4xf32>, vector<4x4xf32>
@ -93,7 +93,7 @@ func @transfer_read_unroll_tensor(%arg0 : tensor<4x4xf32>) -> vector<4x4xf32> {
// CHECK-NEXT: %[[VTW3:.*]] = vector.transfer_write %[[S3]], %[[VTW2]][%[[C2]], %[[C2]]] {{.*}} : vector<2x2xf32>, tensor<4x4xf32>
// CHECK-NEXT: return %[[VTW3]] : tensor<4x4xf32>
func @transfer_write_unroll_tensor(%arg0 : tensor<4x4xf32>,
func.func @transfer_write_unroll_tensor(%arg0 : tensor<4x4xf32>,
%arg1 : vector<4x4xf32>) -> tensor<4x4xf32> {
%c0 = arith.constant 0 : index
%r = vector.transfer_write %arg1, %arg0[%c0, %c0] :
@ -114,7 +114,7 @@ func @transfer_write_unroll_tensor(%arg0 : tensor<4x4xf32>,
// CHECK-NEXT: %[[VTW3:.*]] = vector.transfer_write %[[VTR3]], %[[VTW2]][%[[C2]], %[[C2]]] {{.*}} : vector<2x2xf32>, tensor<4x4xf32>
// CHECK-NEXT: return %[[VTW3]] : tensor<4x4xf32>
func @transfer_readwrite_unroll_tensor(%arg0 : tensor<4x4xf32>, %arg1 : tensor<4x4xf32>) ->
func.func @transfer_readwrite_unroll_tensor(%arg0 : tensor<4x4xf32>, %arg1 : tensor<4x4xf32>) ->
tensor<4x4xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
@ -143,7 +143,7 @@ func @transfer_readwrite_unroll_tensor(%arg0 : tensor<4x4xf32>, %arg1 : tensor<4
// CHECK-NEXT: %[[VEC5:.*]] = vector.insert_strided_slice %[[VTR5]], %[[VEC4]] {offsets = [2, 4], strides = [1, 1]} : vector<2x2xf32> into vector<4x6xf32>
// CHECK-NEXT: return %[[VEC5]] : vector<4x6xf32>
#map0 = affine_map<(d0, d1) -> (d1, d0)>
func @transfer_read_unroll_permutation(%arg0 : memref<6x4xf32>) -> vector<4x6xf32> {
func.func @transfer_read_unroll_permutation(%arg0 : memref<6x4xf32>) -> vector<4x6xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 {permutation_map = #map0} : memref<6x4xf32>, vector<4x6xf32>
@ -169,7 +169,7 @@ func @transfer_read_unroll_permutation(%arg0 : memref<6x4xf32>) -> vector<4x6xf3
// CHECK-NEXT: %[[VEC5:.*]] = vector.insert_strided_slice %[[VTR5]], %[[VEC4]] {offsets = [4, 2], strides = [1, 1]} : vector<2x2xf32> into vector<6x4xf32>
// CHECK-NEXT: return %[[VEC5]] : vector<6x4xf32>
#map0 = affine_map<(d0, d1) -> (0, d1)>
func @transfer_read_unroll_broadcast(%arg0 : memref<6x4xf32>) -> vector<6x4xf32> {
func.func @transfer_read_unroll_broadcast(%arg0 : memref<6x4xf32>) -> vector<6x4xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 {permutation_map = #map0} : memref<6x4xf32>, vector<6x4xf32>
@ -196,7 +196,7 @@ func @transfer_read_unroll_broadcast(%arg0 : memref<6x4xf32>) -> vector<6x4xf32>
// CHECK-NEXT: %[[VEC5:.*]] = vector.insert_strided_slice %[[VTR5]], %[[VEC4]] {offsets = [2, 4], strides = [1, 1]} : vector<2x2xf32> into vector<4x6xf32>
// CHECK-NEXT: return %[[VEC5]] : vector<4x6xf32>
#map0 = affine_map<(d0, d1) -> (0, d0)>
func @transfer_read_unroll_broadcast_permuation(%arg0 : memref<6x4xf32>) -> vector<4x6xf32> {
func.func @transfer_read_unroll_broadcast_permuation(%arg0 : memref<6x4xf32>) -> vector<4x6xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 {permutation_map = #map0} : memref<6x4xf32>, vector<4x6xf32>
@ -223,7 +223,7 @@ func @transfer_read_unroll_broadcast_permuation(%arg0 : memref<6x4xf32>) -> vect
// CHECK-NEXT: %[[VEC5:.*]] = vector.insert_strided_slice %[[VTR5]], %[[VEC4]] {offsets = [4, 2], strides = [1, 1]} : vector<2x2xf32> into vector<6x4xf32>
// CHECK-NEXT: return %[[VEC5]] : vector<6x4xf32>
#map0 = affine_map<(d0, d1, d2) -> (d2, d0)>
func @transfer_read_unroll_different_rank(%arg0 : memref<?x?x?xf32>) -> vector<6x4xf32> {
func.func @transfer_read_unroll_different_rank(%arg0 : memref<?x?x?xf32>) -> vector<6x4xf32> {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0, %c0], %cf0 {permutation_map = #map0} : memref<?x?x?xf32>, vector<6x4xf32>

12
mlir/test/Dialect/Vector/vector-transferop-opt.mlir
View File

@ -7,7 +7,7 @@
// CHECK: }
// CHECK: vector.transfer_write
// CHECK: return
func @forward_dead_store(%arg0: i1, %arg1 : memref<4x4xf32>,
func.func @forward_dead_store(%arg0: i1, %arg1 : memref<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) {
%c1 = arith.constant 1 : index
%c4 = arith.constant 4 : index
@ -35,7 +35,7 @@ func @forward_dead_store(%arg0: i1, %arg1 : memref<4x4xf32>,
