1 ; RUN: opt < %s -loop-vectorize -force-vector-width=4 -force-vector-interleave=1 -instcombine -S | FileCheck %s
3 target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
6 define void @foo(float* noalias %a, float* noalias %b, float* noalias %c, i32 %N) {
8 %cmp.4 = icmp sgt i32 %N, 0
9 br i1 %cmp.4, label %for.body.preheader, label %for.end
11 for.body.preheader: ; preds = %entry
14 for.body: ; preds = %for.body.preheader, %for.body
15 %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %for.body.preheader ]
17 ; Check that we don't lose !nontemporal hint when attempting vectorizing of loads.
18 ; CHECK: load {{.*}} align 4, !nontemporal !0
19 %arrayidx = getelementptr inbounds float, float* %b, i64 %indvars.iv
20 %0 = load float, float* %arrayidx, align 4, !nontemporal !0
22 ; Check that we don't introduce !nontemporal hint when the original scalar loads didn't have it.
23 ; CHECK: load {{.*}} align 4{{$}}
24 %arrayidx2 = getelementptr inbounds float, float* %c, i64 %indvars.iv
25 %1 = load float, float* %arrayidx2, align 4
26 %add = fadd float %0, %1
28 ; Check that we don't lose !nontemporal hint when attempting vectorizing of stores.
29 ; CHECK: store {{.*}} align 4, !nontemporal !0
30 %arrayidx4 = getelementptr inbounds float, float* %a, i64 %indvars.iv
31 store float %add, float* %arrayidx4, align 4, !nontemporal !0
33 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
34 %lftr.wideiv = trunc i64 %indvars.iv.next to i32
35 %exitcond = icmp eq i32 %lftr.wideiv, %N
36 br i1 %exitcond, label %for.end.loopexit, label %for.body
38 for.end.loopexit: ; preds = %for.body
41 for.end: ; preds = %for.end.loopexit, %entry