1 ; RUN: llc -mtriple=amdgcn -mcpu=gfx906 -verify-machineinstrs < %s | FileCheck --check-prefix=OCC %s
2 ; RUN: llc -mtriple=amdgcn -mcpu=gfx906 -verify-machineinstrs -amdgpu-schedule-relaxed-occupancy=true < %s | FileCheck --check-prefix=RELAX %s
5 ; Using -amgpu-schedule-relaxed-occupancy allows scheduler to produce better ILP by further relaxing occupancy target
7 ; GCN-LABEL: {{^}}load_fma_store:
10 ; OCC: NumVGPRsForWavesPerEU: 32
11 ; RELAX: NumVGPRsForWavesPerEU: 64
15 define amdgpu_kernel void @load_fma_store(ptr addrspace(3) nocapture readonly %arg, ptr addrspace(1) nocapture %arg1) #1 {
17 %tmp = getelementptr inbounds float, ptr addrspace(3) %arg, i32 1
18 %tmp2 = load float, ptr addrspace(3) %tmp, align 4
19 %tmp3 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 2
20 %tmp4 = load float, ptr addrspace(3) %tmp3, align 4
21 %tmp5 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 3
22 %tmp6 = load float, ptr addrspace(3) %tmp5, align 4
23 %tmp7 = tail call float @llvm.fmuladd.f32(float %tmp2, float %tmp4, float %tmp6)
24 %tmp8 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 5
25 %tmp9 = load float, ptr addrspace(3) %tmp8, align 4
26 %tmp10 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 6
27 %tmp11 = load float, ptr addrspace(3) %tmp10, align 4
28 %tmp12 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 7
29 %tmp13 = load float, ptr addrspace(3) %tmp12, align 4
30 %tmp14 = tail call float @llvm.fmuladd.f32(float %tmp9, float %tmp11, float %tmp13)
31 %tmp15 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 9
32 %tmp16 = load float, ptr addrspace(3) %tmp15, align 4
33 %tmp17 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 10
34 %tmp18 = load float, ptr addrspace(3) %tmp17, align 4
35 %tmp19 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 11
36 %tmp20 = load float, ptr addrspace(3) %tmp19, align 4
37 %tmp21 = tail call float @llvm.fmuladd.f32(float %tmp16, float %tmp18, float %tmp20)
38 %tmp22 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 13
39 %tmp23 = load float, ptr addrspace(3) %tmp22, align 4
40 %tmp24 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 14
41 %tmp25 = load float, ptr addrspace(3) %tmp24, align 4
42 %tmp26 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 15
43 %tmp27 = load float, ptr addrspace(3) %tmp26, align 4
44 %tmp28 = tail call float @llvm.fmuladd.f32(float %tmp23, float %tmp25, float %tmp27)
45 %tmp29 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 17
46 %tmp30 = load float, ptr addrspace(3) %tmp29, align 4
47 %tmp31 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 18
48 %tmp32 = load float, ptr addrspace(3) %tmp31, align 4
49 %tmp33 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 19
50 %tmp34 = load float, ptr addrspace(3) %tmp33, align 4
51 %tmp35 = tail call float @llvm.fmuladd.f32(float %tmp30, float %tmp32, float %tmp34)
52 %tmp36 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 21
53 %tmp37 = load float, ptr addrspace(3) %tmp36, align 4
54 %tmp38 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 22
55 %tmp39 = load float, ptr addrspace(3) %tmp38, align 4
56 %tmp40 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 23
57 %tmp41 = load float, ptr addrspace(3) %tmp40, align 4
58 %tmp42 = tail call float @llvm.fmuladd.f32(float %tmp37, float %tmp39, float %tmp41)
59 %tmp43 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 25
60 %tmp44 = load float, ptr addrspace(3) %tmp43, align 4
61 %tmp45 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 26
62 %tmp46 = load float, ptr addrspace(3) %tmp45, align 4
63 %tmp47 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 27
64 %tmp48 = load float, ptr addrspace(3) %tmp47, align 4
65 %tmp49 = tail call float @llvm.fmuladd.f32(float %tmp44, float %tmp46, float %tmp48)
66 %tmp50 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 29
67 %tmp51 = load float, ptr addrspace(3) %tmp50, align 4
68 %tmp52 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 30
69 %tmp53 = load float, ptr addrspace(3) %tmp52, align 4
70 %tmp54 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 31
71 %tmp55 = load float, ptr addrspace(3) %tmp54, align 4
72 %tmp56 = tail call float @llvm.fmuladd.f32(float %tmp51, float %tmp53, float %tmp55)
73 %tmp57 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 33
74 %tmp58 = load float, ptr addrspace(3) %tmp57, align 4
75 %tmp59 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 34
76 %tmp60 = load float, ptr addrspace(3) %tmp59, align 4
77 %tmp61 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 35
78 %tmp62 = load float, ptr addrspace(3) %tmp61, align 4
79 %tmp63 = tail call float @llvm.fmuladd.f32(float %tmp58, float %tmp60, float %tmp62)
80 %tmp64 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 37
81 %tmp65 = load float, ptr addrspace(3) %tmp64, align 4
82 %tmp66 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 38
