[PATCH] D141657: [X86] Prefer fpext(splat(X)) to splat(fpext(x)).

[Freddy, Ye via Phabricator via llvm-commits]

FreddyYe updated this revision to Diff 488951.
FreddyYe marked an inline comment as done.
FreddyYe added a comment.

Address comments. THX for review!


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D141657/new/

https://reviews.llvm.org/D141657

Files:
  llvm/lib/Target/X86/X86ISelLowering.cpp
  llvm/lib/Target/X86/X86ISelLowering.h
  llvm/test/CodeGen/X86/prefer-fpext-splat.ll


Index: llvm/test/CodeGen/X86/prefer-fpext-splat.ll
===================================================================
--- /dev/null
+++ llvm/test/CodeGen/X86/prefer-fpext-splat.ll
@@ -0,0 +1,15 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512vl | FileCheck %s
+
+define <4 x double> @prefer(float* %p) {
+; CHECK-LABEL: prefer:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    vcvtps2pd (%rdi){1to4}, %ymm0
+; CHECK-NEXT:    retq
+entry:
+  %0 = load float, float* %p, align 4
+  %vecinit.i = insertelement <4 x float> undef, float %0, i64 0
+  %vecinit3.i = shufflevector <4 x float> %vecinit.i, <4 x float> poison, <4 x i32> zeroinitializer
+  %conv.i = fpext <4 x float> %vecinit3.i to <4 x double>
+  ret <4 x double> %conv.i
+}
Index: llvm/lib/Target/X86/X86ISelLowering.h
===================================================================
--- llvm/lib/Target/X86/X86ISelLowering.h
+++ llvm/lib/Target/X86/X86ISelLowering.h
@@ -1091,6 +1091,8 @@
         unsigned OldShiftOpcode, unsigned NewShiftOpcode,
         SelectionDAG &DAG) const override;
 
+    bool preferScalarizeSplat(unsigned Opc) const override;
+
     bool shouldFoldConstantShiftPairToMask(const SDNode *N,
                                            CombineLevel Level) const override;
 
Index: llvm/lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- llvm/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -6012,6 +6012,10 @@
   return NewShiftOpcode == ISD::SHL;
 }
 
+bool X86TargetLowering::preferScalarizeSplat(unsigned Opc) const {
+  return Opc != ISD::FP_EXTEND;
+}
+
 bool X86TargetLowering::shouldFoldConstantShiftPairToMask(
     const SDNode *N, CombineLevel Level) const {
   assert(((N->getOpcode() == ISD::SHL &&


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D141657.488951.patch
Type: text/x-patch
Size: 1900 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20230113/2695ed73/attachment.bin>