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
150
151
152
153
154
155
156
157
158
159
160
161
162
163
| ; RUN: llc -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=SI %s
; RUN: llc -march=amdgcn -mcpu=bonaire -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=SI %s
; RUN: llc -march=amdgcn -mcpu=fiji -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=VI %s
; RUN: llc -march=amdgcn -mcpu=gfx900 -verify-machineinstrs < %s | FileCheck -enable-var-scope -check-prefix=GCN -check-prefix=VI %s
; GCN-LABEL: {{^}}s_cvt_pknorm_u16_f32:
; GCN-DAG: s_load_dwordx2 s{{\[}}[[SX:[0-9]+]]:[[SY:[0-9]+]]{{\]}}, s[0:1], 0x{{b|2c}}
; GCN: v_mov_b32_e32 [[VY:v[0-9]+]], s[[SY]]
; SI: v_cvt_pknorm_u16_f32_e32 v{{[0-9]+}}, s[[SX]], [[VY]]
; VI: v_cvt_pknorm_u16_f32 v{{[0-9]+}}, s[[SX]], [[VY]]
define amdgpu_kernel void @s_cvt_pknorm_u16_f32(i32 addrspace(1)* %out, float %x, float %y) #0 {
%result = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %x, float %y)
%r = bitcast <2 x i16> %result to i32
store i32 %r, i32 addrspace(1)* %out
ret void
}
; GCN-LABEL: {{^}}s_cvt_pknorm_u16_samereg_f32:
; GCN: s_load_dword [[X:s[0-9]+]]
; GCN: v_cvt_pknorm_u16_f32{{(_e64)*}} v{{[0-9]+}}, [[X]], [[X]]
define amdgpu_kernel void @s_cvt_pknorm_u16_samereg_f32(i32 addrspace(1)* %out, float %x) #0 {
%result = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %x, float %x)
%r = bitcast <2 x i16> %result to i32
store i32 %r, i32 addrspace(1)* %out
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; SI: v_cvt_pknorm_u16_f32_e32 v{{[0-9]+}}, [[A]], [[B]]
; VI: v_cvt_pknorm_u16_f32 v{{[0-9]+}}, [[A]], [[B]]
define amdgpu_kernel void @v_cvt_pknorm_u16_f32(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr, float addrspace(1)* %b.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%b.gep = getelementptr inbounds float, float addrspace(1)* %b.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%b = load volatile float, float addrspace(1)* %b.gep
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %a, float %b)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32_reg_imm:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: v_cvt_pknorm_u16_f32{{(_e64)*}} v{{[0-9]+}}, [[A]], 1.0
define amdgpu_kernel void @v_cvt_pknorm_u16_f32_reg_imm(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %a, float 1.0)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32_imm_reg:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; SI: v_cvt_pknorm_u16_f32_e32 v{{[0-9]+}}, 1.0, [[A]]
; VI: v_cvt_pknorm_u16_f32 v{{[0-9]+}}, 1.0, [[A]]
define amdgpu_kernel void @v_cvt_pknorm_u16_f32_imm_reg(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float 1.0, float %a)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32_fneg_lo:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; GCN: v_cvt_pknorm_u16_f32{{(_e64)*}} v{{[0-9]+}}, -[[A]], [[B]]
define amdgpu_kernel void @v_cvt_pknorm_u16_f32_fneg_lo(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr, float addrspace(1)* %b.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%b.gep = getelementptr inbounds float, float addrspace(1)* %b.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%b = load volatile float, float addrspace(1)* %b.gep
%neg.a = fsub float -0.0, %a
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %neg.a, float %b)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32_fneg_hi:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; GCN: v_cvt_pknorm_u16_f32{{(_e64)*}} v{{[0-9]+}}, [[A]], -[[B]]
define amdgpu_kernel void @v_cvt_pknorm_u16_f32_fneg_hi(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr, float addrspace(1)* %b.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%b.gep = getelementptr inbounds float, float addrspace(1)* %b.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%b = load volatile float, float addrspace(1)* %b.gep
%neg.b = fsub float -0.0, %b
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %a, float %neg.b)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32_fneg_lo_hi:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; GCN: v_cvt_pknorm_u16_f32{{(_e64)*}} v{{[0-9]+}}, -[[A]], -[[B]]
define amdgpu_kernel void @v_cvt_pknorm_u16_f32_fneg_lo_hi(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr, float addrspace(1)* %b.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%b.gep = getelementptr inbounds float, float addrspace(1)* %b.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%b = load volatile float, float addrspace(1)* %b.gep
%neg.a = fsub float -0.0, %a
%neg.b = fsub float -0.0, %b
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %neg.a, float %neg.b)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
; GCN-LABEL: {{^}}v_cvt_pknorm_u16_f32_fneg_fabs_lo_fneg_hi:
; GCN: {{buffer|flat|global}}_load_dword [[A:v[0-9]+]]
; GCN: {{buffer|flat|global}}_load_dword [[B:v[0-9]+]]
; GCN: v_cvt_pknorm_u16_f32{{(_e64)*}} v{{[0-9]+}}, -|[[A]]|, -[[B]]
define amdgpu_kernel void @v_cvt_pknorm_u16_f32_fneg_fabs_lo_fneg_hi(i32 addrspace(1)* %out, float addrspace(1)* %a.ptr, float addrspace(1)* %b.ptr) #0 {
%tid = call i32 @llvm.amdgcn.workitem.id.x()
%tid.ext = sext i32 %tid to i64
%a.gep = getelementptr inbounds float, float addrspace(1)* %a.ptr, i64 %tid.ext
%b.gep = getelementptr inbounds float, float addrspace(1)* %b.ptr, i64 %tid.ext
%out.gep = getelementptr inbounds i32, i32 addrspace(1)* %out, i64 %tid.ext
%a = load volatile float, float addrspace(1)* %a.gep
%b = load volatile float, float addrspace(1)* %b.gep
%fabs.a = call float @llvm.fabs.f32(float %a)
%neg.fabs.a = fsub float -0.0, %fabs.a
%neg.b = fsub float -0.0, %b
%cvt = call <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float %neg.fabs.a, float %neg.b)
%r = bitcast <2 x i16> %cvt to i32
store i32 %r, i32 addrspace(1)* %out.gep
ret void
}
declare <2 x i16> @llvm.amdgcn.cvt.pknorm.u16(float, float) #1
declare float @llvm.fabs.f32(float) #1
declare i32 @llvm.amdgcn.workitem.id.x() #1
attributes #0 = { nounwind }
attributes #1 = { nounwind readnone }
|