Lines Matching +full:0 +full:x92
9 ; X86: # %bb.0: # %entry
10 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
11 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
12 ; X86-NEXT: vfmadd132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x98,0xc1]
14 ; X86-NEXT: retl # encoding: [0xc3]
17 ; X64: # %bb.0: # %entry
18 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
19 ; X64-NEXT: vfmadd132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x98,0xc1]
21 ; X64-NEXT: retq # encoding: [0xc3]
23 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %__…
25 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
26 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__A
32 ; X86: # %bb.0: # %entry
33 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
34 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
35 ; X86-NEXT: vfmsub132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x9a,0xc1]
37 ; X86-NEXT: retl # encoding: [0xc3]
40 ; X64: # %bb.0: # %entry
41 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
42 ; X64-NEXT: vfmsub132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x9a,0xc1]
44 ; X64-NEXT: retq # encoding: [0xc3]
47 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %su…
49 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
50 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__A
56 ; X86: # %bb.0: # %entry
57 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
58 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
59 ; X86-NEXT: vfmadd231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xb8,0xd1]
61 ; X86-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
62 ; X86-NEXT: retl # encoding: [0xc3]
65 ; X64: # %bb.0: # %entry
66 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
67 ; X64-NEXT: vfmadd231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xb8,0xd1]
69 ; X64-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
70 ; X64-NEXT: retq # encoding: [0xc3]
72 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %__…
74 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
75 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__C
81 ; X86: # %bb.0: # %entry
82 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
83 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
84 ; X86-NEXT: vfnmadd231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xbc,0xd1]
86 ; X86-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
87 ; X86-NEXT: retl # encoding: [0xc3]
90 ; X64: # %bb.0: # %entry
91 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
92 ; X64-NEXT: vfnmadd231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xbc,0xd1]
94 ; X64-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
95 ; X64-NEXT: retq # encoding: [0xc3]
98 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %sub.i, <2 x double> %__B, <2 x double> %…
100 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
101 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__C
107 ; X86: # %bb.0: # %entry
108 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
109 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
110 ; X86-NEXT: vfmadd213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xa8,0xc2]
112 ; X86-NEXT: retl # encoding: [0xc3]
115 ; X64: # %bb.0: # %entry
116 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
117 ; X64-NEXT: vfmadd213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xa8,0xc2]
119 ; X64-NEXT: retq # encoding: [0xc3]
121 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %__…
123 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
124 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> zeroinitializer
130 ; X86: # %bb.0: # %entry
131 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
132 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
133 ; X86-NEXT: vfmsub213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xaa,0xc2]
135 ; X86-NEXT: retl # encoding: [0xc3]
138 ; X64: # %bb.0: # %entry
139 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
140 ; X64-NEXT: vfmsub213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xaa,0xc2]
142 ; X64-NEXT: retq # encoding: [0xc3]
145 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %su…
147 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
148 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> zeroinitializer
154 ; X86: # %bb.0: # %entry
155 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
156 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
157 ; X86-NEXT: vfnmadd213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xac,0xc…
159 ; X86-NEXT: retl # encoding: [0xc3]
162 ; X64: # %bb.0: # %entry
163 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
164 ; X64-NEXT: vfnmadd213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xac,0xc…
166 ; X64-NEXT: retq # encoding: [0xc3]
169 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %sub.i, <2 x double> %__B, <2 x double> %…
171 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
172 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> zeroinitializer
178 ; X86: # %bb.0: # %entry
179 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
180 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
181 ; X86-NEXT: vfnmsub213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xae,0xc…
183 ; X86-NEXT: retl # encoding: [0xc3]
186 ; X64: # %bb.0: # %entry
187 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
188 ; X64-NEXT: vfnmsub213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xae,0xc…
190 ; X64-NEXT: retq # encoding: [0xc3]
194 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %sub.i, <2 x double> %__B, <2 x double> %…
196 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
197 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> zeroinitializer
203 ; X86: # %bb.0: # %entry
204 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
205 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
206 ; X86-NEXT: vfmadd132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x98,0xc1]
208 ; X86-NEXT: retl # encoding: [0xc3]
211 ; X64: # %bb.0: # %entry
212 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
213 ; X64-NEXT: vfmadd132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x98,0xc1]
215 ; X64-NEXT: retq # encoding: [0xc3]
217 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %__…
219 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
220 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__A
226 ; X86: # %bb.0: # %entry
227 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
228 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
229 ; X86-NEXT: vfmsub132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x9a,0xc1]
