1 ; RUN: llc -march=amdgcn -mcpu=tahiti -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
2 ; RUN: llc -march=amdgcn -mcpu=tonga -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
4 ; GCN-LABEL: {{^}}vgpr:
5 ; GCN-DAG: v_mov_b32_e32 v1, v0
6 ; GCN-DAG: exp mrt0 v0, v0, v0, v0 done vm
7 ; GCN: s_waitcnt expcnt(0)
8 ; GCN: v_add_f32_e32 v0, 1.0, v1
10 define amdgpu_vs { float, float } @vgpr([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, float %arg3) #0 {
12 call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %arg3, float %arg3, float %arg3, float %arg3, i1 true, i1 true) #0
13 %x = fadd float %arg3, 1.000000e+00
14 %a = insertvalue { float, float } undef, float %x, 0
15 %b = insertvalue { float, float } %a, float %arg3, 1
16 ret { float, float } %b
19 ; GCN-LABEL: {{^}}vgpr_literal:
20 ; GCN: exp mrt0 v0, v0, v0, v0 done vm
22 ; GCN-DAG: v_mov_b32_e32 v0, 1.0
23 ; GCN-DAG: v_mov_b32_e32 v1, 2.0
24 ; GCN-DAG: v_mov_b32_e32 v2, 4.0
25 ; GCN-DAG: v_mov_b32_e32 v3, -1.0
26 ; GCN-DAG: s_waitcnt expcnt(0)
28 define amdgpu_vs { float, float, float, float } @vgpr_literal([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, float %arg3) #0 {
30 call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %arg3, float %arg3, float %arg3, float %arg3, i1 true, i1 true) #0
31 ret { float, float, float, float } { float 1.000000e+00, float 2.000000e+00, float 4.000000e+00, float -1.000000e+00 }
36 ; GCN-NEXT: .long 165584
38 ; GCN-LABEL: {{^}}vgpr_ps_addr0:
39 ; GCN-NOT: v_mov_b32_e32 v0
40 ; GCN-NOT: v_mov_b32_e32 v1
41 ; GCN-NOT: v_mov_b32_e32 v2
42 ; GCN: v_mov_b32_e32 v3, v4
43 ; GCN: v_mov_b32_e32 v4, v6
45 define amdgpu_ps { float, float, float, float, float } @vgpr_ps_addr0([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <2 x i32> %arg3, <2 x i32> %arg4, <2 x i32> %arg5, <3 x i32> %arg6, <2 x i32> %arg7, <2 x i32> %arg8, <2 x i32> %arg9, float %arg10, float %arg11, float %arg12, float %arg13, float %arg14, float %arg15, float %arg16, float %arg17, float %arg18) #1 {
47 %i0 = extractelement <2 x i32> %arg4, i32 0
48 %i1 = extractelement <2 x i32> %arg4, i32 1
49 %i2 = extractelement <2 x i32> %arg7, i32 0
50 %i3 = extractelement <2 x i32> %arg8, i32 0
51 %f0 = bitcast i32 %i0 to float
52 %f1 = bitcast i32 %i1 to float
53 %f2 = bitcast i32 %i2 to float
54 %f3 = bitcast i32 %i3 to float
55 %r0 = insertvalue { float, float, float, float, float } undef, float %f0, 0
56 %r1 = insertvalue { float, float, float, float, float } %r0, float %f1, 1
57 %r2 = insertvalue { float, float, float, float, float } %r1, float %f2, 2
58 %r3 = insertvalue { float, float, float, float, float } %r2, float %f3, 3
59 %r4 = insertvalue { float, float, float, float, float } %r3, float %arg12, 4
60 ret { float, float, float, float, float } %r4
65 ; GCN-NEXT: .long 165584
67 ; GCN-LABEL: {{^}}ps_input_ena_no_inputs:
68 ; GCN: v_mov_b32_e32 v0, 1.0