83 %tmp67 = load float, ptr addrspace(3) %tmp66, align 4
84 %tmp68 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 39
85 %tmp69 = load float, ptr addrspace(3) %tmp68, align 4
86 %tmp70 = tail call float @llvm.fmuladd.f32(float %tmp65, float %tmp67, float %tmp69)
87 %tmp71 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 41
88 %tmp72 = load float, ptr addrspace(3) %tmp71, align 4
89 %tmp73 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 42
90 %tmp74 = load float, ptr addrspace(3) %tmp73, align 4
91 %tmp75 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 43
92 %tmp76 = load float, ptr addrspace(3) %tmp75, align 4
93 %tmp77 = tail call float @llvm.fmuladd.f32(float %tmp72, float %tmp74, float %tmp76)
94 %tmp78 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 45
95 %tmp79 = load float, ptr addrspace(3) %tmp78, align 4
96 %tmp80 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 46
97 %tmp81 = load float, ptr addrspace(3) %tmp80, align 4
98 %tmp82 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 47
99 %tmp83 = load float, ptr addrspace(3) %tmp82, align 4
100 %tmp84 = tail call float @llvm.fmuladd.f32(float %tmp79, float %tmp81, float %tmp83)
101 %tmp85 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 49
102 %tmp86 = load float, ptr addrspace(3) %tmp85, align 4
103 %tmp87 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 50
104 %tmp88 = load float, ptr addrspace(3) %tmp87, align 4
105 %tmp89 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 51
106 %tmp90 = load float, ptr addrspace(3) %tmp89, align 4
107 %tmp91 = tail call float @llvm.fmuladd.f32(float %tmp86, float %tmp88, float %tmp90)
108 %tmp92 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 53
109 %tmp93 = load float, ptr addrspace(3) %tmp92, align 4
110 %tmp94 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 54
111 %tmp95 = load float, ptr addrspace(3) %tmp94, align 4
112 %tmp96 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 55
113 %tmp97 = load float, ptr addrspace(3) %tmp96, align 4
114 %tmp98 = tail call float @llvm.fmuladd.f32(float %tmp93, float %tmp95, float %tmp97)
115 %tmp99 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 57
116 %tmp100 = load float, ptr addrspace(3) %tmp99, align 4
117 %tmp101 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 58
118 %tmp102 = load float, ptr addrspace(3) %tmp101, align 4
119 %tmp103 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 59
120 %tmp104 = load float, ptr addrspace(3) %tmp103, align 4
121 %tmp105 = tail call float @llvm.fmuladd.f32(float %tmp100, float %tmp102, float %tmp104)
122 %tmp106 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 61
123 %tmp107 = load float, ptr addrspace(3) %tmp106, align 4
124 %tmp108 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 62
125 %tmp109 = load float, ptr addrspace(3) %tmp108, align 4
126 %tmp110 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 63
127 %tmp111 = load float, ptr addrspace(3) %tmp110, align 4
128 %tmp112 = tail call float @llvm.fmuladd.f32(float %tmp107, float %tmp109, float %tmp111)
129 %tmp113 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 65
130 %tmp114 = load float, ptr addrspace(3) %tmp113, align 4
131 %tmp115 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 66
132 %tmp116 = load float, ptr addrspace(3) %tmp115, align 4
133 %tmp117 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 67
134 %tmp118 = load float, ptr addrspace(3) %tmp117, align 4
135 %tmp119 = tail call float @llvm.fmuladd.f32(float %tmp114, float %tmp116, float %tmp118)
136 %tmp120 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 69
137 %tmp121 = load float, ptr addrspace(3) %tmp120, align 4
138 %tmp122 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 70
139 %tmp123 = load float, ptr addrspace(3) %tmp122, align 4
140 %tmp124 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 71
141 %tmp125 = load float, ptr addrspace(3) %tmp124, align 4
142 %tmp126 = tail call float @llvm.fmuladd.f32(float %tmp121, float %tmp123, float %tmp125)
143 %tmp127 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 73
144 %tmp128 = load float, ptr addrspace(3) %tmp127, align 4
145 %tmp129 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 74
146 %tmp130 = load float, ptr addrspace(3) %tmp129, align 4
147 %tmp131 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 75
148 %tmp132 = load float, ptr addrspace(3) %tmp131, align 4
149 %tmp133 = tail call float @llvm.fmuladd.f32(float %tmp128, float %tmp130, float %tmp132)
150 %tmp134 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 77
151 %tmp135 = load float, ptr addrspace(3) %tmp134, align 4
152 %tmp136 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 78
153 %tmp137 = load float, ptr addrspace(3) %tmp136, align 4
154 %tmp138 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 79
155 %tmp139 = load float, ptr addrspace(3) %tmp138, align 4