231 ; X86-NEXT: retl # encoding: [0xc3]
234 ; X64: # %bb.0: # %entry
235 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
236 ; X64-NEXT: vfmsub132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x9a,0xc1]
238 ; X64-NEXT: retq # encoding: [0xc3]
241 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %su…
243 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
244 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__A
250 ; X86: # %bb.0: # %entry
251 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
252 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
253 ; X86-NEXT: vfmadd231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xb8,0xd1]
255 ; X86-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
256 ; X86-NEXT: retl # encoding: [0xc3]
259 ; X64: # %bb.0: # %entry
260 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
261 ; X64-NEXT: vfmadd231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xb8,0xd1]
263 ; X64-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
264 ; X64-NEXT: retq # encoding: [0xc3]
266 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %__…
268 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
269 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__C
275 ; X86: # %bb.0: # %entry
276 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
277 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
278 ; X86-NEXT: vfnmadd231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xbc,0xd1]
280 ; X86-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
281 ; X86-NEXT: retl # encoding: [0xc3]
284 ; X64: # %bb.0: # %entry
285 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
286 ; X64-NEXT: vfnmadd231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xbc,0xd1]
288 ; X64-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
289 ; X64-NEXT: retq # encoding: [0xc3]
292 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %sub.i, <4 x double> %__B, <4 x double> %…
294 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
295 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__C
301 ; X86: # %bb.0: # %entry
302 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
303 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
304 ; X86-NEXT: vfmadd213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xa8,0xc2]
306 ; X86-NEXT: retl # encoding: [0xc3]
309 ; X64: # %bb.0: # %entry
310 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
311 ; X64-NEXT: vfmadd213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xa8,0xc2]
313 ; X64-NEXT: retq # encoding: [0xc3]
315 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %__…
317 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
318 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> zeroinitializer
324 ; X86: # %bb.0: # %entry
325 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
326 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
327 ; X86-NEXT: vfmsub213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xaa,0xc2]
329 ; X86-NEXT: retl # encoding: [0xc3]
332 ; X64: # %bb.0: # %entry
333 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
334 ; X64-NEXT: vfmsub213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xaa,0xc2]
336 ; X64-NEXT: retq # encoding: [0xc3]
339 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %su…
341 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
342 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> zeroinitializer
348 ; X86: # %bb.0: # %entry
349 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
350 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
351 ; X86-NEXT: vfnmadd213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xac,0xc…
353 ; X86-NEXT: retl # encoding: [0xc3]
356 ; X64: # %bb.0: # %entry
357 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
358 ; X64-NEXT: vfnmadd213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xac,0xc…
360 ; X64-NEXT: retq # encoding: [0xc3]
363 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %sub.i, <4 x double> %__B, <4 x double> %…
365 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
366 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> zeroinitializer
372 ; X86: # %bb.0: # %entry
373 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
374 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
375 ; X86-NEXT: vfnmsub213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xae,0xc…
377 ; X86-NEXT: retl # encoding: [0xc3]
380 ; X64: # %bb.0: # %entry
381 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
382 ; X64-NEXT: vfnmsub213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xae,0xc…
384 ; X64-NEXT: retq # encoding: [0xc3]
388 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %sub.i, <4 x double> %__B, <4 x double> %…
390 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
391 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> zeroinitializer
397 ; X86: # %bb.0: # %entry
398 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
399 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
400 ; X86-NEXT: vfmadd132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x98,0xc1]
402 ; X86-NEXT: retl # encoding: [0xc3]
405 ; X64: # %bb.0: # %entry
406 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
407 ; X64-NEXT: vfmadd132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x98,0xc1]
409 ; X64-NEXT: retq # encoding: [0xc3]
411 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %__C) #9
413 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
414 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__A
420 ; X86: # %bb.0: # %entry
421 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
422 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
423 ; X86-NEXT: vfmsub132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x9a,0xc1]
425 ; X86-NEXT: retl # encoding: [0xc3]
428 ; X64: # %bb.0: # %entry
429 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
430 ; X64-NEXT: vfmsub132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x9a,0xc1]
432 ; X64-NEXT: retq # encoding: [0xc3]
435 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %sub.i)…
437 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
438 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__A
444 ; X86: # %bb.0: # %entry
445 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
446 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
447 ; X86-NEXT: vfmadd231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xb8,0xd1]
449 ; X86-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
450 ; X86-NEXT: retl # encoding: [0xc3]
453 ; X64: # %bb.0: # %entry
454 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
