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
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
|
; RUN: llc -march=amdgcn -mcpu=fiji -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=VI -check-prefix=VIPLUS %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -mattr=-flat-for-global -verify-machineinstrs < %s | FileCheck -check-prefix=GCN -check-prefix=GFX9 -check-prefix=VIPLUS %s
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_imax_sge_i16:
; VIPLUS: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
define amdgpu_kernel void @v_test_imax_sge_i16(i16 addrspace(1)* %out, i16 addrspace(1)* %aptr, i16 addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr i16, i16 addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr i16, i16 addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr i16, i16 addrspace(1)* %out, i32 %tid
%a = load i16, i16 addrspace(1)* %gep0, align 4
%b = load i16, i16 addrspace(1)* %gep1, align 4
%cmp = icmp sge i16 %a, %b
%val = select i1 %cmp, i16 %a, i16 %b
store i16 %val, i16 addrspace(1)* %outgep, align 4
ret void
}
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_imax_sge_v2i16:
; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; VI: v_max_i16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
; GFX9: v_pk_max_i16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
define amdgpu_kernel void @v_test_imax_sge_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %aptr, <2 x i16> addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr <2 x i16>, <2 x i16> addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr <2 x i16>, <2 x i16> addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr <2 x i16>, <2 x i16> addrspace(1)* %out, i32 %tid
%a = load <2 x i16>, <2 x i16> addrspace(1)* %gep0, align 4
%b = load <2 x i16>, <2 x i16> addrspace(1)* %gep1, align 4
%cmp = icmp sge <2 x i16> %a, %b
%val = select <2 x i1> %cmp, <2 x i16> %a, <2 x i16> %b
store <2 x i16> %val, <2 x i16> addrspace(1)* %outgep, align 4
ret void
}
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_imax_sge_v3i16:
; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; VI: v_max_i16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; VI-NOT: v_max_i16
; GFX9: v_pk_max_i16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; GFX9: v_pk_max_i16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
define amdgpu_kernel void @v_test_imax_sge_v3i16(<3 x i16> addrspace(1)* %out, <3 x i16> addrspace(1)* %aptr, <3 x i16> addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr <3 x i16>, <3 x i16> addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr <3 x i16>, <3 x i16> addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr <3 x i16>, <3 x i16> addrspace(1)* %out, i32 %tid
%a = load <3 x i16>, <3 x i16> addrspace(1)* %gep0, align 4
%b = load <3 x i16>, <3 x i16> addrspace(1)* %gep1, align 4
%cmp = icmp sge <3 x i16> %a, %b
%val = select <3 x i1> %cmp, <3 x i16> %a, <3 x i16> %b
store <3 x i16> %val, <3 x i16> addrspace(1)* %outgep, align 4
ret void
}
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_imax_sge_v4i16:
; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; VI: v_max_i16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
; VI: v_max_i16_e32 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; VI: v_max_i16_sdwa v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}} dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:WORD_1
; GFX9: v_pk_max_i16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
; GFX9: v_pk_max_i16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
define amdgpu_kernel void @v_test_imax_sge_v4i16(<4 x i16> addrspace(1)* %out, <4 x i16> addrspace(1)* %aptr, <4 x i16> addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr <4 x i16>, <4 x i16> addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr <4 x i16>, <4 x i16> addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr <4 x i16>, <4 x i16> addrspace(1)* %out, i32 %tid
%a = load <4 x i16>, <4 x i16> addrspace(1)* %gep0, align 4
%b = load <4 x i16>, <4 x i16> addrspace(1)* %gep1, align 4
%cmp = icmp sge <4 x i16> %a, %b
%val = select <4 x i1> %cmp, <4 x i16> %a, <4 x i16> %b
store <4 x i16> %val, <4 x i16> addrspace(1)* %outgep, align 4
ret void
}
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_imax_sgt_i16:
; VIPLUS: v_max_i16_e32
define amdgpu_kernel void @v_test_imax_sgt_i16(i16 addrspace(1)* %out, i16 addrspace(1)* %aptr, i16 addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr i16, i16 addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr i16, i16 addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr i16, i16 addrspace(1)* %out, i32 %tid
%a = load i16, i16 addrspace(1)* %gep0, align 4
%b = load i16, i16 addrspace(1)* %gep1, align 4
%cmp = icmp sgt i16 %a, %b
%val = select i1 %cmp, i16 %a, i16 %b
store i16 %val, i16 addrspace(1)* %outgep, align 4
ret void
}
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_umax_uge_i16:
; VIPLUS: v_max_u16_e32
define amdgpu_kernel void @v_test_umax_uge_i16(i16 addrspace(1)* %out, i16 addrspace(1)* %aptr, i16 addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr i16, i16 addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr i16, i16 addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr i16, i16 addrspace(1)* %out, i32 %tid
%a = load i16, i16 addrspace(1)* %gep0, align 4
%b = load i16, i16 addrspace(1)* %gep1, align 4
%cmp = icmp uge i16 %a, %b
%val = select i1 %cmp, i16 %a, i16 %b
store i16 %val, i16 addrspace(1)* %outgep, align 4
ret void
}
; FIXME: Need to handle non-uniform case for function below (load without gep).
; GCN-LABEL: {{^}}v_test_umax_ugt_i16:
; VIPLUS: v_max_u16_e32
define amdgpu_kernel void @v_test_umax_ugt_i16(i16 addrspace(1)* %out, i16 addrspace(1)* %aptr, i16 addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr i16, i16 addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr i16, i16 addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr i16, i16 addrspace(1)* %out, i32 %tid
%a = load i16, i16 addrspace(1)* %gep0, align 4
%b = load i16, i16 addrspace(1)* %gep1, align 4
%cmp = icmp ugt i16 %a, %b
%val = select i1 %cmp, i16 %a, i16 %b
store i16 %val, i16 addrspace(1)* %outgep, align 4
ret void
}
; GCN-LABEL: {{^}}v_test_umax_ugt_v2i16:
; VI: v_max_u16_e32
; VI: v_max_u16_sdwa
; GFX9: v_pk_max_u16 v{{[0-9]+}}, v{{[0-9]+}}, v{{[0-9]+}}
define amdgpu_kernel void @v_test_umax_ugt_v2i16(<2 x i16> addrspace(1)* %out, <2 x i16> addrspace(1)* %aptr, <2 x i16> addrspace(1)* %bptr) nounwind {
%tid = call i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
%gep0 = getelementptr <2 x i16>, <2 x i16> addrspace(1)* %aptr, i32 %tid
%gep1 = getelementptr <2 x i16>, <2 x i16> addrspace(1)* %bptr, i32 %tid
%outgep = getelementptr <2 x i16>, <2 x i16> addrspace(1)* %out, i32 %tid
%a = load <2 x i16>, <2 x i16> addrspace(1)* %gep0, align 4
%b = load <2 x i16>, <2 x i16> addrspace(1)* %gep1, align 4
%cmp = icmp ugt <2 x i16> %a, %b
%val = select <2 x i1> %cmp, <2 x i16> %a, <2 x i16> %b
store <2 x i16> %val, <2 x i16> addrspace(1)* %outgep, align 4
ret void
}
declare i32 @llvm.amdgcn.workitem.id.x() nounwind readnone
|
