mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2025-01-27 06:54:30 +00:00
Do not vectorize loops with subtraction reductions
Since subtraction does not commute the loop vectorizer incorrectly vectorizes reductions such as x = A[i] - x. Disabling for now. llvm-svn: 171537
This commit is contained in:
parent
3f90fbe286
commit
6ce33a6ae9
@ -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;
|
||||
|
@ -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
|
||||
|
@ -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
|
||||
}
|
||||
|
Loading…
x
Reference in New Issue
Block a user