1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
|
; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,CI %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefixes=GCN,GFX9 %s
declare i32 @llvm.amdgcn.workitem.id.x() #0
@lds.obj = addrspace(3) global [256 x i32] undef, align 4
; GCN-LABEL: {{^}}write_ds_sub0_offset0_global:
; GCN: v_lshlrev_b32_e32 [[SHL:v[0-9]+]], 2, v0
; CI: v_sub_i32_e32 [[BASEPTR:v[0-9]+]], vcc, 0, [[SHL]]
; GFX9: v_sub_u32_e32 [[BASEPTR:v[0-9]+]], 0, [[SHL]]
; GCN: v_mov_b32_e32 [[VAL:v[0-9]+]], 0x7b
; GCN: ds_write_b32 [[BASEPTR]], [[VAL]] offset:12
define amdgpu_kernel void @write_ds_sub0_offset0_global() #0 {
entry:
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #1
%sub1 = sub i32 0, %x.i
%tmp0 = getelementptr [256 x i32], [256 x i32] addrspace(3)* @lds.obj, i32 0, i32 %sub1
%arrayidx = getelementptr inbounds i32, i32 addrspace(3)* %tmp0, i32 3
store i32 123, i32 addrspace(3)* %arrayidx
ret void
}
; GCN-LABEL: {{^}}add_x_shl_neg_to_sub_max_offset:
; GCN-DAG: v_lshlrev_b32_e32 [[SCALED:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[NEG:v[0-9]+]], vcc, 0, [[SCALED]]
; GFX9-DAG: v_sub_u32_e32 [[NEG:v[0-9]+]], 0, [[SCALED]]
; GCN-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 13
; GCN: ds_write_b8 [[NEG]], [[K]] offset:65535
define amdgpu_kernel void @add_x_shl_neg_to_sub_max_offset() #1 {
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #0
%neg = sub i32 0, %x.i
%shl = shl i32 %neg, 2
%add = add i32 65535, %shl
%ptr = inttoptr i32 %add to i8 addrspace(3)*
store i8 13, i8 addrspace(3)* %ptr
ret void
}
; GCN-LABEL: {{^}}add_x_shl_neg_to_sub_max_offset_p1:
; GCN-DAG: v_lshlrev_b32_e32 [[SCALED:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[NEG:v[0-9]+]], vcc, 0x10000, [[SCALED]]
; GFX9-DAG: v_sub_u32_e32 [[NEG:v[0-9]+]], 0x10000, [[SCALED]]
; GCN-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 13
; GCN: ds_write_b8 [[NEG]], [[K]]{{$}}
define amdgpu_kernel void @add_x_shl_neg_to_sub_max_offset_p1() #1 {
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #0
%neg = sub i32 0, %x.i
%shl = shl i32 %neg, 2
%add = add i32 65536, %shl
%ptr = inttoptr i32 %add to i8 addrspace(3)*
store i8 13, i8 addrspace(3)* %ptr
ret void
}
; GCN-LABEL: {{^}}add_x_shl_neg_to_sub_multi_use:
; GCN-DAG: v_lshlrev_b32_e32 [[SCALED:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[NEG:v[0-9]+]], vcc, 0, [[SCALED]]
; GFX9-DAG: v_sub_u32_e32 [[NEG:v[0-9]+]], 0, [[SCALED]]
; GCN-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 13
; GCN-NOT: v_sub
; GCN: ds_write_b32 [[NEG]], [[K]] offset:123{{$}}
; GCN-NOT: v_sub
; GCN: ds_write_b32 [[NEG]], [[K]] offset:456{{$}}
; GCN: s_endpgm
define amdgpu_kernel void @add_x_shl_neg_to_sub_multi_use() #1 {
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #0
%neg = sub i32 0, %x.i
%shl = shl i32 %neg, 2
%add0 = add i32 123, %shl
%add1 = add i32 456, %shl
%ptr0 = inttoptr i32 %add0 to i32 addrspace(3)*
store volatile i32 13, i32 addrspace(3)* %ptr0
%ptr1 = inttoptr i32 %add1 to i32 addrspace(3)*
store volatile i32 13, i32 addrspace(3)* %ptr1
ret void
}
; GCN-LABEL: {{^}}add_x_shl_neg_to_sub_multi_use_same_offset:
; GCN-DAG: v_lshlrev_b32_e32 [[SCALED:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[NEG:v[0-9]+]], vcc, 0, [[SCALED]]
; GFX9-DAG: v_sub_u32_e32 [[NEG:v[0-9]+]], 0, [[SCALED]]
; GCN-DAG: v_mov_b32_e32 [[K:v[0-9]+]], 13
; GCN-NOT: v_sub
; GCN: ds_write_b32 [[NEG]], [[K]] offset:123{{$}}
; GCN-NOT: v_sub
; GCN: ds_write_b32 [[NEG]], [[K]] offset:123{{$}}
; GCN: s_endpgm
define amdgpu_kernel void @add_x_shl_neg_to_sub_multi_use_same_offset() #1 {
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #0
%neg = sub i32 0, %x.i
%shl = shl i32 %neg, 2
%add = add i32 123, %shl
%ptr = inttoptr i32 %add to i32 addrspace(3)*
store volatile i32 13, i32 addrspace(3)* %ptr
store volatile i32 13, i32 addrspace(3)* %ptr
ret void
}
; GCN-LABEL: {{^}}add_x_shl_neg_to_sub_misaligned_i64_max_offset:
; GCN-DAG: v_lshlrev_b32_e32 [[SCALED:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[NEG:v[0-9]+]], vcc, 0, [[SCALED]]
; GFX9-DAG: v_sub_u32_e32 [[NEG:v[0-9]+]], 0, [[SCALED]]
; GCN: ds_write2_b32 [[NEG]], {{v[0-9]+}}, {{v[0-9]+}} offset0:254 offset1:255
define amdgpu_kernel void @add_x_shl_neg_to_sub_misaligned_i64_max_offset() #1 {
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #0
%neg = sub i32 0, %x.i
%shl = shl i32 %neg, 2
%add = add i32 1019, %shl
%ptr = inttoptr i32 %add to i64 addrspace(3)*
store i64 123, i64 addrspace(3)* %ptr, align 4
ret void
}
; GCN-LABEL: {{^}}add_x_shl_neg_to_sub_misaligned_i64_max_offset_p1:
; GCN-DAG: v_lshlrev_b32_e32 [[SCALED:v[0-9]+]], 2, v0
; CI-DAG: v_sub_i32_e32 [[NEG:v[0-9]+]], vcc, 0x3fc, [[SCALED]]
; GFX9-DAG: v_sub_u32_e32 [[NEG:v[0-9]+]], 0x3fc, [[SCALED]]
; GCN: ds_write2_b32 [[NEG]], {{v[0-9]+}}, {{v[0-9]+}} offset1:1{{$}}
define amdgpu_kernel void @add_x_shl_neg_to_sub_misaligned_i64_max_offset_p1() #1 {
%x.i = call i32 @llvm.amdgcn.workitem.id.x() #0
%neg = sub i32 0, %x.i
%shl = shl i32 %neg, 2
%add = add i32 1020, %shl
%ptr = inttoptr i32 %add to i64 addrspace(3)*
store i64 123, i64 addrspace(3)* %ptr, align 4
ret void
}
attributes #0 = { nounwind readnone }
attributes #1 = { nounwind }
attributes #2 = { nounwind convergent }
|
