• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1; RUN: opt -mtriple=amdgcn-amd-amdhsa -basicaa -load-store-vectorizer -S -o - %s | FileCheck %s
2
3target datalayout = "e-p:32:32-p1:64:64-p2:64:64-p3:32:32-p4:64:64-p5:32:32-p24:64:64-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64"
4
5; Check relative position of the inserted vector load relative to the existing
6; adds. Vectorized loads should be inserted at the position of the first load.
7
8; CHECK-LABEL: @insert_load_point(
9; CHECK: %z = add i32 %x, 4
10; CHECK: load <2 x float>
11; CHECK: %w = add i32 %y, 9
12; CHECK: %foo = add i32 %z, %w
13define void @insert_load_point(float addrspace(1)* nocapture %a, float addrspace(1)* nocapture %b, float addrspace(1)* nocapture readonly %c, i64 %idx, i32 %x, i32 %y) #0 {
14entry:
15  %a.idx.x = getelementptr inbounds float, float addrspace(1)* %a, i64 %idx
16  %c.idx.x = getelementptr inbounds float, float addrspace(1)* %c, i64 %idx
17  %a.idx.x.1 = getelementptr inbounds float, float addrspace(1)* %a.idx.x, i64 1
18  %c.idx.x.1 = getelementptr inbounds float, float addrspace(1)* %c.idx.x, i64 1
19
20  %z = add i32 %x, 4
21  %ld.c = load float, float addrspace(1)* %c.idx.x, align 4
22  %w = add i32 %y, 9
23  %ld.c.idx.1 = load float, float addrspace(1)* %c.idx.x.1, align 4
24  %foo = add i32 %z, %w
25
26  store float 0.0, float addrspace(1)* %a.idx.x, align 4
27  store float 0.0, float addrspace(1)* %a.idx.x.1, align 4
28
29  %add = fadd float %ld.c, %ld.c.idx.1
30  store float %add, float addrspace(1)* %b, align 4
31  store i32 %foo, i32 addrspace(3)* null, align 4
32  ret void
33}
34
35; CHECK-LABEL: @insert_store_point(
36; CHECK: %z = add i32 %x, 4
37; CHECK: %w = add i32 %y, 9
38; CHECK: store <2 x float>
39; CHECK: %foo = add i32 %z, %w
40define void @insert_store_point(float addrspace(1)* nocapture %a, float addrspace(1)* nocapture %b, float addrspace(1)* nocapture readonly %c, i64 %idx, i32 %x, i32 %y) #0 {
41entry:
42  %a.idx.x = getelementptr inbounds float, float addrspace(1)* %a, i64 %idx
43  %c.idx.x = getelementptr inbounds float, float addrspace(1)* %c, i64 %idx
44  %a.idx.x.1 = getelementptr inbounds float, float addrspace(1)* %a.idx.x, i64 1
45  %c.idx.x.1 = getelementptr inbounds float, float addrspace(1)* %c.idx.x, i64 1
46
47  %ld.c = load float, float addrspace(1)* %c.idx.x, align 4
48  %ld.c.idx.1 = load float, float addrspace(1)* %c.idx.x.1, align 4
49
50  %z = add i32 %x, 4
51  store float 0.0, float addrspace(1)* %a.idx.x, align 4
52  %w = add i32 %y, 9
53  store float 0.0, float addrspace(1)* %a.idx.x.1, align 4
54  %foo = add i32 %z, %w
55
56  %add = fadd float %ld.c, %ld.c.idx.1
57  store float %add, float addrspace(1)* %b, align 4
58  store i32 %foo, i32 addrspace(3)* null, align 4
59  ret void
60}
61
62attributes #0 = { nounwind }
63