// CHECK: }
// CHECK: vector.transfer_write
// CHECK: return
func @forward_nested(%arg0: i1, %arg1 : memref<4x4xf32>, %v0 : vector<1x4xf32>,
func.func @forward_nested(%arg0: i1, %arg1 : memref<4x4xf32>, %v0 : vector<1x4xf32>,
%v1 : vector<1x4xf32>, %i : index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -68,7 +68,7 @@ func @forward_nested(%arg0: i1, %arg1 : memref<4x4xf32>, %v0 : vector<1x4xf32>,
// CHECK: }
// CHECK: vector.transfer_write
// CHECK: return
func @forward_nested_negative(%arg0: i1, %arg1 : memref<4x4xf32>,
func.func @forward_nested_negative(%arg0: i1, %arg1 : memref<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -102,7 +102,7 @@ func @forward_nested_negative(%arg0: i1, %arg1 : memref<4x4xf32>,
// CHECK-NOT: vector.transfer_write
// CHECK: vector.transfer_read
// CHECK: return
func @dead_store_region(%arg0: i1, %arg1 : memref<4x4xf32>,
func.func @dead_store_region(%arg0: i1, %arg1 : memref<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index)
-> (vector<1x4xf32>) {
%c0 = arith.constant 0 : index
@ -138,7 +138,7 @@ func @dead_store_region(%arg0: i1, %arg1 : memref<4x4xf32>,
// CHECK: }
// CHECK: vector.transfer_write
// CHECK: return
func @dead_store_negative(%arg0: i1, %arg1 : memref<4x4xf32>,
func.func @dead_store_negative(%arg0: i1, %arg1 : memref<4x4xf32>,
%v0 :vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index
@ -166,7 +166,7 @@ func @dead_store_negative(%arg0: i1, %arg1 : memref<4x4xf32>,
// CHECK: vector.transfer_write
// CHECK: }
// CHECK: return
func @dead_store_nested_region(%arg0: i1, %arg1: i1, %arg2 : memref<4x4xf32>,
func.func @dead_store_nested_region(%arg0: i1, %arg1: i1, %arg2 : memref<4x4xf32>,
%v0 : vector<1x4xf32>, %v1 : vector<1x4xf32>, %i : index) {
%c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index

34
mlir/test/Dialect/Vector/vector-transforms.mlir
View File

@ -13,7 +13,7 @@
// CHECK-NEXT: %[[VEC1:.*]] = vector.insert_strided_slice %[[A2]], %[[VEC0]] {offsets = [2, 0], strides = [1, 1]} : vector<2x2xf32> into vector<4x2xf32>
// CHECK-NEXT: return %[[VEC1:.*]] : vector<4x2xf32>
func @add4x2(%0: vector<4x2xf32>) -> vector<4x2xf32> {
func.func @add4x2(%0: vector<4x2xf32>) -> vector<4x2xf32> {
%1 = arith.addf %0, %0: vector<4x2xf32>
return %1: vector<4x2xf32>
}
@ -56,7 +56,7 @@ func @add4x2(%0: vector<4x2xf32>) -> vector<4x2xf32> {
// CHECK-NEXT: return %[[R4]] : vector<4x4xf32>
func @add4x4(%0: vector<4x4xf32>, %1: vector<4x4xf32>) -> vector<4x4xf32> {
func.func @add4x4(%0: vector<4x4xf32>, %1: vector<4x4xf32>) -> vector<4x4xf32> {
%2 = arith.addf %0, %1: vector<4x4xf32>
%3 = arith.addf %1, %2: vector<4x4xf32>
return %3: vector<4x4xf32>
@ -166,7 +166,7 @@ func @add4x4(%0: vector<4x4xf32>, %1: vector<4x4xf32>) -> vector<4x4xf32> {
// CHECK-NEXT: return %[[VEC4]] : vector<4x4xf32>
func @contraction4x4_ijk(%arg0 : vector<4x6xf32>, %arg1 : vector<6x4xf32>,
func.func @contraction4x4_ijk(%arg0 : vector<4x6xf32>, %arg1 : vector<6x4xf32>,
%arg2 : vector<4x4xf32>, %arg3 : index)
-> (vector<4x4xf32>) {
%lhsm = vector.constant_mask [4, 6] : vector<4x6xi1>
@ -231,7 +231,7 @@ func @contraction4x4_ijk(%arg0 : vector<4x6xf32>, %arg1 : vector<6x4xf32>,
// CHECK-NEXT: %[[VEC3:.*]] = vector.insert_strided_slice %[[R1S22]], %[[VEC2]] {offsets = [2, 2], strides = [1, 1]} : vector<2x2xf32> into vector<4x4xf32>
// CHECK-NEXT: return %[[VEC3]] : vector<4x4xf32>
func @contraction4x4_ikj(%arg0 : vector<4x2xf32>, %arg1 : vector<2x4xf32>,
func.func @contraction4x4_ikj(%arg0 : vector<4x2xf32>, %arg1 : vector<2x4xf32>,
%arg2 : vector<4x4xf32>, %arg3 : index)
-> (vector<4x4xf32>) {
%lhsm = vector.constant_mask [4, 2] : vector<4x2xi1>
@ -271,7 +271,7 @@ func @contraction4x4_ikj(%arg0 : vector<4x2xf32>, %arg1 : vector<2x4xf32>,
// CHECK-NEXT: vector.transfer_write %[[R3]], %{{.*}}[%[[C2]], %[[C2]]] {in_bounds = [true, true]} : vector<2x2xf32>, memref<4x4xf32>
// CHECK-NEXT: return
func @contraction4x4_ikj_xfer_read(%arg0 : memref<4x2xf32>,
func.func @contraction4x4_ikj_xfer_read(%arg0 : memref<4x2xf32>,