455 ; X64-NEXT: vfmadd231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xb8,0xd1]
457 ; X64-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
458 ; X64-NEXT: retq # encoding: [0xc3]
460 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %__C) #9
462 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
463 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__C
469 ; X86: # %bb.0: # %entry
470 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
471 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
472 ; X86-NEXT: vfnmadd231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xbc,0xd1]
474 ; X86-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
475 ; X86-NEXT: retl # encoding: [0xc3]
478 ; X64: # %bb.0: # %entry
479 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
480 ; X64-NEXT: vfnmadd231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xbc,0xd1]
482 ; X64-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
483 ; X64-NEXT: retq # encoding: [0xc3]
486 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %sub.i, <4 x float> %__B, <4 x float> %__C)…
488 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
489 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__C
495 ; X86: # %bb.0: # %entry
496 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
497 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
498 ; X86-NEXT: vfmadd213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xa8,0xc2]
500 ; X86-NEXT: retl # encoding: [0xc3]
503 ; X64: # %bb.0: # %entry
504 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
505 ; X64-NEXT: vfmadd213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xa8,0xc2]
507 ; X64-NEXT: retq # encoding: [0xc3]
509 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %__C) #9
511 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
512 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> zeroinitializer
518 ; X86: # %bb.0: # %entry
519 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
520 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
521 ; X86-NEXT: vfmsub213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xaa,0xc2]
523 ; X86-NEXT: retl # encoding: [0xc3]
526 ; X64: # %bb.0: # %entry
527 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
528 ; X64-NEXT: vfmsub213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xaa,0xc2]
530 ; X64-NEXT: retq # encoding: [0xc3]
533 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %sub.i)…
535 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
536 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> zeroinitializer
542 ; X86: # %bb.0: # %entry
543 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
544 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
545 ; X86-NEXT: vfnmadd213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xac,0xc…
547 ; X86-NEXT: retl # encoding: [0xc3]
550 ; X64: # %bb.0: # %entry
551 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
552 ; X64-NEXT: vfnmadd213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xac,0xc…
554 ; X64-NEXT: retq # encoding: [0xc3]
557 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %sub.i, <4 x float> %__B, <4 x float> %__C)…
559 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
560 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> zeroinitializer
566 ; X86: # %bb.0: # %entry
567 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
568 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
569 ; X86-NEXT: vfnmsub213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xae,0xc…
571 ; X86-NEXT: retl # encoding: [0xc3]
574 ; X64: # %bb.0: # %entry
575 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
576 ; X64-NEXT: vfnmsub213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xae,0xc…
578 ; X64-NEXT: retq # encoding: [0xc3]
582 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %sub.i, <4 x float> %__B, <4 x float> %sub1…
584 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
585 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> zeroinitializer
591 ; X86: # %bb.0: # %entry
592 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
593 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
594 ; X86-NEXT: vfmadd132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x98,0xc1]
596 ; X86-NEXT: retl # encoding: [0xc3]
599 ; X64: # %bb.0: # %entry
600 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
601 ; X64-NEXT: vfmadd132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x98,0xc1]
603 ; X64-NEXT: retq # encoding: [0xc3]
605 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %__C) #9
607 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__A
613 ; X86: # %bb.0: # %entry
614 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
615 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
616 ; X86-NEXT: vfmsub132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x9a,0xc1]
618 ; X86-NEXT: retl # encoding: [0xc3]
621 ; X64: # %bb.0: # %entry
622 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
623 ; X64-NEXT: vfmsub132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x9a,0xc1]
625 ; X64-NEXT: retq # encoding: [0xc3]
628 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %sub.i)…
630 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__A
636 ; X86: # %bb.0: # %entry
637 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
638 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
639 ; X86-NEXT: vfmadd231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xb8,0xd1]
641 ; X86-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
642 ; X86-NEXT: retl # encoding: [0xc3]
645 ; X64: # %bb.0: # %entry
646 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
647 ; X64-NEXT: vfmadd231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xb8,0xd1]
649 ; X64-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
650 ; X64-NEXT: retq # encoding: [0xc3]
652 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %__C) #9
654 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__C
660 ; X86: # %bb.0: # %entry
661 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
662 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
663 ; X86-NEXT: vfnmadd231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xbc,0xd1]
665 ; X86-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
666 ; X86-NEXT: retl # encoding: [0xc3]
669 ; X64: # %bb.0: # %entry
670 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
671 ; X64-NEXT: vfnmadd231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xbc,0xd1]
673 ; X64-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
674 ; X64-NEXT: retq # encoding: [0xc3]