156 %tmp140 = tail call float @llvm.fmuladd.f32(float %tmp135, float %tmp137, float %tmp139)
157 %tmp141 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 81
158 %tmp142 = load float, ptr addrspace(3) %tmp141, align 4
159 %tmp143 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 82
160 %tmp144 = load float, ptr addrspace(3) %tmp143, align 4
161 %tmp145 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 83
162 %tmp146 = load float, ptr addrspace(3) %tmp145, align 4
163 %tmp147 = tail call float @llvm.fmuladd.f32(float %tmp142, float %tmp144, float %tmp146)
164 %tmp148 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 85
165 %tmp149 = load float, ptr addrspace(3) %tmp148, align 4
166 %tmp150 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 86
167 %tmp151 = load float, ptr addrspace(3) %tmp150, align 4
168 %tmp152 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 87
169 %tmp153 = load float, ptr addrspace(3) %tmp152, align 4
170 %tmp154 = tail call float @llvm.fmuladd.f32(float %tmp149, float %tmp151, float %tmp153)
171 %tmp155 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 89
172 %tmp156 = load float, ptr addrspace(3) %tmp155, align 4
173 %tmp157 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 90
174 %tmp158 = load float, ptr addrspace(3) %tmp157, align 4
175 %tmp159 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 91
176 %tmp160 = load float, ptr addrspace(3) %tmp159, align 4
177 %tmp161 = tail call float @llvm.fmuladd.f32(float %tmp156, float %tmp158, float %tmp160)
178 %tmp162 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 93
179 %tmp163 = load float, ptr addrspace(3) %tmp162, align 4
180 %tmp164 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 94
181 %tmp165 = load float, ptr addrspace(3) %tmp164, align 4
182 %tmp166 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 95
183 %tmp167 = load float, ptr addrspace(3) %tmp166, align 4
184 %tmp168 = tail call float @llvm.fmuladd.f32(float %tmp163, float %tmp165, float %tmp167)
185 %tmp169 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 97
186 %tmp170 = load float, ptr addrspace(3) %tmp169, align 4
187 %tmp171 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 98
188 %tmp172 = load float, ptr addrspace(3) %tmp171, align 4
189 %tmp173 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 99
190 %tmp174 = load float, ptr addrspace(3) %tmp173, align 4
191 %tmp175 = tail call float @llvm.fmuladd.f32(float %tmp170, float %tmp172, float %tmp174)
192 %tmp176 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 101
193 %tmp177 = load float, ptr addrspace(3) %tmp176, align 4
194 %tmp178 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 102
195 %tmp179 = load float, ptr addrspace(3) %tmp178, align 4
196 %tmp180 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 103
197 %tmp181 = load float, ptr addrspace(3) %tmp180, align 4
198 %tmp182 = tail call float @llvm.fmuladd.f32(float %tmp177, float %tmp179, float %tmp181)
199 %tmp183 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 105
200 %tmp184 = load float, ptr addrspace(3) %tmp183, align 4
201 %tmp185 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 106
202 %tmp186 = load float, ptr addrspace(3) %tmp185, align 4
203 %tmp187 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 107
204 %tmp188 = load float, ptr addrspace(3) %tmp187, align 4
205 %tmp189 = tail call float @llvm.fmuladd.f32(float %tmp184, float %tmp186, float %tmp188)
206 %tmp190 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 109
207 %tmp191 = load float, ptr addrspace(3) %tmp190, align 4
208 %tmp192 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 110
209 %tmp193 = load float, ptr addrspace(3) %tmp192, align 4
210 %tmp194 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 111
211 %tmp195 = load float, ptr addrspace(3) %tmp194, align 4
212 %tmp196 = tail call float @llvm.fmuladd.f32(float %tmp191, float %tmp193, float %tmp195)
213 %tmp197 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 113
214 %tmp198 = load float, ptr addrspace(3) %tmp197, align 4
215 %tmp199 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 114
216 %tmp200 = load float, ptr addrspace(3) %tmp199, align 4
217 %tmp201 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 115
218 %tmp202 = load float, ptr addrspace(3) %tmp201, align 4
219 %tmp203 = tail call float @llvm.fmuladd.f32(float %tmp198, float %tmp200, float %tmp202)
220 %tmp204 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 117
221 %tmp205 = load float, ptr addrspace(3) %tmp204, align 4
222 %tmp206 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 118
223 %tmp207 = load float, ptr addrspace(3) %tmp206, align 4
224 %tmp208 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 119
225 %tmp209 = load float, ptr addrspace(3) %tmp208, align 4
226 %tmp210 = tail call float @llvm.fmuladd.f32(float %tmp205, float %tmp207, float %tmp209)
227 %tmp211 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 121