%arg1 : memref<2x4xf32>,
%arg2 : memref<4x4xf32>) {
%c0 = arith.constant 0 : index
@ -304,7 +304,7 @@ func @contraction4x4_ikj_xfer_read(%arg0 : memref<4x2xf32>,
// CHECK-COUNT-4: arith.addf
// CHECK-COUNT-4: vector.transfer_write
func @vector_transfers(%arg0: index, %arg1: index) {
func.func @vector_transfers(%arg0: index, %arg1: index) {
%cst = arith.constant 0.000000e+00 : f32
%0 = memref.alloc(%arg0, %arg1) : memref<?x?xf32>
%1 = memref.alloc(%arg0, %arg1) : memref<?x?xf32>
@ -325,7 +325,7 @@ func @vector_transfers(%arg0: index, %arg1: index) {
// CHECK-LABEL: func @cancelling_shape_cast_ops
// CHECK-SAME: %[[A0:.*0]]: vector<2x4xf32>
// CHECK: return %[[A0]] : vector<2x4xf32>
func @cancelling_shape_cast_ops(%arg0 : vector<2x4xf32>) -> vector<2x4xf32> {
func.func @cancelling_shape_cast_ops(%arg0 : vector<2x4xf32>) -> vector<2x4xf32> {
%0 = vector.shape_cast %arg0 : vector<2x4xf32> to vector<8xf32>
%1 = vector.shape_cast %0 : vector<8xf32> to vector<2x4xf32>
return %1 : vector<2x4xf32>
@ -363,7 +363,7 @@ func @cancelling_shape_cast_ops(%arg0 : vector<2x4xf32>) -> vector<2x4xf32> {
// CHECK: vector.transfer_write %[[SEL1]], %[[ARG0]][%[[C0]], %[[C2]]] {{.*}} : vector<2x2xf32>, memref<4x4xf32>
// CHECK: vector.transfer_write %[[SEL2]], %[[ARG0]][%[[C2]], %[[C0]]] {{.*}} : vector<2x2xf32>, memref<4x4xf32>
// CHECK: vector.transfer_write %[[SEL3]], %[[ARG0]][%[[C2]], %[[C2]]] {{.*}} : vector<2x2xf32>, memref<4x4xf32>
func @elementwise_unroll(%arg0 : memref<4x4xf32>, %arg1 : memref<4x4xf32>) {
func.func @elementwise_unroll(%arg0 : memref<4x4xf32>, %arg1 : memref<4x4xf32>) {
%c0 = arith.constant 0 : index
%cf0 = arith.constant 0.0 : f32
%0 = vector.transfer_read %arg0[%c0, %c0], %cf0 : memref<4x4xf32>, vector<4x4xf32>
@ -400,7 +400,7 @@ func @elementwise_unroll(%arg0 : memref<4x4xf32>, %arg1 : memref<4x4xf32>) {
// CHECK-NEXT: %[[VTW3:.*]] = vector.transfer_write %[[R3]], %[[VTW2]][%[[C2]], %[[C2]]] {in_bounds = [true, true]} : vector<2x2xf32>, tensor<4x4xf32>
// CHECK-NEXT: return %[[VTW3]] : tensor<4x4xf32>
func @contraction4x4_ikj_xfer_read_tensor(%arg0 : tensor<4x2xf32>,
func.func @contraction4x4_ikj_xfer_read_tensor(%arg0 : tensor<4x2xf32>,
%arg1 : tensor<2x4xf32>,
%arg2 : tensor<4x4xf32>) ->
tensor<4x4xf32> {
@ -421,7 +421,7 @@ func @contraction4x4_ikj_xfer_read_tensor(%arg0 : tensor<4x2xf32>,
// CHECK-LABEL: func @bubble_down_bitcast_in_extract
// CHECK-SAME: %[[SRC:.+]]: vector<4xf32>
func @bubble_down_bitcast_in_extract(%src: vector<4xf32>) -> (f16, f16) {
func.func @bubble_down_bitcast_in_extract(%src: vector<4xf32>) -> (f16, f16) {
%0 = vector.bitcast %src : vector<4xf32> to vector<8xf16>
// CHECK: %[[EXTRACT1:.+]] = vector.extract %[[SRC]][1] : vector<4xf32>
// CHECK: %[[CAST1:.+]] = vector.bitcast %[[EXTRACT1]] : vector<1xf32> to vector<2xf16>
@ -437,7 +437,7 @@ func @bubble_down_bitcast_in_extract(%src: vector<4xf32>) -> (f16, f16) {
// CHECK-LABEL: func @bubble_down_bitcast_in_strided_slice_extract
// CHECK-SAME: %[[SRC:.+]]: vector<4xf32>
func @bubble_down_bitcast_in_strided_slice_extract(%arg0: vector<4xf32>) -> vector<4xf16> {
func.func @bubble_down_bitcast_in_strided_slice_extract(%arg0: vector<4xf32>) -> vector<4xf16> {
// CHECK: %[[EXTRACT:.+]] = vector.extract_strided_slice %[[SRC]] {offsets = [2], sizes = [2], strides = [1]} : vector<4xf32> to vector<2xf32>
// CHECK: %[[CAST:.+]] = vector.bitcast %[[EXTRACT]] : vector<2xf32> to vector<4xf16>
%cast = vector.bitcast %arg0: vector<4xf32> to vector<8xf16>
@ -448,7 +448,7 @@ func @bubble_down_bitcast_in_strided_slice_extract(%arg0: vector<4xf32>) -> vect
// CHECK-LABEL: func @bubble_down_bitcast_in_strided_slice_extract_full_last_dim
// CHECK-SAME: %[[SRC:.+]]: vector<4x2xf32>
func @bubble_down_bitcast_in_strided_slice_extract_full_last_dim(%arg0: vector<4x2xf32>) -> vector<2x4xf16> {
func.func @bubble_down_bitcast_in_strided_slice_extract_full_last_dim(%arg0: vector<4x2xf32>) -> vector<2x4xf16> {
// CHECK: %[[EXTRACT:.+]] = vector.extract_strided_slice %[[SRC]] {offsets = [1], sizes = [2], strides = [1]} : vector<4x2xf32> to vector<2x2xf32>