677 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %sub.i, <8 x float> %__B, <8 x float> %__C)…
679 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__C
685 ; X86: # %bb.0: # %entry
686 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
687 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
688 ; X86-NEXT: vfmadd213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xa8,0xc2]
690 ; X86-NEXT: retl # encoding: [0xc3]
693 ; X64: # %bb.0: # %entry
694 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
695 ; X64-NEXT: vfmadd213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xa8,0xc2]
697 ; X64-NEXT: retq # encoding: [0xc3]
699 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %__C) #9
701 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> zeroinitializer
707 ; X86: # %bb.0: # %entry
708 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
709 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
710 ; X86-NEXT: vfmsub213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xaa,0xc2]
712 ; X86-NEXT: retl # encoding: [0xc3]
715 ; X64: # %bb.0: # %entry
716 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
717 ; X64-NEXT: vfmsub213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xaa,0xc2]
719 ; X64-NEXT: retq # encoding: [0xc3]
722 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %sub.i)…
724 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> zeroinitializer
730 ; X86: # %bb.0: # %entry
731 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
732 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
733 ; X86-NEXT: vfnmadd213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xac,0xc…
735 ; X86-NEXT: retl # encoding: [0xc3]
738 ; X64: # %bb.0: # %entry
739 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
740 ; X64-NEXT: vfnmadd213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xac,0xc…
742 ; X64-NEXT: retq # encoding: [0xc3]
745 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %sub.i, <8 x float> %__B, <8 x float> %__C)…
747 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> zeroinitializer
753 ; X86: # %bb.0: # %entry
754 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
755 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
756 ; X86-NEXT: vfnmsub213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xae,0xc…
758 ; X86-NEXT: retl # encoding: [0xc3]
761 ; X64: # %bb.0: # %entry
762 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
763 ; X64-NEXT: vfnmsub213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xae,0xc…
765 ; X64-NEXT: retq # encoding: [0xc3]
769 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %sub.i, <8 x float> %__B, <8 x float> %sub1…
771 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> zeroinitializer
777 ; X86: # %bb.0: # %entry
778 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
779 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
780 ; X86-NEXT: vfmaddsub132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x96,0xc1]
782 ; X86-NEXT: retl # encoding: [0xc3]
785 ; X64: # %bb.0: # %entry
786 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
787 ; X64-NEXT: vfmaddsub132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x96,0xc1]
789 ; X64-NEXT: retq # encoding: [0xc3]
791 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %__…
794 %3 = shufflevector <2 x double> %2, <2 x double> %0, <2 x i32> <i32 0, i32 3>
796 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
803 ; X86: # %bb.0: # %entry
804 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
805 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
806 ; X86-NEXT: vfmsubadd132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x97,0xc1]
808 ; X86-NEXT: retl # encoding: [0xc3]
811 ; X64: # %bb.0: # %entry
812 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
813 ; X64-NEXT: vfmsubadd132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x97,0xc1]
815 ; X64-NEXT: retq # encoding: [0xc3]
818 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %su…
820 %2 = shufflevector <2 x double> %1, <2 x double> %0, <2 x i32> <i32 0, i32 3>
822 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
829 ; X86: # %bb.0: # %entry
830 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
831 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
832 ; X86-NEXT: vfmaddsub231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xb6,0xd1]
834 ; X86-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
835 ; X86-NEXT: retl # encoding: [0xc3]
838 ; X64: # %bb.0: # %entry
839 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
840 ; X64-NEXT: vfmaddsub231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xb6,0xd1]
842 ; X64-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
843 ; X64-NEXT: retq # encoding: [0xc3]
845 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %__…
848 %3 = shufflevector <2 x double> %2, <2 x double> %0, <2 x i32> <i32 0, i32 3>
850 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
857 ; X86: # %bb.0: # %entry
858 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
859 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
860 ; X86-NEXT: vfmaddsub213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xa6,0…
862 ; X86-NEXT: retl # encoding: [0xc3]
865 ; X64: # %bb.0: # %entry
866 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
867 ; X64-NEXT: vfmaddsub213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xa6,0…
869 ; X64-NEXT: retq # encoding: [0xc3]
871 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %__…
874 %3 = shufflevector <2 x double> %2, <2 x double> %0, <2 x i32> <i32 0, i32 3>
876 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
883 ; X86: # %bb.0: # %entry
884 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
885 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
886 ; X86-NEXT: vfmsubadd213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xa7,0…
888 ; X86-NEXT: retl # encoding: [0xc3]
891 ; X64: # %bb.0: # %entry
892 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
893 ; X64-NEXT: vfmsubadd213pd %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0x89,0xa7,0…
895 ; X64-NEXT: retq # encoding: [0xc3]
898 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %su…
900 %2 = shufflevector <2 x double> %1, <2 x double> %0, <2 x i32> <i32 0, i32 3>
902 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
909 ; X86: # %bb.0: # %entry
910 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
911 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
912 ; X86-NEXT: vfmaddsub132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x96,0xc1]
914 ; X86-NEXT: retl # encoding: [0xc3]
