6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2024-10-02 17:55:18 +00:00
Code Issues Projects Releases Wiki Activity

Do not vectorize loops with subtraction reductions

Browse Source 
Since subtraction does not commute the loop vectorizer incorrectly vectorizes
reductions such as x = A[i] - x.

Disabling for now.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171537 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Paul Redmond 2013-01-04 22:10:16 +00:00
parent 8cab6edf4f
commit 5767d91956
3 changed files with 51 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
lib/Transforms/Vectorize/LoopVectorize.cpp
View File

@ -1986,7 +1986,6 @@ LoopVectorizationLegality::isReductionInstr(Instruction *I,
// possibly.
return true;
case Instruction::Add:
case Instruction::Sub:
return Kind == IntegerAdd;
case Instruction::Mul:
return Kind == IntegerMult;

2
test/Transforms/LoopVectorize/gcc-examples.ll
View File

@ -241,7 +241,7 @@ define void @example8(i32 %x) nounwind uwtable ssp {
}
;CHECK: @example9
;CHECK: phi <4 x i32>
;CHECK-NOT: phi <4 x i32>
;CHECK: ret i32
define i32 @example9() nounwind uwtable readonly ssp {
br label %1

50
test/Transforms/LoopVectorize/reduction.ll
View File

@ -270,3 +270,53 @@ for.end: ; preds = %for.body, %entry
%result.0.lcssa = phi i32 [ 0, %entry ], [ %xor, %for.body ]
ret i32 %result.0.lcssa
}
;CHECK: @reduction_sub_rhs
;CHECK-NOT: phi <4 x i32>
;CHECK-NOT: sub nsw <4 x i32>
;CHECK: ret i32
define i32 @reduction_sub_rhs(i32 %n, i32* noalias nocapture %A) nounwind uwtable readonly {
entry:
%cmp3 = icmp sgt i32 %n, 0
br i1 %cmp3, label %for.body, label %for.end
for.body: ; preds = %entry, %for.body
%indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
%x.04 = phi i32 [ %sub, %for.body ], [ 0, %entry ]
%arrayidx = getelementptr inbounds i32* %A, i64 %indvars.iv
%0 = load i32* %arrayidx, align 4
%sub = sub nsw i32 %x.04, %0
%indvars.iv.next = add i64 %indvars.iv, 1
%lftr.wideiv = trunc i64 %indvars.iv.next to i32
%exitcond = icmp eq i32 %lftr.wideiv, %n
br i1 %exitcond, label %for.end, label %for.body
for.end: ; preds = %for.body, %entry
%x.0.lcssa = phi i32 [ 0, %entry ], [ %sub, %for.body ]
ret i32 %x.0.lcssa
}
;CHECK: @reduction_sub_lhs
;CHECK-NOT: phi <4 x i32>
;CHECK-NOT: sub nsw <4 x i32>
;CHECK: ret i32
define i32 @reduction_sub_lhs(i32 %n, i32* noalias nocapture %A) nounwind uwtable readonly {
entry:
%cmp4 = icmp sgt i32 %n, 0
br i1 %cmp4, label %for.body, label %for.end
for.body: ; preds = %entry, %for.body
%indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
%x.05 = phi i32 [ %sub, %for.body ], [ 0, %entry ]
%arrayidx = getelementptr inbounds i32* %A, i64 %indvars.iv
%0 = load i32* %arrayidx, align 4
%sub = sub nsw i32 %0, %x.05
%indvars.iv.next = add i64 %indvars.iv, 1
%lftr.wideiv = trunc i64 %indvars.iv.next to i32
%exitcond = icmp eq i32 %lftr.wideiv, %n
br i1 %exitcond, label %for.end, label %for.body
for.end: ; preds = %for.body, %entry
%x.0.lcssa = phi i32 [ 0, %entry ], [ %sub, %for.body ]
ret i32 %x.0.lcssa
}