// CHECK: %[[CAST:.+]] = vector.bitcast %[[EXTRACT]] : vector<2x2xf32> to vector<2x4xf16>
%cast = vector.bitcast %arg0: vector<4x2xf32> to vector<4x4xf16>
@ -458,7 +458,7 @@ func @bubble_down_bitcast_in_strided_slice_extract_full_last_dim(%arg0: vector<4
}
// CHECK-LABEL: func @bubble_down_bitcast_in_strided_slice_extract_odd_offset
func @bubble_down_bitcast_in_strided_slice_extract_odd_offset(%arg0: vector<4xf32>) -> vector<4xf16> {
func.func @bubble_down_bitcast_in_strided_slice_extract_odd_offset(%arg0: vector<4xf32>) -> vector<4xf16> {
// CHECK: vector.bitcast
// CHECK-NEXT: vector.extract_strided_slice
%cast = vector.bitcast %arg0: vector<4xf32> to vector<8xf16>
@ -467,7 +467,7 @@ func @bubble_down_bitcast_in_strided_slice_extract_odd_offset(%arg0: vector<4xf3
}
// CHECK-LABEL: func @bubble_down_bitcast_in_strided_slice_extract_odd_size
func @bubble_down_bitcast_in_strided_slice_extract_odd_size(%arg0: vector<4xf32>) -> vector<3xf16> {
func.func @bubble_down_bitcast_in_strided_slice_extract_odd_size(%arg0: vector<4xf32>) -> vector<3xf16> {
// CHECK: vector.bitcast
// CHECK-NEXT: vector.extract_strided_slice
%cast = vector.bitcast %arg0: vector<4xf32> to vector<8xf16>
@ -477,7 +477,7 @@ func @bubble_down_bitcast_in_strided_slice_extract_odd_size(%arg0: vector<4xf32>
// CHECK-LABEL: func @bubble_up_bitcast_in_strided_slice_insert
// CHECK-SAME: (%[[DST:.+]]: vector<8xf16>, %[[SRC1:.+]]: vector<4xf16>, %[[SRC2:.+]]: vector<4xf16>)
func @bubble_up_bitcast_in_strided_slice_insert(%dst: vector<8xf16>, %src1: vector<4xf16>, %src2: vector<4xf16>) -> vector<4xf32> {
func.func @bubble_up_bitcast_in_strided_slice_insert(%dst: vector<8xf16>, %src1: vector<4xf16>, %src2: vector<4xf16>) -> vector<4xf32> {
// CHECK-DAG: %[[CAST_SRC1:.+]] = vector.bitcast %[[SRC1]] : vector<4xf16> to vector<2xf32>
// CHECK-DAG: %[[CAST_SRC2:.+]] = vector.bitcast %[[SRC2]] : vector<4xf16> to vector<2xf32>
// CHECK-DAG: %[[CAST_DST:.+]] = vector.bitcast %[[DST]] : vector<8xf16> to vector<4xf32>
@ -491,7 +491,7 @@ func @bubble_up_bitcast_in_strided_slice_insert(%dst: vector<8xf16>, %src1: vect
}
// CHECK-LABEL: func @bubble_up_bitcast_in_strided_slice_insert_odd_offset
func @bubble_up_bitcast_in_strided_slice_insert_odd_offset(%dst: vector<8xf16>, %src: vector<4xf16>) -> vector<4xf32> {
func.func @bubble_up_bitcast_in_strided_slice_insert_odd_offset(%dst: vector<8xf16>, %src: vector<4xf16>) -> vector<4xf32> {
// CHECK: vector.insert_strided_slice
// CHECK-NEXT: vector.bitcast
%0 = vector.insert_strided_slice %src, %dst {offsets = [3], strides = [1]} : vector<4xf16> into vector<8xf16>
@ -500,7 +500,7 @@ func @bubble_up_bitcast_in_strided_slice_insert_odd_offset(%dst: vector<8xf16>,
}
// CHECK-LABEL: func @bubble_up_bitcast_in_strided_slice_insert_different_rank
func @bubble_up_bitcast_in_strided_slice_insert_different_rank(%dst: vector<16x4x8xf16>, %src: vector<2x4xf16>) -> vector<16x4x4xf32> {
func.func @bubble_up_bitcast_in_strided_slice_insert_different_rank(%dst: vector<16x4x8xf16>, %src: vector<2x4xf16>) -> vector<16x4x4xf32> {
// CHECK: vector.insert_strided_slice
// CHECK-NEXT: vector.bitcast
%0 = vector.insert_strided_slice %src, %dst {offsets = [0, 0, 2], strides = [1, 1]} : vector<2x4xf16> into vector<16x4x8xf16>

44
mlir/test/Dialect/Vector/vector-transpose-lowering.mlir
View File

@ -19,7 +19,7 @@
// ELTWISE: %[[T10:.*]] = vector.extract %[[A]][1, 2] : vector<2x3xf32>
// ELTWISE: %[[T11:.*]] = vector.insert %[[T10]], %[[T9]] [2, 1] : f32 into vector<3x2xf32>
// ELTWISE: return %[[T11]] : vector<3x2xf32>
func @transpose23(%arg0: vector<2x3xf32>) -> vector<3x2xf32> {
func.func @transpose23(%arg0: vector<2x3xf32>) -> vector<3x2xf32> {
%0 = vector.transpose %arg0, [1, 0] : vector<2x3xf32> to vector<3x2xf32>
return %0 : vector<3x2xf32>
}
@ -28,7 +28,7 @@ func @transpose23(%arg0: vector<2x3xf32>) -> vector<3x2xf32> {
// SHUFFLE-LABEL: func @transpose
// FLAT-LABEL: func @transpose(
func @transpose(%arg0: vector<2x4xf32>) -> vector<4x2xf32> {