917 ; X64: # %bb.0: # %entry
918 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
919 ; X64-NEXT: vfmaddsub132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x96,0xc1]
921 ; X64-NEXT: retq # encoding: [0xc3]
923 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %__…
926 %3 = shufflevector <4 x double> %2, <4 x double> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
928 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
935 ; X86: # %bb.0: # %entry
936 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
937 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
938 ; X86-NEXT: vfmsubadd132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x97,0xc1]
940 ; X86-NEXT: retl # encoding: [0xc3]
943 ; X64: # %bb.0: # %entry
944 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
945 ; X64-NEXT: vfmsubadd132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x97,0xc1]
947 ; X64-NEXT: retq # encoding: [0xc3]
950 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %su…
952 %2 = shufflevector <4 x double> %1, <4 x double> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
954 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
961 ; X86: # %bb.0: # %entry
962 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
963 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
964 ; X86-NEXT: vfmaddsub231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xb6,0xd1]
966 ; X86-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
967 ; X86-NEXT: retl # encoding: [0xc3]
970 ; X64: # %bb.0: # %entry
971 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
972 ; X64-NEXT: vfmaddsub231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xb6,0xd1]
974 ; X64-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
975 ; X64-NEXT: retq # encoding: [0xc3]
977 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %__…
980 %3 = shufflevector <4 x double> %2, <4 x double> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
982 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
989 ; X86: # %bb.0: # %entry
990 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
991 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
992 ; X86-NEXT: vfmaddsub213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xa6,0…
994 ; X86-NEXT: retl # encoding: [0xc3]
997 ; X64: # %bb.0: # %entry
998 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
999 ; X64-NEXT: vfmaddsub213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xa6,0…
1001 ; X64-NEXT: retq # encoding: [0xc3]
1003 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %__…
1006 %3 = shufflevector <4 x double> %2, <4 x double> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1008 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1015 ; X86: # %bb.0: # %entry
1016 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1017 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1018 ; X86-NEXT: vfmsubadd213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xa7,0…
1020 ; X86-NEXT: retl # encoding: [0xc3]
1023 ; X64: # %bb.0: # %entry
1024 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1025 ; X64-NEXT: vfmsubadd213pd %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xf5,0xa9,0xa7,0…
1027 ; X64-NEXT: retq # encoding: [0xc3]
1030 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %su…
1032 %2 = shufflevector <4 x double> %1, <4 x double> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1034 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1041 ; X86: # %bb.0: # %entry
1042 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1043 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1044 ; X86-NEXT: vfmaddsub132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x96,0xc1]
1046 ; X86-NEXT: retl # encoding: [0xc3]
1049 ; X64: # %bb.0: # %entry
1050 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1051 ; X64-NEXT: vfmaddsub132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x96,0xc1]
1053 ; X64-NEXT: retq # encoding: [0xc3]
1055 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %__C) #9
1058 %3 = shufflevector <4 x float> %2, <4 x float> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1060 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1067 ; X86: # %bb.0: # %entry
1068 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1069 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1070 ; X86-NEXT: vfmsubadd132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x97,0xc1]
1072 ; X86-NEXT: retl # encoding: [0xc3]
1075 ; X64: # %bb.0: # %entry
1076 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1077 ; X64-NEXT: vfmsubadd132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x97,0xc1]
1079 ; X64-NEXT: retq # encoding: [0xc3]
1082 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %sub.i)…
1084 %2 = shufflevector <4 x float> %1, <4 x float> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1086 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1093 ; X86: # %bb.0: # %entry
1094 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1095 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1096 ; X86-NEXT: vfmaddsub231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xb6,0xd1]
1098 ; X86-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1099 ; X86-NEXT: retl # encoding: [0xc3]
1102 ; X64: # %bb.0: # %entry
1103 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1104 ; X64-NEXT: vfmaddsub231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xb6,0xd1]
1106 ; X64-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1107 ; X64-NEXT: retq # encoding: [0xc3]
1109 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %__C) #9
1112 %3 = shufflevector <4 x float> %2, <4 x float> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1114 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1121 ; X86: # %bb.0: # %entry
1122 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1123 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1124 ; X86-NEXT: vfmaddsub213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xa6,0…
1126 ; X86-NEXT: retl # encoding: [0xc3]
1129 ; X64: # %bb.0: # %entry
1130 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1131 ; X64-NEXT: vfmaddsub213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xa6,0…
1133 ; X64-NEXT: retq # encoding: [0xc3]
1135 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %__C) #9
1138 %3 = shufflevector <4 x float> %2, <4 x float> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1140 %extract.i = shufflevector <8 x i1> %4, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1147 ; X86: # %bb.0: # %entry
