maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[SLP] Remove bogus assert. 

The removed assert seems bogus - it's perfectly legal for the roots of the
vectorized subtrees to be equal even if the original scalar values aren't,
if the original scalars happen to be equivalent.

This fixes PR31599.

Differential Revision: https://reviews.llvm.org/D28539

llvm-svn: 291692
This commit is contained in:
Michael Kuperstein 2017-01-11 19:23:57 +00:00
parent 3b5edf4fa7
commit f69e64662b
2 changed files with 30 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
llvm/lib/Transforms/Vectorize/SLPVectorizer.cpp
View File

@ -2493,10 +2493,6 @@ Value *BoUpSLP::vectorizeTree(TreeEntry *E) {
Value *LHS = vectorizeTree(LHSVL);
Value *RHS = vectorizeTree(RHSVL);
if (LHS == RHS && isa<Instruction>(LHS)) {
assert((VL0->getOperand(0) == VL0->getOperand(1)) && "Invalid order");
}
if (Value *V = alreadyVectorized(E->Scalars))
return V;

30
llvm/test/Transforms/SLPVectorizer/X86/pr31599.ll Normal file
View File

@ -0,0 +1,30 @@
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
; RUN: opt -slp-vectorizer -S -mtriple=x86_64-unknown-linux-gnu < %s | FileCheck %s
define <2 x float> @foo() {
; CHECK-LABEL: @foo(
; CHECK-NEXT: entry:
; CHECK-NEXT: [[SOURCE:%.*]] = insertelement <2 x float> undef, float undef, i32 0
; CHECK-NEXT: [[TMP0:%.*]] = fsub <2 x float> [[SOURCE]], [[SOURCE]]
; CHECK-NEXT: [[TMP1:%.*]] = extractelement <2 x float> [[TMP0]], i32 0
; CHECK-NEXT: [[RES1:%.*]] = insertelement <2 x float> undef, float [[TMP1]], i32 0
; CHECK-NEXT: [[TMP2:%.*]] = extractelement <2 x float> [[TMP0]], i32 1
; CHECK-NEXT: [[RES2:%.*]] = insertelement <2 x float> [[RES1]], float [[TMP2]], i32 1
; CHECK-NEXT: ret <2 x float> [[RES2]]
;
entry:
%source = insertelement <2 x float> undef, float undef, i32 0
%e0 = extractelement <2 x float> %source, i32 0
%e0.dup = extractelement <2 x float> %source, i32 0
%sub1 = fsub float %e0, %e0.dup
%e1 = extractelement <2 x float> %source, i32 1
%e1.dup = extractelement <2 x float> %source, i32 1
%sub2 = fsub float %e1, %e1.dup
%res1 = insertelement <2 x float> undef, float %sub1, i32 0
%res2 = insertelement <2 x float> %res1, float %sub2, i32 1
ret <2 x float> %res2
}
!llvm.ident = !{!0, !0}
!0 = !{!"clang version 4.0.0 "}