maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/Transforms/LowerMatrixIntrinsics/multiply-double-contraction...

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

70 lines
6.5 KiB
LLVM
Raw Normal View History

[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -lower-matrix-intrinsics -S < %s | FileCheck %s
; RUN: opt -passes='lower-matrix-intrinsics' -S < %s | FileCheck %s
define <4 x double> @multiply_2x2(<4 x double> %a, <4 x double> %b) {
; CHECK-LABEL: @multiply_2x2(
; CHECK-NEXT: entry:
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[SPLIT:%.*]] = shufflevector <4 x double> [[A:%.*]], <4 x double> poison, <2 x i32> <i32 0, i32 1>
; CHECK-NEXT: [[SPLIT1:%.*]] = shufflevector <4 x double> [[A]], <4 x double> poison, <2 x i32> <i32 2, i32 3>
; CHECK-NEXT: [[SPLIT2:%.*]] = shufflevector <4 x double> [[B:%.*]], <4 x double> poison, <2 x i32> <i32 0, i32 1>
; CHECK-NEXT: [[SPLIT3:%.*]] = shufflevector <4 x double> [[B]], <4 x double> poison, <2 x i32> <i32 2, i32 3>
; CHECK-NEXT: [[BLOCK:%.*]] = shufflevector <2 x double> [[SPLIT]], <2 x double> poison, <1 x i32> zeroinitializer
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP0:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 0
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT:%.*]] = insertelement <1 x double> poison, double [[TMP0]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP1:%.*]] = fmul contract <1 x double> [[BLOCK]], [[SPLAT_SPLAT]]
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK4:%.*]] = shufflevector <2 x double> [[SPLIT1]], <2 x double> poison, <1 x i32> zeroinitializer
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 1
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT5:%.*]] = insertelement <1 x double> poison, double [[TMP2]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT6:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT5]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP3:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK4]], <1 x double> [[SPLAT_SPLAT6]], <1 x double> [[TMP1]])
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <1 x double> [[TMP3]], <1 x double> poison, <2 x i32> <i32 0, i32 undef>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP4]], <2 x i32> <i32 2, i32 1>
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK7:%.*]] = shufflevector <2 x double> [[SPLIT]], <2 x double> poison, <1 x i32> <i32 1>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP6:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 0
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT8:%.*]] = insertelement <1 x double> poison, double [[TMP6]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT9:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT8]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP7:%.*]] = fmul contract <1 x double> [[BLOCK7]], [[SPLAT_SPLAT9]]
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK10:%.*]] = shufflevector <2 x double> [[SPLIT1]], <2 x double> poison, <1 x i32> <i32 1>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP8:%.*]] = extractelement <2 x double> [[SPLIT2]], i64 1
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT11:%.*]] = insertelement <1 x double> poison, double [[TMP8]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT12:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT11]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP9:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK10]], <1 x double> [[SPLAT_SPLAT12]], <1 x double> [[TMP7]])
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[TMP10:%.*]] = shufflevector <1 x double> [[TMP9]], <1 x double> poison, <2 x i32> <i32 0, i32 undef>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP11:%.*]] = shufflevector <2 x double> [[TMP5]], <2 x double> [[TMP10]], <2 x i32> <i32 0, i32 2>
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK13:%.*]] = shufflevector <2 x double> [[SPLIT]], <2 x double> poison, <1 x i32> zeroinitializer
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP12:%.*]] = extractelement <2 x double> [[SPLIT3]], i64 0
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT14:%.*]] = insertelement <1 x double> poison, double [[TMP12]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT15:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT14]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP13:%.*]] = fmul contract <1 x double> [[BLOCK13]], [[SPLAT_SPLAT15]]