1148 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1149 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1150 ; X86-NEXT: vfmsubadd213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xa7,0…
1152 ; X86-NEXT: retl # encoding: [0xc3]
1155 ; X64: # %bb.0: # %entry
1156 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1157 ; X64-NEXT: vfmsubadd213ps %xmm2, %xmm1, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0xa7,0…
1159 ; X64-NEXT: retq # encoding: [0xc3]
1162 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %sub.i)…
1164 %2 = shufflevector <4 x float> %1, <4 x float> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1166 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1173 ; X86: # %bb.0: # %entry
1174 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1175 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1176 ; X86-NEXT: vfmaddsub132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x96,0xc1]
1178 ; X86-NEXT: retl # encoding: [0xc3]
1181 ; X64: # %bb.0: # %entry
1182 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1183 ; X64-NEXT: vfmaddsub132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x96,0xc1]
1185 ; X64-NEXT: retq # encoding: [0xc3]
1187 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %__C) #9
1190 …%3 = shufflevector <8 x float> %2, <8 x float> %0, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, …
1198 ; X86: # %bb.0: # %entry
1199 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1200 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1201 ; X86-NEXT: vfmsubadd132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x97,0xc1]
1203 ; X86-NEXT: retl # encoding: [0xc3]
1206 ; X64: # %bb.0: # %entry
1207 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1208 ; X64-NEXT: vfmsubadd132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x97,0xc1]
1210 ; X64-NEXT: retq # encoding: [0xc3]
1213 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %sub.i)…
1215 …%2 = shufflevector <8 x float> %1, <8 x float> %0, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, …
1223 ; X86: # %bb.0: # %entry
1224 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1225 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1226 ; X86-NEXT: vfmaddsub231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xb6,0xd1]
1228 ; X86-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1229 ; X86-NEXT: retl # encoding: [0xc3]
1232 ; X64: # %bb.0: # %entry
1233 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1234 ; X64-NEXT: vfmaddsub231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xb6,0xd1]
1236 ; X64-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1237 ; X64-NEXT: retq # encoding: [0xc3]
1239 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %__C) #9
1242 …%3 = shufflevector <8 x float> %2, <8 x float> %0, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, …
1250 ; X86: # %bb.0: # %entry
1251 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1252 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1253 ; X86-NEXT: vfmaddsub213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xa6,0…
1255 ; X86-NEXT: retl # encoding: [0xc3]
1258 ; X64: # %bb.0: # %entry
1259 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1260 ; X64-NEXT: vfmaddsub213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xa6,0…
1262 ; X64-NEXT: retq # encoding: [0xc3]
1264 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %__C) #9
1267 …%3 = shufflevector <8 x float> %2, <8 x float> %0, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, …
1275 ; X86: # %bb.0: # %entry
1276 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1277 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1278 ; X86-NEXT: vfmsubadd213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xa7,0…
1280 ; X86-NEXT: retl # encoding: [0xc3]
1283 ; X64: # %bb.0: # %entry
1284 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1285 ; X64-NEXT: vfmsubadd213ps %ymm2, %ymm1, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0xa7,0…
1287 ; X64-NEXT: retq # encoding: [0xc3]
1290 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %sub.i)…
1292 …%2 = shufflevector <8 x float> %1, <8 x float> %0, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, …
1300 ; X86: # %bb.0: # %entry
1301 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1302 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1303 ; X86-NEXT: vfmsub231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xba,0xd1]
1305 ; X86-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
1306 ; X86-NEXT: retl # encoding: [0xc3]
1309 ; X64: # %bb.0: # %entry
1310 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1311 ; X64-NEXT: vfmsub231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xba,0xd1]
1313 ; X64-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
1314 ; X64-NEXT: retq # encoding: [0xc3]
1317 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %su…
1319 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
1320 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__C
1326 ; X86: # %bb.0: # %entry
1327 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1328 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1329 ; X86-NEXT: vfmsub231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xba,0xd1]
1331 ; X86-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
1332 ; X86-NEXT: retl # encoding: [0xc3]
1335 ; X64: # %bb.0: # %entry
1336 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1337 ; X64-NEXT: vfmsub231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xba,0xd1]
1339 ; X64-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
1340 ; X64-NEXT: retq # encoding: [0xc3]
1343 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %su…
1345 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1346 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__C
1352 ; X86: # %bb.0: # %entry
1353 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1354 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1355 ; X86-NEXT: vfmsub231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xba,0xd1]
1357 ; X86-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1358 ; X86-NEXT: retl # encoding: [0xc3]
1361 ; X64: # %bb.0: # %entry
1362 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1363 ; X64-NEXT: vfmsub231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xba,0xd1]
1365 ; X64-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1366 ; X64-NEXT: retq # encoding: [0xc3]
1369 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %sub.i)…
1371 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1372 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__C
1378 ; X86: # %bb.0: # %entry
