diff options
| author | Paul Redmond <paul.redmond@intel.com> | 2013-01-04 22:10:16 +0000 |
|---|---|---|
| committer | Paul Redmond <paul.redmond@intel.com> | 2013-01-04 22:10:16 +0000 |
| commit | 5767d919569f7b4be51c0a13bff24f722625d9f3 (patch) | |
| tree | fdeba97aeb1434f7c6b6adcc693d69a129d2f627 /test/Transforms/LoopVectorize/reduction.ll | |
| parent | 8cab6edf4fa46d9bd06ff3ce0e1dd4557be58b65 (diff) | |
| download | llvm-5767d919569f7b4be51c0a13bff24f722625d9f3.tar.gz llvm-5767d919569f7b4be51c0a13bff24f722625d9f3.tar.bz2 llvm-5767d919569f7b4be51c0a13bff24f722625d9f3.tar.xz | |
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.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@171537 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/Transforms/LoopVectorize/reduction.ll')
| -rw-r--r-- | test/Transforms/LoopVectorize/reduction.ll | 50 |
1 files changed, 50 insertions, 0 deletions
diff --git a/test/Transforms/LoopVectorize/reduction.ll b/test/Transforms/LoopVectorize/reduction.ll index bc1c0290de..91fc953877 100644 --- a/test/Transforms/LoopVectorize/reduction.ll +++ b/test/Transforms/LoopVectorize/reduction.ll @@ -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 +} |