xref: /llvm-project/llvm/test/CodeGen/AMDGPU/promote-alloca-no-opts.ll (revision 4490003a22658dcd12527029b2c8682b63d8a9d6)
1; RUN: llc -O0 -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -mattr=+promote-alloca < %s | FileCheck -check-prefix=NOOPTS -check-prefix=ALL %s
2; RUN: llc -O1 -mtriple=amdgcn-unknown-amdhsa -mcpu=kaveri -mattr=+promote-alloca < %s | FileCheck -check-prefix=OPTS -check-prefix=ALL %s
3
4; ALL-LABEL: {{^}}promote_alloca_i32_array_array:
5; NOOPTS: .amdhsa_group_segment_fixed_size 0
6; NOOPTS-NOT: ds_write
7; OPTS: ds_write
8define amdgpu_kernel void @promote_alloca_i32_array_array(ptr addrspace(1) %out, i32 %index) #0 {
9entry:
10  %alloca = alloca [2 x [2 x i32]], addrspace(5)
11  %gep1 = getelementptr inbounds [2 x [2 x i32]], ptr addrspace(5) %alloca, i32 0, i32 0, i32 1
12  store i32 0, ptr addrspace(5) %alloca
13  store i32 1, ptr addrspace(5) %gep1
14  %gep2 = getelementptr inbounds [2 x [2 x i32]], ptr addrspace(5) %alloca, i32 0, i32 0, i32 %index
15  %load = load i32, ptr addrspace(5) %gep2
16  store i32 %load, ptr addrspace(1) %out
17  ret void
18}
19
20; ALL-LABEL: {{^}}optnone_promote_alloca_i32_array_array:
21; ALL: .amdhsa_group_segment_fixed_size 0
22; ALL-NOT: ds_write
23define amdgpu_kernel void @optnone_promote_alloca_i32_array_array(ptr addrspace(1) %out, i32 %index) #1 {
24entry:
25  %alloca = alloca [2 x [2 x i32]], addrspace(5)
26  %gep1 = getelementptr inbounds [2 x [2 x i32]], ptr addrspace(5) %alloca, i32 0, i32 0, i32 1
27  store i32 0, ptr addrspace(5) %alloca
28  store i32 1, ptr addrspace(5) %gep1
29  %gep2 = getelementptr inbounds [2 x [2 x i32]], ptr addrspace(5) %alloca, i32 0, i32 0, i32 %index
30  %load = load i32, ptr addrspace(5) %gep2
31  store i32 %load, ptr addrspace(1) %out
32  ret void
33}
34
35attributes #0 = { nounwind "amdgpu-flat-work-group-size"="64,64" }
36attributes #1 = { nounwind optnone noinline "amdgpu-flat-work-group-size"="64,64" }
37
38!llvm.module.flags = !{!0}
39!0 = !{i32 1, !"amdhsa_code_object_version", i32 400}
40