1 ; RUN: llc -march=amdgcn < %s | FileCheck %s
2 ; RUN: opt -S -mtriple=amdgcn-- -amdgpu-lower-intrinsics < %s | FileCheck -check-prefix=OPT %s
6 ; OPT-LABEL: @zext_grp_size_128
7 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.x(), !range !0
8 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.y(), !range !0
9 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.z(), !range !0
10 define amdgpu_kernel void @zext_grp_size_128(i32 addrspace(1)* nocapture %arg) #0 {
12 %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
13 %tmp1 = and i32 %tmp, 127
14 store i32 %tmp1, i32 addrspace(1)* %arg, align 4
15 %tmp2 = tail call i32 @llvm.amdgcn.workitem.id.y()
16 %tmp3 = and i32 %tmp2, 127
17 %tmp4 = getelementptr inbounds i32, i32 addrspace(1)* %arg, i64 1
18 store i32 %tmp3, i32 addrspace(1)* %tmp4, align 4
19 %tmp5 = tail call i32 @llvm.amdgcn.workitem.id.z()
20 %tmp6 = and i32 %tmp5, 127
21 %tmp7 = getelementptr inbounds i32, i32 addrspace(1)* %arg, i64 2
22 store i32 %tmp6, i32 addrspace(1)* %tmp7, align 4
26 ; OPT-LABEL: @zext_grp_size_32x4x1
27 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.x(), !range !2
28 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.y(), !range !3
29 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.z(), !range !4
30 define amdgpu_kernel void @zext_grp_size_32x4x1(i32 addrspace(1)* nocapture %arg) #0 !reqd_work_group_size !0 {
32 %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
33 %tmp1 = and i32 %tmp, 31
34 store i32 %tmp1, i32 addrspace(1)* %arg, align 4
35 %tmp2 = tail call i32 @llvm.amdgcn.workitem.id.y()
36 %tmp3 = and i32 %tmp2, 3
37 %tmp4 = getelementptr inbounds i32, i32 addrspace(1)* %arg, i64 1
38 store i32 %tmp3, i32 addrspace(1)* %tmp4, align 4
39 %tmp5 = tail call i32 @llvm.amdgcn.workitem.id.z()
40 %tmp6 = and i32 %tmp5, 1
41 %tmp7 = getelementptr inbounds i32, i32 addrspace(1)* %arg, i64 2
42 store i32 %tmp6, i32 addrspace(1)* %tmp7, align 4
46 ; OPT-LABEL: @zext_grp_size_512
47 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.x(), !range !5
48 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.y(), !range !5
49 ; OPT: tail call i32 @llvm.amdgcn.workitem.id.z(), !range !5
50 define amdgpu_kernel void @zext_grp_size_512(i32 addrspace(1)* nocapture %arg) #1 {
52 %tmp = tail call i32 @llvm.amdgcn.workitem.id.x()
53 %tmp1 = and i32 %tmp, 65535
54 store i32 %tmp1, i32 addrspace(1)* %arg, align 4
55 %tmp2 = tail call i32 @llvm.amdgcn.workitem.id.y()
56 %tmp3 = and i32 %tmp2, 65535
57 %tmp4 = getelementptr inbounds i32, i32 addrspace(1)* %arg, i64 1
58 store i32 %tmp3, i32 addrspace(1)* %tmp4, align 4
59 %tmp5 = tail call i32 @llvm.amdgcn.workitem.id.z()
60 %tmp6 = and i32 %tmp5, 65535
61 %tmp7 = getelementptr inbounds i32, i32 addrspace(1)* %arg, i64 2
62 store i32 %tmp6, i32 addrspace(1)* %tmp7, align 4
66 declare i32 @llvm.amdgcn.workitem.id.x() #2
68 declare i32 @llvm.amdgcn.workitem.id.y() #2
70 declare i32 @llvm.amdgcn.workitem.id.z() #2
72 attributes #0 = { nounwind "amdgpu-flat-work-group-size"="64,128" }
73 attributes #1 = { nounwind "amdgpu-flat-work-group-size"="512,512" }
74 attributes #2 = { nounwind readnone speculatable }
75 attributes #3 = { nounwind readnone }
77 !0 = !{i32 32, i32 4, i32 1}
79 ; OPT: !0 = !{i32 0, i32 128}
80 ; OPT: !1 = !{i32 32, i32 4, i32 1}
81 ; OPT: !2 = !{i32 0, i32 32}
82 ; OPT: !3 = !{i32 0, i32 4}
83 ; OPT: !4 = !{i32 0, i32 1}
84 ; OPT: !5 = !{i32 0, i32 512}