[PATCH] D105520: [AArch64][SVE][InstCombine] last{a,b} of a splat vector

[Usman Nadeem via Phabricator via llvm-commits]

mnadeem created this revision.
Herald added subscribers: psnobl, hiraditya, kristof.beyls, tschuett.
Herald added a reviewer: efriedma.
mnadeem requested review of this revision.
Herald added a project: LLVM.
Herald added a subscriber: llvm-commits.

Replace last{a,b}(splat(X)) with X, irrespective of the predicate.


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D105520

Files:
  llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
  llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-lasta-lastb.ll


Index: llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-lasta-lastb.ll
===================================================================
--- llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-lasta-lastb.ll
+++ llvm/test/Transforms/InstCombine/AArch64/sve-intrinsic-opts-lasta-lastb.ll
@@ -144,6 +144,25 @@
   ret i8 %e
 }
 
+; Return the splatted value irrespective of the predicate.
+define i8 @lasta_splat(<vscale x 16 x i1> %pg, i8 %a) #0 {
+; OPT-LABEL: @lasta_splat(
+; OPT-NEXT:    ret i8 %a
+  %splat_insert = insertelement <vscale x 16 x i8> poison, i8 %a, i32 0
+  %splat = shufflevector <vscale x 16 x i8> %splat_insert, <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
+  %last = tail call i8 @llvm.aarch64.sve.lasta.nxv16i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %splat)
+  ret i8 %last
+}
+
+define i8 @lastb_splat(<vscale x 16 x i1> %pg, i8 %a) #0 {
+; OPT-LABEL: @lastb_splat(
+; OPT-NEXT:    ret i8 %a
+  %splat_insert = insertelement <vscale x 16 x i8> poison, i8 %a, i32 0
+  %splat = shufflevector <vscale x 16 x i8> %splat_insert, <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
+  %last = tail call i8 @llvm.aarch64.sve.lastb.nxv16i8(<vscale x 16 x i1> %pg, <vscale x 16 x i8> %splat)
+  ret i8 %last
+}
+
 declare <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32)
 declare i8 @llvm.aarch64.sve.lasta.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>)
 declare i8 @llvm.aarch64.sve.lastb.nxv16i8(<vscale x 16 x i1>, <vscale x 16 x i8>)
Index: llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
===================================================================
--- llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
+++ llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp
@@ -501,6 +501,10 @@
   Value *Vec = II.getArgOperand(1);
   bool IsAfter = II.getIntrinsicID() == Intrinsic::aarch64_sve_lasta;
 
+  // lastX(splat(X)) --> X
+  if (auto *SplatVal = getSplatValue(Vec))
+    return IC.replaceInstUsesWith(II, SplatVal);
+
   auto *C = dyn_cast<Constant>(Pg);
   if (IsAfter && C && C->isNullValue()) {
     // The intrinsic is extracting lane 0 so use an extract instead.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D105520.356847.patch
Type: text/x-patch
Size: 2169 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210707/197414c8/attachment.bin>