diff options
author | Karthik Bhat <kv.bhat@samsung.com> | 2014-05-30 04:31:24 +0000 |
---|---|---|
committer | Karthik Bhat <kv.bhat@samsung.com> | 2014-05-30 04:31:24 +0000 |
commit | 8ffc96a163a9d4387feb06007d36ac1e419a8cd1 (patch) | |
tree | bc3b0ed97b2b882dd5f275b5dc4ad2377cb7ad74 /test/Transforms/LoopVectorize | |
parent | 102187efba7e084e7b2f59a435cb0b3583189d7b (diff) | |
download | llvm-8ffc96a163a9d4387feb06007d36ac1e419a8cd1.tar.gz llvm-8ffc96a163a9d4387feb06007d36ac1e419a8cd1.tar.bz2 llvm-8ffc96a163a9d4387feb06007d36ac1e419a8cd1.tar.xz |
Allow vectorization of intrinsics such as powi,cttz and ctlz in Loop and SLP Vectorizer.
This patch adds support to vectorize intrinsics such as powi, cttz and ctlz in Vectorizer. These intrinsics are different from other
intrinsics as second argument to these function must be same in order to vectorize them and it should be represented as a scalar.
Review: http://reviews.llvm.org/D3851#inline-32769 and http://reviews.llvm.org/D3937#inline-32857
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@209873 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/Transforms/LoopVectorize')
-rw-r--r-- | test/Transforms/LoopVectorize/intrinsic.ll | 102 |
1 files changed, 102 insertions, 0 deletions
diff --git a/test/Transforms/LoopVectorize/intrinsic.ll b/test/Transforms/LoopVectorize/intrinsic.ll index c3d570c03a..7dfaf03b0f 100644 --- a/test/Transforms/LoopVectorize/intrinsic.ll +++ b/test/Transforms/LoopVectorize/intrinsic.ll @@ -1090,3 +1090,105 @@ for.end: ; preds = %for.body ret void } +declare double @llvm.powi.f64(double %Val, i32 %power) nounwind readnone + +;CHECK-LABEL: @powi_f64( +;CHECK: llvm.powi.v4f64 +;CHECK: ret void +define void @powi_f64(i32 %n, double* noalias %y, double* noalias %x, i32 %P) nounwind uwtable { +entry: + %cmp9 = icmp sgt i32 %n, 0 + br i1 %cmp9, label %for.body, label %for.end + +for.body: ; preds = %entry, %for.body + %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] + %arrayidx = getelementptr inbounds double* %y, i64 %indvars.iv + %0 = load double* %arrayidx, align 8 + %call = tail call double @llvm.powi.f64(double %0, i32 %P) nounwind readnone + %arrayidx4 = getelementptr inbounds double* %x, i64 %indvars.iv + store double %call, double* %arrayidx4, align 8 + %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 + ret void +} + +;CHECK-LABEL: @powi_f64_neg( +;CHECK-NOT: llvm.powi.v4f64 +;CHECK: ret void +define void @powi_f64_neg(i32 %n, double* noalias %y, double* noalias %x) nounwind uwtable { +entry: + %cmp9 = icmp sgt i32 %n, 0 + br i1 %cmp9, label %for.body, label %for.end + +for.body: ; preds = %entry, %for.body + %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] + %arrayidx = getelementptr inbounds double* %y, i64 %indvars.iv + %0 = load double* %arrayidx, align 8 + %1 = trunc i64 %indvars.iv to i32 + %call = tail call double @llvm.powi.f64(double %0, i32 %1) nounwind readnone + %arrayidx4 = getelementptr inbounds double* %x, i64 %indvars.iv + store double %call, double* %arrayidx4, align 8 + %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 + ret void +} + +declare i64 @llvm.cttz.i64 (i64, i1) nounwind readnone + +;CHECK-LABEL: @cttz_f64( +;CHECK: llvm.cttz.v4i64 +;CHECK: ret void +define void @cttz_f64(i32 %n, i64* noalias %y, i64* noalias %x) nounwind uwtable { +entry: + %cmp9 = icmp sgt i32 %n, 0 + br i1 %cmp9, label %for.body, label %for.end + +for.body: ; preds = %entry, %for.body + %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] + %arrayidx = getelementptr inbounds i64* %y, i64 %indvars.iv + %0 = load i64* %arrayidx, align 8 + %call = tail call i64 @llvm.cttz.i64(i64 %0, i1 true) nounwind readnone + %arrayidx4 = getelementptr inbounds i64* %x, i64 %indvars.iv + store i64 %call, i64* %arrayidx4, align 8 + %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 + ret void +} + +declare i64 @llvm.ctlz.i64 (i64, i1) nounwind readnone + +;CHECK-LABEL: @ctlz_f64( +;CHECK: llvm.ctlz.v4i64 +;CHECK: ret void +define void @ctlz_f64(i32 %n, i64* noalias %y, i64* noalias %x) nounwind uwtable { +entry: + %cmp9 = icmp sgt i32 %n, 0 + br i1 %cmp9, label %for.body, label %for.end + +for.body: ; preds = %entry, %for.body + %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ] + %arrayidx = getelementptr inbounds i64* %y, i64 %indvars.iv + %0 = load i64* %arrayidx, align 8 + %call = tail call i64 @llvm.ctlz.i64(i64 %0, i1 true) nounwind readnone + %arrayidx4 = getelementptr inbounds i64* %x, i64 %indvars.iv + store i64 %call, i64* %arrayidx4, align 8 + %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 + ret void +} |