2 ; RUN: opt -mtriple=s390x-unknown-linux -mcpu=z13 -loop-vectorize \
3 ; RUN: -force-vector-width=4 -debug-only=loop-vectorize \
4 ; RUN: -disable-output -enable-interleaved-mem-accesses=false < %s 2>&1 | \
7 ; Check that a scalarized load/store does not get a cost for insterts/
8 ; extracts, since z13 supports element load/store.
10 define void @fun(i32* %data, i64 %n) {
15 %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ]
16 %tmp0 = getelementptr inbounds i32, i32* %data, i64 %i
17 %tmp1 = load i32, i32* %tmp0, align 4
18 %tmp2 = add i32 %tmp1, 1
19 store i32 %tmp2, i32* %tmp0, align 4
20 %i.next = add nuw nsw i64 %i, 2
21 %cond = icmp slt i64 %i.next, %n
22 br i1 %cond, label %for.body, label %for.end
27 ; CHECK: LV: Scalarizing: %tmp1 = load i32, i32* %tmp0, align 4
28 ; CHECK: LV: Scalarizing: store i32 %tmp2, i32* %tmp0, align 4
30 ; CHECK: LV: Found an estimated cost of 4 for VF 4 For instruction: %tmp1 = load i32, i32* %tmp0, align 4
31 ; CHECK: LV: Found an estimated cost of 4 for VF 4 For instruction: store i32 %tmp2, i32* %tmp0, align 4