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

[mlir][ods] Use lambda in element type check pred rather than repeated casts 

Avoids multiple cast & getElementType calls. Just a local change for ShapedType
containers but reduces one model case from 24.7 to 24.04s.

Resultant code generated change:
https://gist.github.com/jpienaar/7ffd2e9b0737134ba2ea2729b91c9572

Differential Revision: https://reviews.llvm.org/D113621
This commit is contained in:
Jacques Pienaar 2021-11-10 16:27:37 -08:00
parent 4a0c225616
commit 32b327e4ed
2 changed files with 9 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
mlir/include/mlir/IR/OpBase.td
View File

@ -566,20 +566,17 @@ class ContainerType<Type etype, Pred containerPred, code elementTypeCall,
Type<And<[containerPred,
SubstLeaves<"$_self", !cast<string>(elementTypeCall),
etype.predicate>]>,
descr # " of " # etype.summary # " values", cppClassName> {
// The type of elements in the container.
Type elementType = etype;
// Call to retrieve.
code getElementTypeCall = elementTypeCall;
}
descr # " of " # etype.summary # " values", cppClassName>;
class ShapedContainerType<list<Type> allowedTypes,
Pred containerPred, string descr,
string cppClassName = "::mlir::Type"> :
ContainerType<AnyTypeOf<allowedTypes>, containerPred,
"$_self.cast<::mlir::ShapedType>().getElementType()", descr,
cppClassName>;
Type<And<[containerPred,
Concat<"[](::mlir::Type elementType) { return ",
SubstLeaves<"$_self", "elementType",
AnyTypeOf<allowedTypes>.predicate>,
"; }($_self.cast<::mlir::ShapedType>().getElementType())">]>,
descr # " of " # AnyTypeOf<allowedTypes>.summary # " values", cppClassName>;
// Whether a shaped type is ranked.
def HasRankPred : CPred<"$_self.cast<::mlir::ShapedType>().hasRank()">;

4
mlir/test/mlir-tblgen/predicate.td
View File

@ -25,11 +25,11 @@ def OpA : NS_Op<"op_for_CPred_containing_multiple_same_placeholder", []> {
// CHECK-NOT: return op->emitOpError(valueKind) << " #" << valueGroupStartIndex << " must be 32-bit integer or floating-point type, but got " << type;
// CHECK: static ::mlir::LogicalResult [[$TENSOR_CONSTRAINT:__mlir_ods_local_type_constraint.*]](
// CHECK-NEXT: if (!(((type.isa<::mlir::TensorType>())) && ((true)))) {
// CHECK-NEXT: if (!(((type.isa<::mlir::TensorType>())) && ([](Type elementType) { return (true); }(type.cast<::mlir::ShapedType>().getElementType())))) {
// CHECK-NEXT: return op->emitOpError(valueKind) << " #" << valueGroupStartIndex << " must be tensor of any type values, but got " << type;
// CHECK: static ::mlir::LogicalResult [[$TENSOR_INTEGER_FLOAT_CONSTRAINT:__mlir_ods_local_type_constraint.*]](
// CHECK-NEXT: if (!(((type.isa<::mlir::TensorType>())) && (((type.cast<::mlir::ShapedType>().getElementType().isF32())) || ((type.cast<::mlir::ShapedType>().getElementType().isSignlessInteger(32)))))) {
// CHECK-NEXT: if (!(((type.isa<::mlir::TensorType>())) && ([](Type elementType) { return ((elementType.isF32())) || ((elementType.isSignlessInteger(32))); }(type.cast<::mlir::ShapedType>().getElementType())))) {
// CHECK-NEXT: return op->emitOpError(valueKind) << " #" << valueGroupStartIndex << " must be tensor of 32-bit float or 32-bit signless integer values, but got " << type;
// CHECK-LABEL: OpA::verify