func.func @transpose(%arg0: vector<2x4xf32>) -> vector<4x2xf32> {
// SHUFFLE: vector.shape_cast %{{.*}} : vector<2x4xf32> to vector<8xf32>
// 0 4
// 0 1 2 3 1 5
@ -47,7 +47,7 @@ func @transpose(%arg0: vector<2x4xf32>) -> vector<4x2xf32> {
// -----
// AVX2-LABEL: func @transpose4x8
func @transpose4x8xf32(%arg0: vector<4x8xf32>) -> vector<8x4xf32> {
func.func @transpose4x8xf32(%arg0: vector<4x8xf32>) -> vector<8x4xf32> {
// AVX2: vector.extract {{.*}}[0]
// AVX2-NEXT: vector.extract {{.*}}[1]
// AVX2-NEXT: vector.extract {{.*}}[2]
@ -77,7 +77,7 @@ func @transpose4x8xf32(%arg0: vector<4x8xf32>) -> vector<8x4xf32> {
// -----
// AVX2-LABEL: func @transpose021_1x4x8
func @transpose021_1x4x8xf32(%arg0: vector<1x4x8xf32>) -> vector<1x8x4xf32> {
func.func @transpose021_1x4x8xf32(%arg0: vector<1x4x8xf32>) -> vector<1x8x4xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[0, 1]
// AVX2-NEXT: vector.extract {{.*}}[0, 2]
@ -107,7 +107,7 @@ func @transpose021_1x4x8xf32(%arg0: vector<1x4x8xf32>) -> vector<1x8x4xf32> {
// -----
// AVX2-LABEL: func @transpose8x8
func @transpose8x8xf32(%arg0: vector<8x8xf32>) -> vector<8x8xf32> {
func.func @transpose8x8xf32(%arg0: vector<8x8xf32>) -> vector<8x8xf32> {
// AVX2: vector.extract {{.*}}[0]
// AVX2-NEXT: vector.extract {{.*}}[1]
// AVX2-NEXT: vector.extract {{.*}}[2]
@ -150,7 +150,7 @@ func @transpose8x8xf32(%arg0: vector<8x8xf32>) -> vector<8x8xf32> {
// -----
// AVX2-LABEL: func @transpose021_1x8x8
func @transpose021_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<1x8x8xf32> {
func.func @transpose021_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<1x8x8xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[0, 1]
// AVX2-NEXT: vector.extract {{.*}}[0, 2]
@ -194,7 +194,7 @@ func @transpose021_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<1x8x8xf32> {
// -----
// AVX2-LABEL: func @transpose120_8x1x8
func @transpose120_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<1x8x8xf32> {
func.func @transpose120_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<1x8x8xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[1, 0]
// AVX2-NEXT: vector.extract {{.*}}[2, 0]
@ -238,7 +238,7 @@ func @transpose120_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<1x8x8xf32> {
// -----
// AVX2-LABEL: func @transpose120_8x8x1
func @transpose120_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x1x8xf32> {
func.func @transpose120_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x1x8xf32> {
// AVX2: vector.shape_cast %{{.*}} : vector<8x8x1xf32> to vector<8x8xf32>
// AVX2-NEXT: vector.extract {{.*}}[0]
// AVX2-NEXT: vector.extract {{.*}}[1]
@ -283,7 +283,7 @@ func @transpose120_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x1x8xf32> {
// -----
// AVX2-LABEL: func @transpose102_8x8x1
func @transpose102_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x8x1xf32> {
func.func @transpose102_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x8x1xf32> {
// AVX2: vector.shape_cast %{{.*}} : vector<8x8x1xf32> to vector<8x8xf32>
// AVX2-NEXT: vector.extract {{.*}}[0]
// AVX2-NEXT: vector.extract {{.*}}[1]
@ -328,7 +328,7 @@ func @transpose102_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x8x1xf32> {
// -----
// AVX2-LABEL: func @transpose201_8x1x8
func @transpose201_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x8x1xf32> {
func.func @transpose201_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x8x1xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[1, 0]
// AVX2-NEXT: vector.extract {{.*}}[2, 0]
@ -372,7 +372,7 @@ func @transpose201_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x8x1xf32> {
// -----
// AVX2-LABEL: func @transpose201_1x8x8
func @transpose201_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x1x8xf32> {
func.func @transpose201_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x1x8xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[0, 1]
// AVX2-NEXT: vector.extract {{.*}}[0, 2]
@ -416,7 +416,7 @@ func @transpose201_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x1x8xf32> {
// -----
// AVX2-LABEL: func @transpose210_8x1x8
