1 ; RUN: opt -mtriple=s390x-unknown-linux -mcpu=z13 -passes=loop-vectorize \
2 ; RUN: -force-vector-width=4 -debug-only=loop-vectorize \
3 ; RUN: -enable-interleaved-mem-accesses=false -disable-output < %s 2>&1 \
7 ; Check that a scalarized load does not get a zero cost in a vectorized
8 ; loop. It can only be folded into the add operand in the scalar loop.
10 define i32 @fun(ptr %data, i64 %n, i64 %s, ptr %Src) {
15 %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ]
16 %acc = phi i32 [ 0, %entry ], [ %acc_next, %for.body ]
17 %gep = getelementptr inbounds i32, ptr %Src, i64 %iv
18 %ld = load i32, ptr %gep
19 %acc_next = add i32 %acc, %ld
20 %iv.next = add nuw nsw i64 %iv, 2
21 %cmp110.us = icmp slt i64 %iv.next, %n
22 br i1 %cmp110.us, label %for.body, label %for.end
27 ; CHECK: Found an estimated cost of 4 for VF 4 For instruction: %ld = load i32, ptr %gep