228 %tmp212 = load float, ptr addrspace(3) %tmp211, align 4
229 %tmp213 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 122
230 %tmp214 = load float, ptr addrspace(3) %tmp213, align 4
231 %tmp215 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 123
232 %tmp216 = load float, ptr addrspace(3) %tmp215, align 4
233 %tmp217 = tail call float @llvm.fmuladd.f32(float %tmp212, float %tmp214, float %tmp216)
234 %tmp218 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 125
235 %tmp219 = load float, ptr addrspace(3) %tmp218, align 4
236 %tmp220 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 126
237 %tmp221 = load float, ptr addrspace(3) %tmp220, align 4
238 %tmp222 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 127
239 %tmp223 = load float, ptr addrspace(3) %tmp222, align 4
240 %tmp224 = tail call float @llvm.fmuladd.f32(float %tmp219, float %tmp221, float %tmp223)
241 %tmp225 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 129
242 %tmp226 = load float, ptr addrspace(3) %tmp225, align 4
243 %tmp227 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 130
244 %tmp228 = load float, ptr addrspace(3) %tmp227, align 4
245 %tmp229 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 131
246 %tmp230 = load float, ptr addrspace(3) %tmp229, align 4
247 %tmp231 = tail call float @llvm.fmuladd.f32(float %tmp226, float %tmp228, float %tmp230)
248 %tmp232 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 133
249 %tmp233 = load float, ptr addrspace(3) %tmp232, align 4
250 %tmp234 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 134
251 %tmp235 = load float, ptr addrspace(3) %tmp234, align 4
252 %tmp236 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 135
253 %tmp237 = load float, ptr addrspace(3) %tmp236, align 4
254 %tmp238 = tail call float @llvm.fmuladd.f32(float %tmp233, float %tmp235, float %tmp237)
255 %tmp239 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 137
256 %tmp240 = load float, ptr addrspace(3) %tmp239, align 4
257 %tmp241 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 138
258 %tmp242 = load float, ptr addrspace(3) %tmp241, align 4
259 %tmp243 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 139
260 %tmp244 = load float, ptr addrspace(3) %tmp243, align 4
261 %tmp245 = tail call float @llvm.fmuladd.f32(float %tmp240, float %tmp242, float %tmp244)
262 %tmp246 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 141
263 %tmp247 = load float, ptr addrspace(3) %tmp246, align 4
264 %tmp248 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 142
265 %tmp249 = load float, ptr addrspace(3) %tmp248, align 4
266 %tmp250 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 143
267 %tmp251 = load float, ptr addrspace(3) %tmp250, align 4
268 %tmp252 = tail call float @llvm.fmuladd.f32(float %tmp247, float %tmp249, float %tmp251)
269 %tmp253 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 145
270 %tmp254 = load float, ptr addrspace(3) %tmp253, align 4
271 %tmp255 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 146
272 %tmp256 = load float, ptr addrspace(3) %tmp255, align 4
273 %tmp257 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 147
274 %tmp258 = load float, ptr addrspace(3) %tmp257, align 4
275 %tmp259 = tail call float @llvm.fmuladd.f32(float %tmp254, float %tmp256, float %tmp258)
276 %tmp260 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 149
277 %tmp261 = load float, ptr addrspace(3) %tmp260, align 4
278 %tmp262 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 150
279 %tmp263 = load float, ptr addrspace(3) %tmp262, align 4
280 %tmp264 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 151
281 %tmp265 = load float, ptr addrspace(3) %tmp264, align 4
282 %tmp266 = tail call float @llvm.fmuladd.f32(float %tmp261, float %tmp263, float %tmp265)
283 %tmp267 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 153
284 %tmp268 = load float, ptr addrspace(3) %tmp267, align 4
285 %tmp269 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 154
286 %tmp270 = load float, ptr addrspace(3) %tmp269, align 4
287 %tmp271 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 155
288 %tmp272 = load float, ptr addrspace(3) %tmp271, align 4
289 %tmp273 = tail call float @llvm.fmuladd.f32(float %tmp268, float %tmp270, float %tmp272)
290 %tmp274 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 157
291 %tmp275 = load float, ptr addrspace(3) %tmp274, align 4
292 %tmp276 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 158
293 %tmp277 = load float, ptr addrspace(3) %tmp276, align 4
294 %tmp278 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 159
295 %tmp279 = load float, ptr addrspace(3) %tmp278, align 4
296 %tmp280 = tail call float @llvm.fmuladd.f32(float %tmp275, float %tmp277, float %tmp279)
297 %tmp281 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 161
298 %tmp282 = load float, ptr addrspace(3) %tmp281, align 4
299 %tmp283 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 162