func @transpose210_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x1x8xf32> {
func.func @transpose210_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x1x8xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[1, 0]
// AVX2-NEXT: vector.extract {{.*}}[2, 0]
@ -460,7 +460,7 @@ func @transpose210_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x1x8xf32> {
// -----
// AVX2-LABEL: func @transpose210_8x8x1
func @transpose210_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<1x8x8xf32> {
func.func @transpose210_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<1x8x8xf32> {
// AVX2: vector.shape_cast %{{.*}} : vector<8x8x1xf32> to vector<8x8xf32>
// AVX2-NEXT: vector.extract {{.*}}[0]
// AVX2-NEXT: vector.extract {{.*}}[1]
@ -505,7 +505,7 @@ func @transpose210_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<1x8x8xf32> {
// -----
// AVX2-LABEL: func @transpose210_1x8x8
func @transpose210_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x8x1xf32> {
func.func @transpose210_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x8x1xf32> {
// AVX2: vector.extract {{.*}}[0, 0]
// AVX2-NEXT: vector.extract {{.*}}[0, 1]
// AVX2-NEXT: vector.extract {{.*}}[0, 2]
@ -548,7 +548,7 @@ func @transpose210_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x8x1xf32> {
// -----
func @do_not_lower_nonf32_to_avx2(%arg0: vector<4x8xi32>) -> vector<8x4xi32> {
func.func @do_not_lower_nonf32_to_avx2(%arg0: vector<4x8xi32>) -> vector<8x4xi32> {
%0 = vector.transpose %arg0, [1, 0] : vector<4x8xi32> to vector<8x4xi32>
return %0 : vector<8x4xi32>
}
@ -558,7 +558,7 @@ func @do_not_lower_nonf32_to_avx2(%arg0: vector<4x8xi32>) -> vector<8x4xi32> {
// -----
// AVX2-LABEL: func @transpose021_8x1x8
func @transpose021_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x8x1xf32> {
func.func @transpose021_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x8x1xf32> {
%0 = vector.transpose %arg0, [0, 2, 1] : vector<8x1x8xf32> to vector<8x8x1xf32>
return %0 : vector<8x8x1xf32>
}
@ -568,7 +568,7 @@ func @transpose021_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<8x8x1xf32> {
// -----
// AVX2-LABEL: func @transpose021_8x8x1
func @transpose021_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x1x8xf32> {
func.func @transpose021_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x1x8xf32> {
%0 = vector.transpose %arg0, [0, 2, 1] : vector<8x8x1xf32> to vector<8x1x8xf32>
return %0 : vector<8x1x8xf32>
}
@ -579,7 +579,7 @@ func @transpose021_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<8x1x8xf32> {
// ELTWISE-LABEL: func @transpose102_1x8x8xf32
// AVX2-LABEL: func @transpose102_1x8x8
func @transpose102_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x1x8xf32> {
func.func @transpose102_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x1x8xf32> {
// ELTWISE: vector.extract {{.*}}[0, 0] : vector<1x8x8xf32>
// ELTWISE-NEXT: vector.insert {{.*}} [0, 0] : vector<8xf32> into vector<8x1x8xf32>
// ELTWISE-NEXT: vector.extract {{.*}}[0, 1] : vector<1x8x8xf32>
@ -606,7 +606,7 @@ func @transpose102_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x1x8xf32> {
// ELTWISE-LABEL: func @transpose102_8x1x8xf32
// AVX2-LABEL: func @transpose102_8x1x8
func @transpose102_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<1x8x8xf32> {
func.func @transpose102_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<1x8x8xf32> {
// ELTWISE: vector.extract {{.*}}[0, 0] : vector<8x1x8xf32>
// ELTWISE-NEXT: vector.insert {{.*}} [0, 0] : vector<8xf32> into vector<1x8x8xf32>
// ELTWISE-NEXT: vector.extract {{.*}}[1, 0] : vector<8x1x8xf32>
@ -633,7 +633,7 @@ func @transpose102_8x1x8xf32(%arg0: vector<8x1x8xf32>) -> vector<1x8x8xf32> {
// ELTWISE-LABEL: func @transpose1023_1x1x8x8xf32(
// AVX2-LABEL: func @transpose1023_1x1x8x8
func @transpose1023_1x1x8x8xf32(%arg0: vector<1x1x8x8xf32>) -> vector<1x1x8x8xf32> {
func.func @transpose1023_1x1x8x8xf32(%arg0: vector<1x1x8x8xf32>) -> vector<1x1x8x8xf32> {
// Note the single 2-D extract/insert pair since 2 and 3 are not transposed!
