[PATCH] D114577: [AArch64][SVEIntrinsicOpts] Fix: predicated SVE mul/fmul are not commutative

Igor Kirillov via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Nov 25 03:42:12 PST 2021


igor.kirillov created this revision.
Herald added subscribers: ctetreau, hiraditya, kristof.beyls, tschuett.
igor.kirillov requested review of this revision.
Herald added a project: LLVM.
Herald added a subscriber: llvm-commits.

We can not swap multiplicand and multiplier because the sve intrinsics
are predicated. Imagine lanes in vectors having the following values:

  pg = 0
  multiplicand = 1 (from dup)
  multiplier = 2

The resulting value should be 1, but if we swap multiplicand and multiplier it will become 2,
which is incorrect.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D114577

Files:
  llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
  llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
  llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll


Index: llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
===================================================================
--- llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
+++ llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-mul-idempotency.ll
@@ -32,7 +32,8 @@
 
 define <vscale x 2 x i64> @idempotent_mul_different_argument_order(<vscale x 2 x i1> %pg, <vscale x 2 x i64> %a) #0 {
 ; CHECK-LABEL: @idempotent_mul_different_argument_order(
-; CHECK-NEXT:    ret <vscale x 2 x i64> [[A:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 1, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer), <vscale x 2 x i64> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 ;
   %1 = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 1)
   ; Different argument order to the above tests.
Index: llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
===================================================================
--- llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
+++ llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-fmul-idempotency.ll
@@ -32,7 +32,8 @@
 
 define <vscale x 2 x double> @idempotent_fmul_different_argument_order(<vscale x 2 x i1> %pg, <vscale x 2 x double> %a) #0 {
 ; CHECK-LABEL: @idempotent_fmul_different_argument_order(
-; CHECK-NEXT:    ret <vscale x 2 x double> [[A:%.*]]
+; CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[PG:%.*]], <vscale x 2 x double> shufflevector (<vscale x 2 x double> insertelement (<vscale x 2 x double> poison, double 1.000000e+00, i32 0), <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer), <vscale x 2 x double> [[A:%.*]])
+; CHECK-NEXT:    ret <vscale x 2 x double> [[TMP1]]
 ;
   %1 = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double 1.0)
   ; Different argument order to the above tests.
Index: llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
===================================================================
--- llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -833,11 +833,6 @@
     return match(SplatValue, m_FPOne()) || match(SplatValue, m_One());
   };
 
-  // The OpMultiplier variable should always point to the dup (if any), so
-  // swap if necessary.
-  if (IsUnitDup(OpMultiplicand) || IsUnitSplat(OpMultiplicand))
-    std::swap(OpMultiplier, OpMultiplicand);
-
   if (IsUnitSplat(OpMultiplier)) {
     // [f]mul pg (dupx 1) %n => %n
     OpMultiplicand->takeName(&II);


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D114577.389722.patch
Type: text/x-patch
Size: 2825 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20211125/0c0a3ec2/attachment.bin>


More information about the llvm-commits mailing list