[llvm] r191122 - LoopVectorizer: Only allow vectorization of intrinsics. We can't know for sure that the functions 'abs' or 'round' are the functions from libm.
Nadav Rotem
nrotem at apple.com
Fri Sep 20 17:27:05 PDT 2013
Author: nadav
Date: Fri Sep 20 19:27:05 2013
New Revision: 191122
URL: http://llvm.org/viewvc/llvm-project?rev=191122&view=rev
Log:
LoopVectorizer: Only allow vectorization of intrinsics. We can't know for sure that the functions 'abs' or 'round' are the functions from libm.
rdar://15012650
Modified:
llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp
llvm/trunk/test/Transforms/LoopVectorize/intrinsic.ll
Modified: llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp?rev=191122&r1=191121&r2=191122&view=diff
==============================================================================
--- llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp (original)
+++ llvm/trunk/lib/Transforms/Vectorize/LoopVectorize.cpp Fri Sep 20 19:27:05 2013
@@ -2925,9 +2925,18 @@ bool LoopVectorizationLegality::canVecto
// We still don't handle functions. However, we can ignore dbg intrinsic
// calls and we do handle certain intrinsic and libm functions.
CallInst *CI = dyn_cast<CallInst>(it);
- if (CI && !getIntrinsicIDForCall(CI, TLI) && !isa<DbgInfoIntrinsic>(CI)) {
+ if (CI) {
DEBUG(dbgs() << "LV: Found a call site.\n");
- return false;
+
+ if (!isa<IntrinsicInst>(it)) {
+ DEBUG(dbgs() << "LV: We only vectorize intrinsics.\n");
+ return false;
+ }
+
+ if (!getIntrinsicIDForCall(CI, TLI) && !isa<DbgInfoIntrinsic>(CI)) {
+ DEBUG(dbgs() << "LV: Found an unknown intrinsic.\n");
+ return false;
+ }
}
// Check that the instruction return type is vectorizable.
Modified: llvm/trunk/test/Transforms/LoopVectorize/intrinsic.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/Transforms/LoopVectorize/intrinsic.ll?rev=191122&r1=191121&r2=191122&view=diff
==============================================================================
--- llvm/trunk/test/Transforms/LoopVectorize/intrinsic.ll (original)
+++ llvm/trunk/test/Transforms/LoopVectorize/intrinsic.ll Fri Sep 20 19:27:05 2013
@@ -1018,7 +1018,7 @@ for.body:
%indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
%arrayidx = getelementptr inbounds float* %x, i64 %indvars.iv
%0 = load float* %arrayidx, align 4
- %call = tail call float @fabsf(float %0) nounwind readnone
+ %call = tail call float @llvm.fabs.f32(float %0) nounwind readnone
store float %call, float* %arrayidx, align 4
%indvars.iv.next = add i64 %indvars.iv, 1
%lftr.wideiv = trunc i64 %indvars.iv.next to i32
@@ -1029,6 +1029,31 @@ for.end:
ret void
}
-declare float @fabsf(float) nounwind readnone
-
declare double @llvm.pow.f64(double, double) nounwind readnone
+
+
+;CHECK: @not_intrin
+;CHECK: @round
+;CHECK-NOT: @round
+;CHECK: ret
+define void @not_intrin(i32* nocapture %A) nounwind ssp uwtable {
+ br label %1
+
+; <label>:1 ; preds = %1, %0
+ %indvars.iv = phi i64 [ 0, %0 ], [ %indvars.iv.next, %1 ]
+ %2 = getelementptr inbounds i32* %A, i64 %indvars.iv
+ %3 = load i32* %2, align 4
+ %4 = add nsw i32 %3, 3
+ store i32 %4, i32* %2, align 4
+ %5 = trunc i64 %indvars.iv to i32
+ tail call void @round(i32 %5) nounwind
+ %indvars.iv.next = add i64 %indvars.iv, 1
+ %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+ %exitcond = icmp eq i32 %lftr.wideiv, 256
+ br i1 %exitcond, label %6, label %1
+
+; <label>:6 ; preds = %1
+ ret void
+}
+
+declare void @round(i32)
More information about the llvm-commits
mailing list