1379 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1380 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1381 ; X86-NEXT: vfmsub231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xba,0xd1]
1383 ; X86-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1384 ; X86-NEXT: retl # encoding: [0xc3]
1387 ; X64: # %bb.0: # %entry
1388 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1389 ; X64-NEXT: vfmsub231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xba,0xd1]
1391 ; X64-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1392 ; X64-NEXT: retq # encoding: [0xc3]
1395 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %sub.i)…
1397 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__C
1403 ; X86: # %bb.0: # %entry
1404 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1405 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1406 ; X86-NEXT: vfmsubadd231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xb7,0xd1]
1408 ; X86-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
1409 ; X86-NEXT: retl # encoding: [0xc3]
1412 ; X64: # %bb.0: # %entry
1413 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1414 ; X64-NEXT: vfmsubadd231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xb7,0xd1]
1416 ; X64-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
1417 ; X64-NEXT: retq # encoding: [0xc3]
1420 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %__B, <2 x double> %su…
1422 %2 = shufflevector <2 x double> %1, <2 x double> %0, <2 x i32> <i32 0, i32 3>
1424 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
1431 ; X86: # %bb.0: # %entry
1432 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1433 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1434 ; X86-NEXT: vfmsubadd231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xb7,0xd1]
1436 ; X86-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
1437 ; X86-NEXT: retl # encoding: [0xc3]
1440 ; X64: # %bb.0: # %entry
1441 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1442 ; X64-NEXT: vfmsubadd231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xb7,0xd1]
1444 ; X64-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
1445 ; X64-NEXT: retq # encoding: [0xc3]
1448 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %__B, <4 x double> %su…
1450 %2 = shufflevector <4 x double> %1, <4 x double> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1452 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1459 ; X86: # %bb.0: # %entry
1460 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1461 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1462 ; X86-NEXT: vfmsubadd231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xb7,0xd1]
1464 ; X86-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1465 ; X86-NEXT: retl # encoding: [0xc3]
1468 ; X64: # %bb.0: # %entry
1469 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1470 ; X64-NEXT: vfmsubadd231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xb7,0xd1]
1472 ; X64-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1473 ; X64-NEXT: retq # encoding: [0xc3]
1476 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %__B, <4 x float> %sub.i)…
1478 %2 = shufflevector <4 x float> %1, <4 x float> %0, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
1480 %extract.i = shufflevector <8 x i1> %3, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1487 ; X86: # %bb.0: # %entry
1488 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1489 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1490 ; X86-NEXT: vfmsubadd231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xb7,0xd1]
1492 ; X86-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1493 ; X86-NEXT: retl # encoding: [0xc3]
1496 ; X64: # %bb.0: # %entry
1497 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1498 ; X64-NEXT: vfmsubadd231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xb7,0xd1]
1500 ; X64-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1501 ; X64-NEXT: retq # encoding: [0xc3]
1504 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %__B, <8 x float> %sub.i)…
1506 …%2 = shufflevector <8 x float> %1, <8 x float> %0, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, …
1514 ; X86: # %bb.0: # %entry
1515 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1516 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1517 ; X86-NEXT: vfnmadd132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x9c,0xc1]
1519 ; X86-NEXT: retl # encoding: [0xc3]
1522 ; X64: # %bb.0: # %entry
1523 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1524 ; X64-NEXT: vfnmadd132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x9c,0xc1]
1526 ; X64-NEXT: retq # encoding: [0xc3]
1529 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %sub.i, <2 x double> %…
1531 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
1532 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__A
1538 ; X86: # %bb.0: # %entry
1539 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1540 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1541 ; X86-NEXT: vfnmadd132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x9c,0xc1]
1543 ; X86-NEXT: retl # encoding: [0xc3]
1546 ; X64: # %bb.0: # %entry
1547 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1548 ; X64-NEXT: vfnmadd132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x9c,0xc1]
1550 ; X64-NEXT: retq # encoding: [0xc3]
1553 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %sub.i, <4 x double> %…
1555 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1556 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__A
1562 ; X86: # %bb.0: # %entry
1563 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1564 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1565 ; X86-NEXT: vfnmadd132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x9c,0xc1]
1567 ; X86-NEXT: retl # encoding: [0xc3]
1570 ; X64: # %bb.0: # %entry
1571 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1572 ; X64-NEXT: vfnmadd132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x9c,0xc1]
1574 ; X64-NEXT: retq # encoding: [0xc3]
1577 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %sub.i, <4 x float> %__C)…
1579 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1580 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__A
1586 ; X86: # %bb.0: # %entry
1587 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1588 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1589 ; X86-NEXT: vfnmadd132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x9c,0xc1]
1591 ; X86-NEXT: retl # encoding: [0xc3]
1594 ; X64: # %bb.0: # %entry