70 define amdgpu_ps float @ps_input_ena_no_inputs([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <2 x i32> %arg3, <2 x i32> %arg4, <2 x i32> %arg5, <3 x i32> %arg6, <2 x i32> %arg7, <2 x i32> %arg8, <2 x i32> %arg9, float %arg10, float %arg11, float %arg12, float %arg13, float %arg14, float %arg15, float %arg16, float %arg17, float %arg18) #1 {
72 ret float 1.000000e+00
76 ; GCN-NEXT: .long 2081
77 ; GCN-NEXT: .long 165584
78 ; GCN-NEXT: .long 2081
79 ; GCN-LABEL: {{^}}ps_input_ena_pos_w:
80 ; GCN-DAG: v_mov_b32_e32 v0, v4
81 ; GCN-DAG: v_mov_b32_e32 v1, v2
82 ; GCN-DAG: v_mov_b32_e32 v2, v3
84 define amdgpu_ps { float, <2 x float> } @ps_input_ena_pos_w([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <2 x i32> %arg3, <2 x i32> %arg4, <2 x i32> %arg5, <3 x i32> %arg6, <2 x i32> %arg7, <2 x i32> %arg8, <2 x i32> %arg9, float %arg10, float %arg11, float %arg12, float %arg13, float %arg14, float %arg15, float %arg16, float %arg17, float %arg18) #1 {
86 %f = bitcast <2 x i32> %arg8 to <2 x float>
87 %s = insertvalue { float, <2 x float> } undef, float %arg14, 0
88 %s1 = insertvalue { float, <2 x float> } %s, <2 x float> %f, 1
89 ret { float, <2 x float> } %s1
94 ; GCN-NEXT: .long 165584
96 ; GCN-LABEL: {{^}}vgpr_ps_addr1:
97 ; GCN-DAG: v_mov_b32_e32 v0, v2
98 ; GCN-DAG: v_mov_b32_e32 v1, v3
99 ; GCN: v_mov_b32_e32 v2, v4
100 ; GCN-DAG: v_mov_b32_e32 v3, v6
101 ; GCN-DAG: v_mov_b32_e32 v4, v8
103 define amdgpu_ps { float, float, float, float, float } @vgpr_ps_addr1([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <2 x i32> %arg3, <2 x i32> %arg4, <2 x i32> %arg5, <3 x i32> %arg6, <2 x i32> %arg7, <2 x i32> %arg8, <2 x i32> %arg9, float %arg10, float %arg11, float %arg12, float %arg13, float %arg14, float %arg15, float %arg16, float %arg17, float %arg18) #2 {
105 %i0 = extractelement <2 x i32> %arg4, i32 0
106 %i1 = extractelement <2 x i32> %arg4, i32 1
107 %i2 = extractelement <2 x i32> %arg7, i32 0
108 %i3 = extractelement <2 x i32> %arg8, i32 0
109 %f0 = bitcast i32 %i0 to float
110 %f1 = bitcast i32 %i1 to float
111 %f2 = bitcast i32 %i2 to float
112 %f3 = bitcast i32 %i3 to float
113 %r0 = insertvalue { float, float, float, float, float } undef, float %f0, 0
114 %r1 = insertvalue { float, float, float, float, float } %r0, float %f1, 1
115 %r2 = insertvalue { float, float, float, float, float } %r1, float %f2, 2
116 %r3 = insertvalue { float, float, float, float, float } %r2, float %f3, 3
117 %r4 = insertvalue { float, float, float, float, float } %r3, float %arg12, 4
118 ret { float, float, float, float, float } %r4
122 ; GCN-NEXT: .long 562
123 ; GCN-NEXT: .long 165584
124 ; GCN-NEXT: .long 631
125 ; GCN-LABEL: {{^}}vgpr_ps_addr119:
126 ; GCN-DAG: v_mov_b32_e32 v0, v2
127 ; GCN-DAG: v_mov_b32_e32 v1, v3
128 ; GCN-DAG: v_mov_b32_e32 v2, v6
129 ; GCN-DAG: v_mov_b32_e32 v3, v8
130 ; GCN-DAG: v_mov_b32_e32 v4, v12
