2 ; RUN: opt < %s -passes=loop-vectorize -force-vector-interleave=1 -S -debug-only=loop-vectorize 2>&1 | FileCheck %s
4 target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
5 target triple = "aarch64--linux-gnu"
7 ; CHECK-LABEL: all_scalar
8 ; CHECK: LV: Found scalar instruction: %i.next = add nuw nsw i64 %i, 2
9 ; CHECK: LV: Found an estimated cost of 1 for VF 2 For instruction: %i.next = add nuw nsw i64 %i, 2
10 ; CHECK: LV: Not considering vector loop of width 2 because it will not generate any vector instructions
12 define void @all_scalar(ptr %a, i64 %n) {
17 %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ]
18 %tmp0 = getelementptr i64, ptr %a, i64 %i
19 store i64 0, ptr %tmp0, align 1
20 %i.next = add nuw nsw i64 %i, 2
21 %cond = icmp eq i64 %i.next, %n
22 br i1 %cond, label %for.end, label %for.body
28 ; CHECK-LABEL: PR33193
29 ; CHECK: LV: Found scalar instruction: %i.next = zext i32 %j.next to i64
30 ; CHECK: LV: Found an estimated cost of 0 for VF 8 For instruction: %i.next = zext i32 %j.next to i64
31 ; CHECK: LV: Not considering vector loop of width 8 because it will not generate any vector instructions
32 %struct.a = type { i32, i8 }
33 define void @PR33193(ptr %a, i64 %n) {
38 %i = phi i64 [ 0, %entry ], [ %i.next, %for.body ]
39 %j = phi i32 [ 0, %entry ], [ %j.next, %for.body ]
40 %tmp0 = getelementptr inbounds %struct.a, ptr %a, i64 %i, i32 1
41 store i8 0, ptr %tmp0, align 4
42 %j.next = add i32 %j, 1
43 %i.next = zext i32 %j.next to i64
44 %cond = icmp ugt i64 %n, %i.next
45 br i1 %cond, label %for.body, label %for.end