1595 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1596 ; X64-NEXT: vfnmadd132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x9c,0xc1]
1598 ; X64-NEXT: retq # encoding: [0xc3]
1601 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %sub.i, <8 x float> %__C)…
1603 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__A
1609 ; X86: # %bb.0: # %entry
1610 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1611 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1612 ; X86-NEXT: vfnmsub132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x9e,0xc1]
1614 ; X86-NEXT: retl # encoding: [0xc3]
1617 ; X64: # %bb.0: # %entry
1618 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1619 ; X64-NEXT: vfnmsub132pd %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0xed,0x09,0x9e,0xc1]
1621 ; X64-NEXT: retq # encoding: [0xc3]
1625 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %sub.i, <2 x double> %…
1627 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
1628 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__A
1634 ; X86: # %bb.0: # %entry
1635 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1636 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1637 ; X86-NEXT: vfnmsub231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xbe,0xd1]
1639 ; X86-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
1640 ; X86-NEXT: retl # encoding: [0xc3]
1643 ; X64: # %bb.0: # %entry
1644 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1645 ; X64-NEXT: vfnmsub231pd %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0xbe,0xd1]
1647 ; X64-NEXT: vmovapd %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc2]
1648 ; X64-NEXT: retq # encoding: [0xc3]
1652 …%0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %__A, <2 x double> %sub.i, <2 x double> %…
1654 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <2 x i32> <i32 0, i32 1>
1655 %2 = select <2 x i1> %extract.i, <2 x double> %0, <2 x double> %__C
1661 ; X86: # %bb.0: # %entry
1662 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1663 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1664 ; X86-NEXT: vfnmsub132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x9e,0xc1]
1666 ; X86-NEXT: retl # encoding: [0xc3]
1669 ; X64: # %bb.0: # %entry
1670 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1671 ; X64-NEXT: vfnmsub132pd %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0xed,0x29,0x9e,0xc1]
1673 ; X64-NEXT: retq # encoding: [0xc3]
1677 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %sub.i, <4 x double> %…
1679 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1680 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__A
1686 ; X86: # %bb.0: # %entry
1687 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1688 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1689 ; X86-NEXT: vfnmsub231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xbe,0xd1]
1691 ; X86-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
1692 ; X86-NEXT: retl # encoding: [0xc3]
1695 ; X64: # %bb.0: # %entry
1696 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1697 ; X64-NEXT: vfnmsub231pd %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0xbe,0xd1]
1699 ; X64-NEXT: vmovapd %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc2]
1700 ; X64-NEXT: retq # encoding: [0xc3]
1704 …%0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %__A, <4 x double> %sub.i, <4 x double> %…
1706 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1707 %2 = select <4 x i1> %extract.i, <4 x double> %0, <4 x double> %__C
1713 ; X86: # %bb.0: # %entry
1714 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1715 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1716 ; X86-NEXT: vfnmsub132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x9e,0xc1]
1718 ; X86-NEXT: retl # encoding: [0xc3]
1721 ; X64: # %bb.0: # %entry
1722 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1723 ; X64-NEXT: vfnmsub132ps %xmm1, %xmm2, %xmm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x09,0x9e,0xc1]
1725 ; X64-NEXT: retq # encoding: [0xc3]
1729 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %sub.i, <4 x float> %sub1…
1731 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1732 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__A
1738 ; X86: # %bb.0: # %entry
1739 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1740 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1741 ; X86-NEXT: vfnmsub231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xbe,0xd1]
1743 ; X86-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1744 ; X86-NEXT: retl # encoding: [0xc3]
1747 ; X64: # %bb.0: # %entry
1748 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1749 ; X64-NEXT: vfnmsub231ps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x09,0xbe,0xd1]
1751 ; X64-NEXT: vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
1752 ; X64-NEXT: retq # encoding: [0xc3]
1756 …%0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %__A, <4 x float> %sub.i, <4 x float> %sub1…
1758 %extract.i = shufflevector <8 x i1> %1, <8 x i1> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
1759 %2 = select <4 x i1> %extract.i, <4 x float> %0, <4 x float> %__C
1765 ; X86: # %bb.0: # %entry
1766 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1767 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1768 ; X86-NEXT: vfnmsub132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x9e,0xc1]
1770 ; X86-NEXT: retl # encoding: [0xc3]
1773 ; X64: # %bb.0: # %entry
1774 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1775 ; X64-NEXT: vfnmsub132ps %ymm1, %ymm2, %ymm0 {%k1} # encoding: [0x62,0xf2,0x6d,0x29,0x9e,0xc1]
1777 ; X64-NEXT: retq # encoding: [0xc3]
1781 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %sub.i, <8 x float> %sub1…
1783 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__A
1789 ; X86: # %bb.0: # %entry
1790 ; X86-NEXT: movzbl {{[0-9]+}}(%esp), %eax # encoding: [0x0f,0xb6,0x44,0x24,0x04]
1791 ; X86-NEXT: kmovw %eax, %k1 # encoding: [0xc5,0xf8,0x92,0xc8]
1792 ; X86-NEXT: vfnmsub231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xbe,0xd1]
1794 ; X86-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1795 ; X86-NEXT: retl # encoding: [0xc3]
1798 ; X64: # %bb.0: # %entry
1799 ; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
1800 ; X64-NEXT: vfnmsub231ps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0x7d,0x29,0xbe,0xd1]
1802 ; X64-NEXT: vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
1803 ; X64-NEXT: retq # encoding: [0xc3]
1807 …%0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %__A, <8 x float> %sub.i, <8 x float> %sub1…
1809 %2 = select <8 x i1> %1, <8 x float> %0, <8 x float> %__C