132 define amdgpu_ps { float, float, float, float, float } @vgpr_ps_addr119([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <2 x i32> %arg3, <2 x i32> %arg4, <2 x i32> %arg5, <3 x i32> %arg6, <2 x i32> %arg7, <2 x i32> %arg8, <2 x i32> %arg9, float %arg10, float %arg11, float %arg12, float %arg13, float %arg14, float %arg15, float %arg16, float %arg17, float %arg18) #3 {
134 %i0 = extractelement <2 x i32> %arg4, i32 0
135 %i1 = extractelement <2 x i32> %arg4, i32 1
136 %i2 = extractelement <2 x i32> %arg7, i32 0
137 %i3 = extractelement <2 x i32> %arg8, i32 0
138 %f0 = bitcast i32 %i0 to float
139 %f1 = bitcast i32 %i1 to float
140 %f2 = bitcast i32 %i2 to float
141 %f3 = bitcast i32 %i3 to float
142 %r0 = insertvalue { float, float, float, float, float } undef, float %f0, 0
143 %r1 = insertvalue { float, float, float, float, float } %r0, float %f1, 1
144 %r2 = insertvalue { float, float, float, float, float } %r1, float %f2, 2
145 %r3 = insertvalue { float, float, float, float, float } %r2, float %f3, 3
146 %r4 = insertvalue { float, float, float, float, float } %r3, float %arg12, 4
147 ret { float, float, float, float, float } %r4
151 ; GCN-NEXT: .long 562
152 ; GCN-NEXT: .long 165584
153 ; GCN-NEXT: .long 946
154 ; GCN-LABEL: {{^}}vgpr_ps_addr418:
155 ; GCN-NOT: v_mov_b32_e32 v0
156 ; GCN-NOT: v_mov_b32_e32 v1
157 ; GCN-NOT: v_mov_b32_e32 v2
158 ; GCN: v_mov_b32_e32 v3, v4
159 ; GCN: v_mov_b32_e32 v4, v8
161 define amdgpu_ps { float, float, float, float, float } @vgpr_ps_addr418([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, <2 x i32> %arg3, <2 x i32> %arg4, <2 x i32> %arg5, <3 x i32> %arg6, <2 x i32> %arg7, <2 x i32> %arg8, <2 x i32> %arg9, float %arg10, float %arg11, float %arg12, float %arg13, float %arg14, float %arg15, float %arg16, float %arg17, float %arg18) #4 {
163 %i0 = extractelement <2 x i32> %arg4, i32 0
164 %i1 = extractelement <2 x i32> %arg4, i32 1
165 %i2 = extractelement <2 x i32> %arg7, i32 0
166 %i3 = extractelement <2 x i32> %arg8, i32 0
167 %f0 = bitcast i32 %i0 to float
168 %f1 = bitcast i32 %i1 to float
169 %f2 = bitcast i32 %i2 to float
170 %f3 = bitcast i32 %i3 to float
171 %r0 = insertvalue { float, float, float, float, float } undef, float %f0, 0
172 %r1 = insertvalue { float, float, float, float, float } %r0, float %f1, 1
173 %r2 = insertvalue { float, float, float, float, float } %r1, float %f2, 2
174 %r3 = insertvalue { float, float, float, float, float } %r2, float %f3, 3
175 %r4 = insertvalue { float, float, float, float, float } %r3, float %arg12, 4
176 ret { float, float, float, float, float } %r4
179 ; GCN-LABEL: {{^}}sgpr:
180 ; GCN-DAG: s_mov_b32 s2, s3
181 ; GCN-DAG: s_add_{{i|u}}32 s0, s3, 2
183 define amdgpu_vs { i32, i32, i32 } @sgpr([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, float %arg3) #0 {
185 %x = add i32 %arg2, 2
186 %a = insertvalue { i32, i32, i32 } undef, i32 %x, 0