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK16:%.*]] = shufflevector <2 x double> [[SPLIT1]], <2 x double> poison, <1 x i32> zeroinitializer
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP14:%.*]] = extractelement <2 x double> [[SPLIT3]], i64 1
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT17:%.*]] = insertelement <1 x double> poison, double [[TMP14]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT18:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT17]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP15:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK16]], <1 x double> [[SPLAT_SPLAT18]], <1 x double> [[TMP13]])
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[TMP16:%.*]] = shufflevector <1 x double> [[TMP15]], <1 x double> poison, <2 x i32> <i32 0, i32 undef>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP17:%.*]] = shufflevector <2 x double> undef, <2 x double> [[TMP16]], <2 x i32> <i32 2, i32 1>
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK19:%.*]] = shufflevector <2 x double> [[SPLIT]], <2 x double> poison, <1 x i32> <i32 1>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP18:%.*]] = extractelement <2 x double> [[SPLIT3]], i64 0
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT20:%.*]] = insertelement <1 x double> poison, double [[TMP18]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT21:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT20]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP19:%.*]] = fmul contract <1 x double> [[BLOCK19]], [[SPLAT_SPLAT21]]
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[BLOCK22:%.*]] = shufflevector <2 x double> [[SPLIT1]], <2 x double> poison, <1 x i32> <i32 1>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP20:%.*]] = extractelement <2 x double> [[SPLIT3]], i64 1
[IR] Let IRBuilder's CreateVectorSplat/CreateShuffleVector use poison as placeholder This patch updates IRBuilder to create insertelement/shufflevector using poison as a placeholder. Reviewed By: nikic Differential Revision: https://reviews.llvm.org/D93793
2020-12-24 08:33:58 +08:00
; CHECK-NEXT: [[SPLAT_SPLATINSERT23:%.*]] = insertelement <1 x double> poison, double [[TMP20]], i32 0
; CHECK-NEXT: [[SPLAT_SPLAT24:%.*]] = shufflevector <1 x double> [[SPLAT_SPLATINSERT23]], <1 x double> poison, <1 x i32> zeroinitializer
[Matrix] Preserve existing fast-math flags during lowering This patch makes it so, floating-point instructions created in LowerMatrixIntrinsics retain fast-math flags from instructions that are higher up the chain. Fixes https://bugs.llvm.org/show_bug.cgi?id=49738 Reviewed By: fhahn Differential Revision: https://reviews.llvm.org/D103233
2021-06-03 22:11:00 +08:00
; CHECK-NEXT: [[TMP21:%.*]] = call contract <1 x double> @llvm.fmuladd.v1f64(<1 x double> [[BLOCK22]], <1 x double> [[SPLAT_SPLAT24]], <1 x double> [[TMP19]])
Use unary CreateShuffleVector if possible As mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be used instead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`. Let's update them. Actually, it would have been more natural if the patches were made in this order: (1) let them use unary CreateShuffleVector first (2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793) The order is swapped, but in terms of correctness it is still fine. Reviewed By: spatel Differential Revision: https://reviews.llvm.org/D93923
2020-12-30 06:28:17 +08:00
; CHECK-NEXT: [[TMP22:%.*]] = shufflevector <1 x double> [[TMP21]], <1 x double> poison, <2 x i32> <i32 0, i32 undef>
[Matrix] Use fmuladd for matrix.multiply if allowed. If the matrix.multiply calls have the contract fast math flag, we can use fmuladd. This als adds a command line option to force fmuladd generation. We can retire this option once there is a clang-level option. Reviewers: anemet, Gerolf, hfinkel, andrew.w.kaylor Reviewed By: anemet Differential Revision: https://reviews.llvm.org/D70951
2019-12-23 21:28:56 +08:00
; CHECK-NEXT: [[TMP23:%.*]] = shufflevector <2 x double> [[TMP17]], <2 x double> [[TMP22]], <2 x i32> <i32 0, i32 2>
; CHECK-NEXT: [[TMP24:%.*]] = shufflevector <2 x double> [[TMP11]], <2 x double> [[TMP23]], <4 x i32> <i32 0, i32 1, i32 2, i32 3>
; CHECK-NEXT: ret <4 x double> [[TMP24]]
;
entry:
%c = call contract <4 x double> @llvm.matrix.multiply.v4f64.v4f64.v4f64(<4 x double> %a, <4 x double> %b, i32 2, i32 2, i32 2)
ret <4 x double> %c
}
declare <4 x double> @llvm.matrix.multiply.v4f64.v4f64.v4f64(<4 x double>, <4 x double>, i32, i32, i32)