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
|
; RUN: llc < %s -march=amdgcn -mcpu=gfx908 -verify-machineinstrs | FileCheck %s -check-prefix=GCN
; RUN: llc < %s -march=amdgcn -mcpu=gfx90a -verify-machineinstrs | FileCheck %s -check-prefix=GCN
declare float @llvm.amdgcn.buffer.atomic.fadd.f32(float, <4 x i32>, i32, i32, i1)
declare <2 x half> @llvm.amdgcn.buffer.atomic.fadd.v2f16(<2 x half>, <4 x i32>, i32, i32, i1)
declare float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1), float)
declare <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1), <2 x half>)
declare float @llvm.amdgcn.flat.atomic.fadd.f32.p0.f32(ptr, float)
; GCN-LABEL: {{^}}buffer_atomic_add_f32:
; GCN: buffer_atomic_add_f32 v0, v1, s[0:3], 0 idxen
define amdgpu_ps void @buffer_atomic_add_f32(<4 x i32> inreg %rsrc, float %data, i32 %vindex) {
main_body:
%ret = call float @llvm.amdgcn.buffer.atomic.fadd.f32(float %data, <4 x i32> %rsrc, i32 %vindex, i32 0, i1 0)
ret void
}
; GCN-LABEL: {{^}}buffer_atomic_add_f32_off4_slc:
; GCN: buffer_atomic_add_f32 v0, v1, s[0:3], 0 idxen offset:4 slc
define amdgpu_ps void @buffer_atomic_add_f32_off4_slc(<4 x i32> inreg %rsrc, float %data, i32 %vindex) {
main_body:
%ret = call float @llvm.amdgcn.buffer.atomic.fadd.f32(float %data, <4 x i32> %rsrc, i32 %vindex, i32 4, i1 1)
ret void
}
; GCN-LABEL: {{^}}buffer_atomic_pk_add_v2f16:
; GCN: buffer_atomic_pk_add_f16 v0, v1, s[0:3], 0 idxen
define amdgpu_ps void @buffer_atomic_pk_add_v2f16(<4 x i32> inreg %rsrc, <2 x half> %data, i32 %vindex) {
main_body:
%ret = call <2 x half> @llvm.amdgcn.buffer.atomic.fadd.v2f16(<2 x half> %data, <4 x i32> %rsrc, i32 %vindex, i32 0, i1 0)
ret void
}
; GCN-LABEL: {{^}}buffer_atomic_pk_add_v2f16_off4_slc:
; GCN: buffer_atomic_pk_add_f16 v0, v1, s[0:3], 0 idxen offset:4 slc
define amdgpu_ps void @buffer_atomic_pk_add_v2f16_off4_slc(<4 x i32> inreg %rsrc, <2 x half> %data, i32 %vindex) {
main_body:
%ret = call <2 x half> @llvm.amdgcn.buffer.atomic.fadd.v2f16(<2 x half> %data, <4 x i32> %rsrc, i32 %vindex, i32 4, i1 1)
ret void
}
; GCN-LABEL: {{^}}global_atomic_add_f32:
; GCN: global_atomic_add_f32 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}}
define amdgpu_kernel void @global_atomic_add_f32(ptr addrspace(1) %ptr, float %data) {
main_body:
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %ptr, float %data)
ret void
}
; GCN-LABEL: {{^}}global_atomic_add_f32_off4:
; GCN: global_atomic_add_f32 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}} offset:4
define amdgpu_kernel void @global_atomic_add_f32_off4(ptr addrspace(1) %ptr, float %data) {
main_body:
%p = getelementptr float, ptr addrspace(1) %ptr, i64 1
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %p, float %data)
ret void
}
; GCN-LABEL: {{^}}global_atomic_add_f32_offneg4:
; GCN: global_atomic_add_f32 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}} offset:-4
define amdgpu_kernel void @global_atomic_add_f32_offneg4(ptr addrspace(1) %ptr, float %data) {
main_body:
%p = getelementptr float, ptr addrspace(1) %ptr, i64 -1
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %p, float %data)
ret void
}
; GCN-LABEL: {{^}}global_atomic_pk_add_v2f16:
; GCN: global_atomic_pk_add_f16 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]$}}
define amdgpu_kernel void @global_atomic_pk_add_v2f16(ptr addrspace(1) %ptr, <2 x half> %data) {
main_body:
%ret = call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %ptr, <2 x half> %data)
ret void
}
; GCN-LABEL: {{^}}global_atomic_pk_add_v2f16_off4:
; GCN: global_atomic_pk_add_f16 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}} offset:4
define amdgpu_kernel void @global_atomic_pk_add_v2f16_off4(ptr addrspace(1) %ptr, <2 x half> %data) {
main_body:
%p = getelementptr <2 x half>, ptr addrspace(1) %ptr, i64 1
%ret = call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %p, <2 x half> %data)
ret void
}
; GCN-LABEL: {{^}}global_atomic_pk_add_v2f16_offneg4:
; GCN: global_atomic_pk_add_f16 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]}} offset:-4{{$}}
define amdgpu_kernel void @global_atomic_pk_add_v2f16_offneg4(ptr addrspace(1) %ptr, <2 x half> %data) {
main_body:
%p = getelementptr <2 x half>, ptr addrspace(1) %ptr, i64 -1
%ret = call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %p, <2 x half> %data)
ret void
}
; Make sure this artificially selects with an incorrect subtarget, but
; the feature set.
; GCN-LABEL: {{^}}global_atomic_fadd_f32_wrong_subtarget:
; GCN: global_atomic_add_f32 v{{[0-9]+}}, v{{[0-9]+}}, s{{\[[0-9]+:[0-9]+\]$}}
define amdgpu_kernel void @global_atomic_fadd_f32_wrong_subtarget(ptr addrspace(1) %ptr, float %data) #0 {
%ret = call float @llvm.amdgcn.global.atomic.fadd.f32.p1.f32(ptr addrspace(1) %ptr, float %data)
ret void
}
; GCN-LABEL: {{^}}flat_atomic_fadd_f32_wrong_subtarget:
; GCN: flat_atomic_add_f32 v{{\[[0-9]+:[0-9]+\]}}, v{{[0-9]+}}
define amdgpu_kernel void @flat_atomic_fadd_f32_wrong_subtarget(ptr %ptr, float %data) #1 {
%ret = call float @llvm.amdgcn.flat.atomic.fadd.f32.p0.f32(ptr %ptr, float %data)
ret void
}
attributes #0 = { "target-cpu"="gfx803" "target-features"="+atomic-fadd-no-rtn-insts"}
attributes #1 = { "target-cpu"="gfx803" "target-features"="+flat-atomic-fadd-f32-inst"}
|
