2 ; RUN: opt -S -loop-vectorize -debug-only=loop-vectorize -mcpu=core-avx2 %s 2>&1 | FileCheck %s
3 target datalayout = "e-m:e-p:32:32-f64:32:64-f80:32-n8:16:32-S128"
4 target triple = "i386-unknown-linux-gnu"
6 @i64src = common local_unnamed_addr global [120 x i64] zeroinitializer, align 4
7 @i64dst = common local_unnamed_addr global [120 x i64] zeroinitializer, align 4
9 ; Function Attrs: norecurse nounwind
10 define void @stride2i64(i64 %k, i32 %width_) {
13 ; CHECK: Found an estimated cost of 8 for VF 4 For instruction: %0 = load i64
14 ; CHECK: Found an estimated cost of 8 for VF 4 For instruction: store i64
16 %cmp27 = icmp sgt i32 %width_, 0
17 br i1 %cmp27, label %for.body.lr.ph, label %for.cond.cleanup
19 for.body.lr.ph: ; preds = %entry
22 for.cond.cleanup: ; preds = %for.body, %entry
26 %i.028 = phi i32 [ 0, %for.body.lr.ph ], [ %add16, %for.body ]
27 %arrayidx = getelementptr inbounds [120 x i64], [120 x i64]* @i64src, i32 0, i32 %i.028
28 %0 = load i64, i64* %arrayidx, align 4
29 %arrayidx2 = getelementptr inbounds [120 x i64], [120 x i64]* @i64dst, i32 0, i32 %i.028
30 store i64 %0, i64* %arrayidx2, align 4
31 %add4 = add nuw nsw i32 %i.028, 1
32 %arrayidx5 = getelementptr inbounds [120 x i64], [120 x i64]* @i64src, i32 0, i32 %add4
33 %1 = load i64, i64* %arrayidx5, align 4
34 %arrayidx8 = getelementptr inbounds [120 x i64], [120 x i64]* @i64dst, i32 0, i32 %add4
35 store i64 %1, i64* %arrayidx8, align 4
36 %add16 = add nuw nsw i32 %i.028, 2
37 %cmp = icmp slt i32 %add16, %width_
38 br i1 %cmp, label %for.body, label %for.cond.cleanup