// ELTWISE: vector.extract {{.*}}[0, 0] : vector<1x1x8x8xf32>
// ELTWISE-NEXT: vector.insert {{.*}} [0, 0] : vector<8x8xf32> into vector<1x1x8x8xf32>
@ -646,7 +646,7 @@ func @transpose1023_1x1x8x8xf32(%arg0: vector<1x1x8x8xf32>) -> vector<1x1x8x8xf3
// -----
// AVX2-LABEL: func @transpose120_1x8x8
func @transpose120_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x8x1xf32> {
func.func @transpose120_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x8x1xf32> {
%0 = vector.transpose %arg0, [1, 2, 0] : vector<1x8x8xf32> to vector<8x8x1xf32>
return %0 : vector<8x8x1xf32>
@ -657,7 +657,7 @@ func @transpose120_1x8x8xf32(%arg0: vector<1x8x8xf32>) -> vector<8x8x1xf32> {
// -----
// AVX2-LABEL: func @transpose201_8x8x1
func @transpose201_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<1x8x8xf32> {
func.func @transpose201_8x8x1xf32(%arg0: vector<8x8x1xf32>) -> vector<1x8x8xf32> {
%0 = vector.transpose %arg0, [2, 0, 1] : vector<8x8x1xf32> to vector<1x8x8xf32>
return %0 : vector<1x8x8xf32>
}

12
mlir/test/Dialect/Vector/vector-unroll-options.mlir
View File

@ -1,6 +1,6 @@
// RUN: mlir-opt %s -test-vector-unrolling-patterns=unroll-based-on-type | FileCheck %s
func @vector_contract_f32(%lhs : vector<8x8xf32>, %rhs : vector<8x8xf32>,
func.func @vector_contract_f32(%lhs : vector<8x8xf32>, %rhs : vector<8x8xf32>,
%init : vector<8x8xf32>) -> vector<8x8xf32> {
%0 = vector.contract
{indexing_maps = [affine_map<(i, j, k) -> (i, k)>,
@ -45,7 +45,7 @@ func @vector_contract_f32(%lhs : vector<8x8xf32>, %rhs : vector<8x8xf32>,
// CHECK-SAME: vector<4x2xf32>, vector<4x2xf32> into vector<4x4xf32>
// CHECK: return
func @vector_contract_f16(%lhs : vector<8x8xf16>, %rhs : vector<8x8xf16>,
func.func @vector_contract_f16(%lhs : vector<8x8xf16>, %rhs : vector<8x8xf16>,
%init : vector<8x8xf16>) -> vector<8x8xf16> {
%0 = vector.contract
{indexing_maps = [affine_map<(i, j, k) -> (i, k)>,
@ -74,14 +74,14 @@ func @vector_contract_f16(%lhs : vector<8x8xf16>, %rhs : vector<8x8xf16>,
// CHECK-SAME: vector<4x4xf16>, vector<4x4xf16> into vector<4x4xf16>
// CHECK: return
func @vector_fma(%a: vector<4x4xf32>, %b: vector<4x4xf32>, %c: vector<4x4xf32>) -> vector<4x4xf32> {
func.func @vector_fma(%a: vector<4x4xf32>, %b: vector<4x4xf32>, %c: vector<4x4xf32>) -> vector<4x4xf32> {
%0 = vector.fma %a, %b, %c: vector<4x4xf32>
return %0 : vector<4x4xf32>
}
// CHECK-LABEL: func @vector_fma
// CHECK-COUNT-4: vector.fma %{{.+}}, %{{.+}}, %{{.+}} : vector<2x2xf32>
func @vector_multi_reduction(%v : vector<4x6xf32>) -> vector<4xf32> {
func.func @vector_multi_reduction(%v : vector<4x6xf32>) -> vector<4xf32> {
%0 = vector.multi_reduction #vector.kind<add>, %v [1] : vector<4x6xf32> to vector<4xf32>
return %0 : vector<4xf32>
}
@ -108,7 +108,7 @@ func @vector_multi_reduction(%v : vector<4x6xf32>) -> vector<4xf32> {
// CHECK: return %[[V2]] : vector<4xf32>
func @vector_reduction(%v : vector<8xf32>) -> f32 {
func.func @vector_reduction(%v : vector<8xf32>) -> f32 {
%0 = vector.reduction <add>, %v : vector<8xf32> into f32
return %0 : f32
}
@ -127,7 +127,7 @@ func @vector_reduction(%v : vector<8xf32>) -> f32 {
// CHECK: %[[add3:.*]] = arith.addf %[[add2]], %[[r3]]
// CHECK: return %[[add3]]
func @vector_tranpose(%v : vector<2x4x3x8xf32>) -> vector<2x3x8x4xf32> {
func.func @vector_tranpose(%v : vector<2x4x3x8xf32>) -> vector<2x3x8x4xf32> {
%t = vector.transpose %v, [0, 2, 3, 1] : vector<2x4x3x8xf32> to vector<2x3x8x4xf32>
return %t : vector<2x3x8x4xf32>
}

10
mlir/test/Dialect/X86Vector/legalize-for-llvm.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt %s -convert-vector-to-llvm="enable-x86vector" | mlir-opt | FileCheck %s
// CHECK-LABEL: func @avx512_mask_rndscale
func @avx512_mask_rndscale(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16, %i8: i8)
func.func @avx512_mask_rndscale(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16, %i8: i8)
