[PATCH] D147152: [LoopVectorize] Fix zext/sext cost calculations when types are shrunk

[David Sherwood via Phabricator via llvm-commits]

david-arm created this revision.
david-arm added reviewers: sdesmalen, hassnaa-arm, fhahn, dmgreen.
Herald added subscribers: shiva0217, StephenFan, hiraditya.
Herald added a project: All.
david-arm requested review of this revision.
Herald added subscribers: llvm-commits, pcwang-thead.
Herald added a project: LLVM.

In getInstructionCost if we know a zext/sext is going to be shrunk
we should only be changing the destination type, and leave the
source type unchanged. For example, we may change a zext from

  zext <16 x i8> %a to <16 x i32>

to

  zext <16 x i8> %a to <16 x i16>

However, we were previously calculating the cost for doing

  zext <16 x i16> %a to <16 x i16>

which is incorrect.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D147152

Files:
  llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
  llvm/test/Transforms/LoopVectorize/AArch64/type-shrinkage-zext-costs.ll


Index: llvm/test/Transforms/LoopVectorize/AArch64/type-shrinkage-zext-costs.ll
===================================================================
--- llvm/test/Transforms/LoopVectorize/AArch64/type-shrinkage-zext-costs.ll
+++ llvm/test/Transforms/LoopVectorize/AArch64/type-shrinkage-zext-costs.ll
@@ -15,7 +15,7 @@
 ; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 1 For instruction:   %conv = zext i8 %0 to i32
 ; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 2 For instruction:   %conv = zext i8 %0 to i32
 ; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 4 For instruction:   %conv = zext i8 %0 to i32
-; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 8 For instruction:   %conv = zext i8 %0 to i32
+; CHECK-COST: LV: Found an estimated cost of 0 for VF vscale x 8 For instruction:   %conv = zext i8 %0 to i32
 
 ; CHECK-LABEL: define void @zext_i8_i16
 ; CHECK-SAME: (ptr noalias nocapture readonly [[P:%.*]], ptr noalias nocapture [[Q:%.*]], i32 [[LEN:%.*]]) #[[ATTR0:[0-9]+]] {
@@ -105,7 +105,7 @@
 ; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 1 For instruction:   %conv = sext i8 %0 to i32
 ; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 2 For instruction:   %conv = sext i8 %0 to i32
 ; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 4 For instruction:   %conv = sext i8 %0 to i32
-; CHECK-COST: LV: Found an estimated cost of 1 for VF vscale x 8 For instruction:   %conv = sext i8 %0 to i32
+; CHECK-COST: LV: Found an estimated cost of 0 for VF vscale x 8 For instruction:   %conv = sext i8 %0 to i32
 
 ; CHECK-LABEL: define void @sext_i8_i16
 ; CHECK-SAME: (ptr noalias nocapture readonly [[P:%.*]], ptr noalias nocapture [[Q:%.*]], i32 [[LEN:%.*]]) #[[ATTR0]] {
Index: llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
===================================================================
--- llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
+++ llvm/lib/Transforms/Vectorize/LoopVectorize.cpp
@@ -7373,7 +7373,8 @@
         VectorTy =
             largestIntegerVectorType(ToVectorTy(I->getType(), VF), MinVecTy);
       } else if (Opcode == Instruction::ZExt || Opcode == Instruction::SExt) {
-        SrcVecTy = largestIntegerVectorType(SrcVecTy, MinVecTy);
+        // Leave SrcVecTy unchanged - we only shrink the destination element
+        // type.
         VectorTy =
             smallestIntegerVectorType(ToVectorTy(I->getType(), VF), MinVecTy);
       }


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D147152.509341.patch
Type: text/x-patch
Size: 2481 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230329/8f943ca1/attachment.bin>