187 %b = insertvalue { i32, i32, i32 } %a, i32 %arg1, 1
188 %c = insertvalue { i32, i32, i32 } %a, i32 %arg2, 2
189 ret { i32, i32, i32 } %c
192 ; GCN-LABEL: {{^}}sgpr_literal:
193 ; GCN: s_mov_b32 s0, 5
194 ; GCN-NOT: s_mov_b32 s0, s0
195 ; GCN-DAG: s_mov_b32 s1, 6
196 ; GCN-DAG: s_mov_b32 s2, 7
197 ; GCN-DAG: s_mov_b32 s3, 8
199 define amdgpu_vs { i32, i32, i32, i32 } @sgpr_literal([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, float %arg3) #0 {
201 %x = add i32 %arg2, 2
202 ret { i32, i32, i32, i32 } { i32 5, i32 6, i32 7, i32 8 }
205 ; GCN-LABEL: {{^}}both:
206 ; GCN-DAG: exp mrt0 v0, v0, v0, v0 done vm
207 ; GCN-DAG: v_mov_b32_e32 v1, v0
208 ; GCN-DAG: s_mov_b32 s1, s2
209 ; GCN-DAG: s_waitcnt expcnt(0)
210 ; GCN-DAG: v_add_f32_e32 v0, 1.0, v1
211 ; GCN-DAG: s_add_{{i|u}}32 s0, s3, 2
212 ; GCN-DAG: s_mov_b32 s2, s3
214 define amdgpu_vs { float, i32, float, i32, i32 } @both([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, float %arg3) #0 {
216 call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %arg3, float %arg3, float %arg3, float %arg3, i1 true, i1 true) #0
217 %v = fadd float %arg3, 1.000000e+00
218 %s = add i32 %arg2, 2
219 %a0 = insertvalue { float, i32, float, i32, i32 } undef, float %v, 0
220 %a1 = insertvalue { float, i32, float, i32, i32 } %a0, i32 %s, 1
221 %a2 = insertvalue { float, i32, float, i32, i32 } %a1, float %arg3, 2
222 %a3 = insertvalue { float, i32, float, i32, i32 } %a2, i32 %arg1, 3
223 %a4 = insertvalue { float, i32, float, i32, i32 } %a3, i32 %arg2, 4
224 ret { float, i32, float, i32, i32 } %a4
227 ; GCN-LABEL: {{^}}structure_literal:
228 ; GCN: exp mrt0 v0, v0, v0, v0 done vm
230 ; GCN-DAG: v_mov_b32_e32 v0, 1.0
231 ; GCN-DAG: s_mov_b32 s0, 2
232 ; GCN-DAG: s_mov_b32 s1, 3
233 ; GCN-DAG: v_mov_b32_e32 v1, 2.0
234 ; GCN-DAG: v_mov_b32_e32 v2, 4.0
235 ; GCN-DAG: s_waitcnt expcnt(0)
236 define amdgpu_vs { { float, i32 }, { i32, <2 x float> } } @structure_literal([9 x <16 x i8>] addrspace(4)* inreg %arg, i32 inreg %arg1, i32 inreg %arg2, float %arg3) #0 {
238 call void @llvm.amdgcn.exp.f32(i32 0, i32 15, float %arg3, float %arg3, float %arg3, float %arg3, i1 true, i1 true) #0
239 ret { { float, i32 }, { i32, <2 x float> } } { { float, i32 } { float 1.000000e+00, i32 2 }, { i32, <2 x float> } { i32 3, <2 x float> <float 2.000000e+00, float 4.000000e+00> } }
242 ; GCN-LABEL: {{^}}ret_return_to_epilog_pseudo_size:
243 ; GCN: codeLenInByte = 0{{$}}
244 define amdgpu_ps float @ret_return_to_epilog_pseudo_size() #0 {
248 declare void @llvm.amdgcn.exp.f32(i32, i32, float, float, float, float, i1, i1) #0
250 attributes #0 = { nounwind }
251 attributes #1 = { nounwind "InitialPSInputAddr"="0" }
252 attributes #2 = { nounwind "InitialPSInputAddr"="1" }
253 attributes #3 = { nounwind "InitialPSInputAddr"="119" }
254 attributes #4 = { nounwind "InitialPSInputAddr"="418" }