300 %tmp284 = load float, ptr addrspace(3) %tmp283, align 4
301 %tmp285 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 163
302 %tmp286 = load float, ptr addrspace(3) %tmp285, align 4
303 %tmp287 = tail call float @llvm.fmuladd.f32(float %tmp282, float %tmp284, float %tmp286)
304 %tmp288 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 165
305 %tmp289 = load float, ptr addrspace(3) %tmp288, align 4
306 %tmp290 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 166
307 %tmp291 = load float, ptr addrspace(3) %tmp290, align 4
308 %tmp292 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 167
309 %tmp293 = load float, ptr addrspace(3) %tmp292, align 4
310 %tmp294 = tail call float @llvm.fmuladd.f32(float %tmp289, float %tmp291, float %tmp293)
311 %tmp295 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 169
312 %tmp296 = load float, ptr addrspace(3) %tmp295, align 4
313 %tmp297 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 170
314 %tmp298 = load float, ptr addrspace(3) %tmp297, align 4
315 %tmp299 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 171
316 %tmp300 = load float, ptr addrspace(3) %tmp299, align 4
317 %tmp301 = tail call float @llvm.fmuladd.f32(float %tmp296, float %tmp298, float %tmp300)
318 %tmp302 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 173
319 %tmp303 = load float, ptr addrspace(3) %tmp302, align 4
320 %tmp304 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 174
321 %tmp305 = load float, ptr addrspace(3) %tmp304, align 4
322 %tmp306 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 175
323 %tmp307 = load float, ptr addrspace(3) %tmp306, align 4
324 %tmp308 = tail call float @llvm.fmuladd.f32(float %tmp303, float %tmp305, float %tmp307)
325 %tmp309 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 177
326 %tmp310 = load float, ptr addrspace(3) %tmp309, align 4
327 %tmp311 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 178
328 %tmp312 = load float, ptr addrspace(3) %tmp311, align 4
329 %tmp313 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 179
330 %tmp314 = load float, ptr addrspace(3) %tmp313, align 4
331 %tmp315 = tail call float @llvm.fmuladd.f32(float %tmp310, float %tmp312, float %tmp314)
332 %tmp316 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 181
333 %tmp317 = load float, ptr addrspace(3) %tmp316, align 4
334 %tmp318 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 182
335 %tmp319 = load float, ptr addrspace(3) %tmp318, align 4
336 %tmp320 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 183
337 %tmp321 = load float, ptr addrspace(3) %tmp320, align 4
338 %tmp322 = tail call float @llvm.fmuladd.f32(float %tmp317, float %tmp319, float %tmp321)
339 %tmp323 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 185
340 %tmp324 = load float, ptr addrspace(3) %tmp323, align 4
341 %tmp325 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 186
342 %tmp326 = load float, ptr addrspace(3) %tmp325, align 4
343 %tmp327 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 187
344 %tmp328 = load float, ptr addrspace(3) %tmp327, align 4
345 %tmp329 = tail call float @llvm.fmuladd.f32(float %tmp324, float %tmp326, float %tmp328)
346 %tmp330 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 189
347 %tmp331 = load float, ptr addrspace(3) %tmp330, align 4
348 %tmp332 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 190
349 %tmp333 = load float, ptr addrspace(3) %tmp332, align 4
350 %tmp334 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 191
351 %tmp335 = load float, ptr addrspace(3) %tmp334, align 4
352 %tmp336 = tail call float @llvm.fmuladd.f32(float %tmp331, float %tmp333, float %tmp335)
353 %tmp337 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 193
354 %tmp338 = load float, ptr addrspace(3) %tmp337, align 4
355 %tmp339 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 194
356 %tmp340 = load float, ptr addrspace(3) %tmp339, align 4
357 %tmp341 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 195
358 %tmp342 = load float, ptr addrspace(3) %tmp341, align 4
359 %tmp343 = tail call float @llvm.fmuladd.f32(float %tmp338, float %tmp340, float %tmp342)
360 %tmp344 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 197
361 %tmp345 = load float, ptr addrspace(3) %tmp344, align 4
362 %tmp346 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 198
363 %tmp347 = load float, ptr addrspace(3) %tmp346, align 4
364 %tmp348 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 199
365 %tmp349 = load float, ptr addrspace(3) %tmp348, align 4
366 %tmp350 = tail call float @llvm.fmuladd.f32(float %tmp345, float %tmp347, float %tmp349)
367 %tmp351 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 201
368 %tmp352 = load float, ptr addrspace(3) %tmp351, align 4
369 %tmp353 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 202
370 %tmp354 = load float, ptr addrspace(3) %tmp353, align 4