-> (vector<16xf32>, vector<8xf64>, vector<16xf32>, vector<8xf64>)
{
// CHECK: x86vector.avx512.intr.mask.rndscale.ps.512
@ -19,7 +19,7 @@ func @avx512_mask_rndscale(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i1
}
// CHECK-LABEL: func @avx512_mask_compress
func @avx512_mask_compress(%k1: vector<16xi1>, %a1: vector<16xf32>,
func.func @avx512_mask_compress(%k1: vector<16xi1>, %a1: vector<16xf32>,
%k2: vector<8xi1>, %a2: vector<8xi64>)
-> (vector<16xf32>, vector<16xf32>, vector<8xi64>)
{
@ -33,7 +33,7 @@ func @avx512_mask_compress(%k1: vector<16xi1>, %a1: vector<16xf32>,
}
// CHECK-LABEL: func @avx512_vp2intersect
func @avx512_vp2intersect(%a: vector<16xi32>, %b: vector<8xi64>)
func.func @avx512_vp2intersect(%a: vector<16xi32>, %b: vector<8xi64>)
-> (vector<16xi1>, vector<16xi1>, vector<8xi1>, vector<8xi1>)
{
// CHECK: x86vector.avx512.intr.vp2intersect.d.512
@ -44,7 +44,7 @@ func @avx512_vp2intersect(%a: vector<16xi32>, %b: vector<8xi64>)
}
// CHECK-LABEL: func @avx_rsqrt
func @avx_rsqrt(%a: vector<8xf32>) -> (vector<8xf32>)
func.func @avx_rsqrt(%a: vector<8xf32>) -> (vector<8xf32>)
{
// CHECK: x86vector.avx.intr.rsqrt.ps.256
%0 = x86vector.avx.rsqrt %a : vector<8xf32>
@ -52,7 +52,7 @@ func @avx_rsqrt(%a: vector<8xf32>) -> (vector<8xf32>)
}
// CHECK-LABEL: func @avx_dot
func @avx_dot(%a: vector<8xf32>, %b: vector<8xf32>) -> (vector<8xf32>)
func.func @avx_dot(%a: vector<8xf32>, %b: vector<8xf32>) -> (vector<8xf32>)
{
// CHECK: x86vector.avx.intr.dp.ps.256
%0 = x86vector.avx.intr.dot %a, %b : vector<8xf32>

12
mlir/test/Dialect/X86Vector/roundtrip.mlir
View File

@ -1,7 +1,7 @@
// RUN: mlir-opt -verify-diagnostics %s | mlir-opt | FileCheck %s
// CHECK-LABEL: func @avx512_mask_rndscale
func @avx512_mask_rndscale(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16, %i8: i8)
func.func @avx512_mask_rndscale(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16, %i8: i8)
-> (vector<16xf32>, vector<8xf64>)
{
// CHECK: x86vector.avx512.mask.rndscale {{.*}}: vector<16xf32>
@ -12,7 +12,7 @@ func @avx512_mask_rndscale(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i1
}
// CHECK-LABEL: func @avx512_scalef
func @avx512_scalef(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16, %i8: i8)
func.func @avx512_scalef(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16, %i8: i8)
-> (vector<16xf32>, vector<8xf64>)
{
// CHECK: x86vector.avx512.mask.scalef {{.*}}: vector<16xf32>
@ -23,7 +23,7 @@ func @avx512_scalef(%a: vector<16xf32>, %b: vector<8xf64>, %i32: i32, %i16: i16,
}
// CHECK-LABEL: func @avx512_vp2intersect
func @avx512_vp2intersect(%a: vector<16xi32>, %b: vector<8xi64>)
func.func @avx512_vp2intersect(%a: vector<16xi32>, %b: vector<8xi64>)
-> (vector<16xi1>, vector<16xi1>, vector<8xi1>, vector<8xi1>)
{
// CHECK: x86vector.avx512.vp2intersect {{.*}} : vector<16xi32>
@ -34,7 +34,7 @@ func @avx512_vp2intersect(%a: vector<16xi32>, %b: vector<8xi64>)
}
// CHECK-LABEL: func @avx512_mask_compress
func @avx512_mask_compress(%k1: vector<16xi1>, %a1: vector<16xf32>,
func.func @avx512_mask_compress(%k1: vector<16xi1>, %a1: vector<16xf32>,
%k2: vector<8xi1>, %a2: vector<8xi64>)
-> (vector<16xf32>, vector<16xf32>, vector<8xi64>)
{
@ -48,7 +48,7 @@ func @avx512_mask_compress(%k1: vector<16xi1>, %a1: vector<16xf32>,
}
// CHECK-LABEL: func @avx_rsqrt
func @avx_rsqrt(%a: vector<8xf32>) -> (vector<8xf32>)
func.func @avx_rsqrt(%a: vector<8xf32>) -> (vector<8xf32>)
{
// CHECK: x86vector.avx.rsqrt {{.*}} : vector<8xf32>
%0 = x86vector.avx.rsqrt %a : vector<8xf32>
@ -56,7 +56,7 @@ func @avx_rsqrt(%a: vector<8xf32>) -> (vector<8xf32>)
}
// CHECK-LABEL: func @avx_dot
func @avx_dot(%a: vector<8xf32>, %b: vector<8xf32>) -> (vector<8xf32>)
func.func @avx_dot(%a: vector<8xf32>, %b: vector<8xf32>) -> (vector<8xf32>)
{
// CHECK: x86vector.avx.intr.dot {{.*}} : vector<8xf32>
%0 = x86vector.avx.intr.dot %a, %b : vector<8xf32>