371 %tmp355 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 203
372 %tmp356 = load float, ptr addrspace(3) %tmp355, align 4
373 %tmp357 = tail call float @llvm.fmuladd.f32(float %tmp352, float %tmp354, float %tmp356)
374 %tmp358 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 205
375 %tmp359 = load float, ptr addrspace(3) %tmp358, align 4
376 %tmp360 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 206
377 %tmp361 = load float, ptr addrspace(3) %tmp360, align 4
378 %tmp362 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 207
379 %tmp363 = load float, ptr addrspace(3) %tmp362, align 4
380 %tmp364 = tail call float @llvm.fmuladd.f32(float %tmp359, float %tmp361, float %tmp363)
381 %tmp365 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 209
382 %tmp366 = load float, ptr addrspace(3) %tmp365, align 4
383 %tmp367 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 210
384 %tmp368 = load float, ptr addrspace(3) %tmp367, align 4
385 %tmp369 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 211
386 %tmp370 = load float, ptr addrspace(3) %tmp369, align 4
387 %tmp371 = tail call float @llvm.fmuladd.f32(float %tmp366, float %tmp368, float %tmp370)
388 %tmp372 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 213
389 %tmp373 = load float, ptr addrspace(3) %tmp372, align 4
390 %tmp374 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 214
391 %tmp375 = load float, ptr addrspace(3) %tmp374, align 4
392 %tmp376 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 215
393 %tmp377 = load float, ptr addrspace(3) %tmp376, align 4
394 %tmp378 = tail call float @llvm.fmuladd.f32(float %tmp373, float %tmp375, float %tmp377)
395 %tmp379 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 217
396 %tmp380 = load float, ptr addrspace(3) %tmp379, align 4
397 %tmp381 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 218
398 %tmp382 = load float, ptr addrspace(3) %tmp381, align 4
399 %tmp383 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 219
400 %tmp384 = load float, ptr addrspace(3) %tmp383, align 4
401 %tmp385 = tail call float @llvm.fmuladd.f32(float %tmp380, float %tmp382, float %tmp384)
402 %tmp386 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 221
403 %tmp387 = load float, ptr addrspace(3) %tmp386, align 4
404 %tmp388 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 222
405 %tmp389 = load float, ptr addrspace(3) %tmp388, align 4
406 %tmp390 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 223
407 %tmp391 = load float, ptr addrspace(3) %tmp390, align 4
408 %tmp392 = tail call float @llvm.fmuladd.f32(float %tmp387, float %tmp389, float %tmp391)
409 %tmp393 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 225
410 %tmp394 = load float, ptr addrspace(3) %tmp393, align 4
411 %tmp395 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 226
412 %tmp396 = load float, ptr addrspace(3) %tmp395, align 4
413 %tmp397 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 227
414 %tmp398 = load float, ptr addrspace(3) %tmp397, align 4
415 %tmp399 = tail call float @llvm.fmuladd.f32(float %tmp394, float %tmp396, float %tmp398)
416 %tmp400 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 229
417 %tmp401 = load float, ptr addrspace(3) %tmp400, align 4
418 %tmp402 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 230
419 %tmp403 = load float, ptr addrspace(3) %tmp402, align 4
420 %tmp404 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 231
421 %tmp405 = load float, ptr addrspace(3) %tmp404, align 4
422 %tmp406 = tail call float @llvm.fmuladd.f32(float %tmp401, float %tmp403, float %tmp405)
423 %tmp407 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 233
424 %tmp408 = load float, ptr addrspace(3) %tmp407, align 4
425 %tmp409 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 234
426 %tmp410 = load float, ptr addrspace(3) %tmp409, align 4
427 %tmp411 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 235
428 %tmp412 = load float, ptr addrspace(3) %tmp411, align 4
429 %tmp413 = tail call float @llvm.fmuladd.f32(float %tmp408, float %tmp410, float %tmp412)
430 %tmp414 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 237
431 %tmp415 = load float, ptr addrspace(3) %tmp414, align 4
432 %tmp416 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 238
433 %tmp417 = load float, ptr addrspace(3) %tmp416, align 4
434 %tmp418 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 239
435 %tmp419 = load float, ptr addrspace(3) %tmp418, align 4
436 %tmp420 = tail call float @llvm.fmuladd.f32(float %tmp415, float %tmp417, float %tmp419)
437 %tmp421 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 241
438 %tmp422 = load float, ptr addrspace(3) %tmp421, align 4
439 %tmp423 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 242
440 %tmp424 = load float, ptr addrspace(3) %tmp423, align 4
441 %tmp425 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 243
442 %tmp426 = load float, ptr addrspace(3) %tmp425, align 4
443 %tmp427 = tail call float @llvm.fmuladd.f32(float %tmp422, float %tmp424, float %tmp426)
444 %tmp428 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 245
445 %tmp429 = load float, ptr addrspace(3) %tmp428, align 4
446 %tmp430 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 246
447 %tmp431 = load float, ptr addrspace(3) %tmp430, align 4
448 %tmp432 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 247
449 %tmp433 = load float, ptr addrspace(3) %tmp432, align 4
450 %tmp434 = tail call float @llvm.fmuladd.f32(float %tmp429, float %tmp431, float %tmp433)
451 %tmp435 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 249
452 %tmp436 = load float, ptr addrspace(3) %tmp435, align 4
453 %tmp437 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 250
454 %tmp438 = load float, ptr addrspace(3) %tmp437, align 4
455 %tmp439 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 251
456 %tmp440 = load float, ptr addrspace(3) %tmp439, align 4
457 %tmp441 = tail call float @llvm.fmuladd.f32(float %tmp436, float %tmp438, float %tmp440)
458 %tmp442 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 253
459 %tmp443 = load float, ptr addrspace(3) %tmp442, align 4
460 %tmp444 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 254
461 %tmp445 = load float, ptr addrspace(3) %tmp444, align 4
462 %tmp446 = getelementptr inbounds float, ptr addrspace(3) %arg, i32 255
463 %tmp447 = load float, ptr addrspace(3) %tmp446, align 4
464 %tmp448 = tail call float @llvm.fmuladd.f32(float %tmp443, float %tmp445, float %tmp447)
465 store float %tmp7, ptr addrspace(1) %arg1, align 4
466 %tmp449 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 1
467 store float %tmp14, ptr addrspace(1) %tmp449, align 4
468 %tmp450 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 2
469 store float %tmp21, ptr addrspace(1) %tmp450, align 4
470 %tmp451 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 3
471 store float %tmp28, ptr addrspace(1) %tmp451, align 4
472 %tmp452 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 4
473 store float %tmp35, ptr addrspace(1) %tmp452, align 4
474 %tmp453 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 5
475 store float %tmp42, ptr addrspace(1) %tmp453, align 4
476 %tmp454 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 6
477 store float %tmp49, ptr addrspace(1) %tmp454, align 4
478 %tmp455 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 7
479 store float %tmp56, ptr addrspace(1) %tmp455, align 4
480 %tmp456 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 8
481 store float %tmp63, ptr addrspace(1) %tmp456, align 4
482 %tmp457 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 9
483 store float %tmp70, ptr addrspace(1) %tmp457, align 4
484 %tmp458 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 10
485 store float %tmp77, ptr addrspace(1) %tmp458, align 4
486 %tmp459 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 11
487 store float %tmp84, ptr addrspace(1) %tmp459, align 4
488 %tmp460 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 12
489 store float %tmp91, ptr addrspace(1) %tmp460, align 4
490 %tmp461 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 13
491 store float %tmp98, ptr addrspace(1) %tmp461, align 4
492 %tmp462 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 14
493 store float %tmp105, ptr addrspace(1) %tmp462, align 4
494 %tmp463 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 15
495 store float %tmp112, ptr addrspace(1) %tmp463, align 4
496 %tmp464 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 16
497 store float %tmp119, ptr addrspace(1) %tmp464, align 4
498 %tmp465 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 17
499 store float %tmp126, ptr addrspace(1) %tmp465, align 4
500 %tmp466 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 18
501 store float %tmp133, ptr addrspace(1) %tmp466, align 4
502 %tmp467 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 19
503 store float %tmp140, ptr addrspace(1) %tmp467, align 4
504 %tmp468 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 20
505 store float %tmp147, ptr addrspace(1) %tmp468, align 4
506 %tmp469 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 21
507 store float %tmp154, ptr addrspace(1) %tmp469, align 4
508 %tmp470 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 22
509 store float %tmp161, ptr addrspace(1) %tmp470, align 4
510 %tmp471 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 23
511 store float %tmp168, ptr addrspace(1) %tmp471, align 4
512 %tmp472 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 24
513 store float %tmp175, ptr addrspace(1) %tmp472, align 4
514 %tmp473 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 25
515 store float %tmp182, ptr addrspace(1) %tmp473, align 4
516 %tmp474 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 26
517 store float %tmp189, ptr addrspace(1) %tmp474, align 4
518 %tmp475 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 27
519 store float %tmp196, ptr addrspace(1) %tmp475, align 4
520 %tmp476 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 28
521 store float %tmp203, ptr addrspace(1) %tmp476, align 4
522 %tmp477 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 29
523 store float %tmp210, ptr addrspace(1) %tmp477, align 4
524 %tmp478 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 30
525 store float %tmp217, ptr addrspace(1) %tmp478, align 4
526 %tmp479 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 31
527 store float %tmp224, ptr addrspace(1) %tmp479, align 4
528 %tmp480 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 32
529 store float %tmp231, ptr addrspace(1) %tmp480, align 4
530 %tmp481 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 33
531 store float %tmp238, ptr addrspace(1) %tmp481, align 4
532 %tmp482 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 34
533 store float %tmp245, ptr addrspace(1) %tmp482, align 4
534 %tmp483 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 35
535 store float %tmp252, ptr addrspace(1) %tmp483, align 4
536 %tmp484 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 36
537 store float %tmp259, ptr addrspace(1) %tmp484, align 4
538 %tmp485 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 37
539 store float %tmp266, ptr addrspace(1) %tmp485, align 4
540 %tmp486 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 38
541 store float %tmp273, ptr addrspace(1) %tmp486, align 4
542 %tmp487 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 39
543 store float %tmp280, ptr addrspace(1) %tmp487, align 4
544 %tmp488 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 40
545 store float %tmp287, ptr addrspace(1) %tmp488, align 4
546 %tmp489 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 41
547 store float %tmp294, ptr addrspace(1) %tmp489, align 4
548 %tmp490 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 42
549 store float %tmp301, ptr addrspace(1) %tmp490, align 4
550 %tmp491 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 43
551 store float %tmp308, ptr addrspace(1) %tmp491, align 4
552 %tmp492 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 44
553 store float %tmp315, ptr addrspace(1) %tmp492, align 4
554 %tmp493 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 45
555 store float %tmp322, ptr addrspace(1) %tmp493, align 4
556 %tmp494 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 46
557 store float %tmp329, ptr addrspace(1) %tmp494, align 4
558 %tmp495 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 47
559 store float %tmp336, ptr addrspace(1) %tmp495, align 4
560 %tmp496 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 48
561 store float %tmp343, ptr addrspace(1) %tmp496, align 4
562 %tmp497 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 49
563 store float %tmp350, ptr addrspace(1) %tmp497, align 4
564 %tmp498 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 50
565 store float %tmp357, ptr addrspace(1) %tmp498, align 4
566 %tmp499 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 51
567 store float %tmp364, ptr addrspace(1) %tmp499, align 4
568 %tmp500 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 52
569 store float %tmp371, ptr addrspace(1) %tmp500, align 4
570 %tmp501 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 53
571 store float %tmp378, ptr addrspace(1) %tmp501, align 4
572 %tmp502 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 54
573 store float %tmp385, ptr addrspace(1) %tmp502, align 4
574 %tmp503 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 55
575 store float %tmp392, ptr addrspace(1) %tmp503, align 4
576 %tmp504 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 56
577 store float %tmp399, ptr addrspace(1) %tmp504, align 4
578 %tmp505 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 57
579 store float %tmp406, ptr addrspace(1) %tmp505, align 4
580 %tmp506 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 58
581 store float %tmp413, ptr addrspace(1) %tmp506, align 4
582 %tmp507 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 59
583 store float %tmp420, ptr addrspace(1) %tmp507, align 4
584 %tmp508 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 60
585 store float %tmp427, ptr addrspace(1) %tmp508, align 4
586 %tmp509 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 61
587 store float %tmp434, ptr addrspace(1) %tmp509, align 4
588 %tmp510 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 62
589 store float %tmp441, ptr addrspace(1) %tmp510, align 4
590 %tmp511 = getelementptr inbounds float, ptr addrspace(1) %arg1, i64 63
591 store float %tmp448, ptr addrspace(1) %tmp511, align 4
595 ; Function Attrs: nounwind readnone
596 declare float @llvm.fmuladd.f32(float, float, float) #0
598 attributes #0 = { nounwind readnone }
599 attributes #1 = { "amdgpu-wave-limiter"="true" }