1 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512fp16 --fp-contract=fast --enable-unsafe-fp-math | FileCheck %s
4 define dso_local <32 x half> @test1(<32 x half> %acc.coerce, <32 x half> %lhs.coerce.conj, <32 x half> %rhs.coerce) local_unnamed_addr #0 {
6 ; CHECK: # %bb.0: # %entry
7 ; CHECK-NEXT: vfcmaddcph %zmm1, %zmm2, %zmm0
10 %0 = bitcast <32 x half> %lhs.coerce.conj to <16 x i32>
11 %xor.i.i = xor <16 x i32> %0, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
12 %1 = bitcast <16 x i32> %xor.i.i to <16 x float>
13 %2 = bitcast <32 x half> %rhs.coerce to <16 x float>
14 %3 = tail call fast <16 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.512(<16 x float> %1, <16 x float> %2, <16 x float> zeroinitializer, i16 -1, i32 4) #2
15 %4 = bitcast <16 x float> %3 to <32 x half>
16 %add = fadd fast <32 x half> %4, %acc.coerce
20 define dso_local <32 x half> @test2(<32 x half> %acc.coerce, <32 x half> %lhs.coerce.conj, <32 x half> %rhs.coerce) local_unnamed_addr #0 {
22 ; CHECK: # %bb.0: # %entry
23 ; CHECK-NEXT: vfcmaddcph %zmm1, %zmm2, %zmm0
26 %0 = bitcast <32 x half> %lhs.coerce.conj to <16 x i32>
27 %xor.i.i = xor <16 x i32> %0, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
28 %1 = bitcast <16 x i32> %xor.i.i to <16 x float>
29 %2 = bitcast <32 x half> %rhs.coerce to <16 x float>
30 %3 = tail call fast <16 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.512(<16 x float> %2, <16 x float> %1, <16 x float> zeroinitializer, i16 -1, i32 4) #2
31 %4 = bitcast <16 x float> %3 to <32 x half>
32 %add = fadd fast <32 x half> %4, %acc.coerce
36 define dso_local <16 x half> @test3(<16 x half> %acc.coerce, <16 x half> %lhs.coerce.conj, <16 x half> %rhs.coerce) local_unnamed_addr #0 {
38 ; CHECK: # %bb.0: # %entry
39 ; CHECK-NEXT: vfcmaddcph %ymm1, %ymm2, %ymm0
42 %0 = bitcast <16 x half> %lhs.coerce.conj to <8 x i32>
43 %xor.i.i = xor <8 x i32> %0, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
44 %1 = bitcast <8 x i32> %xor.i.i to <8 x float>
45 %2 = bitcast <16 x half> %rhs.coerce to <8 x float>
46 %3 = tail call fast <8 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.256(<8 x float> %1, <8 x float> %2, <8 x float> zeroinitializer, i8 -1) #2
47 %4 = bitcast <8 x float> %3 to <16 x half>
48 %add = fadd fast <16 x half> %4, %acc.coerce
52 define dso_local <8 x half> @test4(<8 x half> %acc.coerce, <8 x half> %lhs.coerce.conj, <8 x half> %rhs.coerce) local_unnamed_addr #0 {
54 ; CHECK: # %bb.0: # %entry
55 ; CHECK-NEXT: vfcmaddcph %xmm1, %xmm2, %xmm0
58 %0 = bitcast <8 x half> %lhs.coerce.conj to <4 x i32>
59 %xor.i.i = xor <4 x i32> %0, <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>
60 %1 = bitcast <4 x i32> %xor.i.i to <4 x float>
61 %2 = bitcast <8 x half> %rhs.coerce to <4 x float>
62 %3 = tail call fast <4 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.128(<4 x float> %1, <4 x float> %2, <4 x float> zeroinitializer, i8 -1) #2
63 %4 = bitcast <4 x float> %3 to <8 x half>
64 %add = fadd fast <8 x half> %4, %acc.coerce
68 define dso_local <8 x half> @test5(<8 x half> %acc.coerce, <8 x half> %lhs.coerce.conj, <8 x half> %rhs.coerce) local_unnamed_addr #0 {
70 ; CHECK: # %bb.0: # %entry
71 ; CHECK-NEXT: vfcmaddcph %xmm1, %xmm2, %xmm0
74 %0 = bitcast <8 x half> %lhs.coerce.conj to <4 x i32>
75 %xor.i.i = xor <4 x i32> <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>, %0
76 %1 = bitcast <4 x i32> %xor.i.i to <4 x float>
77 %2 = bitcast <8 x half> %rhs.coerce to <4 x float>
78 %3 = tail call fast <4 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.128(<4 x float> %1, <4 x float> %2, <4 x float> zeroinitializer, i8 -1) #2
79 %4 = bitcast <4 x float> %3 to <8 x half>
80 %add = fadd fast <8 x half> %4, %acc.coerce
84 define dso_local <8 x half> @test6(<8 x half> %acc.coerce, <8 x half> %lhs.coerce.conj, <8 x half> %rhs.coerce) local_unnamed_addr #0 {
86 ; CHECK: # %bb.0: # %entry
87 ; CHECK-NEXT: vxorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm1, %xmm1
88 ; CHECK-NEXT: vfmaddcph %xmm2, %xmm1, %xmm0
91 %0 = bitcast <8 x half> %lhs.coerce.conj to <4 x i32>
92 %xor.i.i = xor <4 x i32> <i32 1, i32 1, i32 1, i32 1>, %0
93 %1 = bitcast <4 x i32> %xor.i.i to <4 x float>
94 %2 = bitcast <8 x half> %rhs.coerce to <4 x float>
95 %3 = tail call fast <4 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.128(<4 x float> %1, <4 x float> %2, <4 x float> zeroinitializer, i8 -1) #2
96 %4 = bitcast <4 x float> %3 to <8 x half>
97 %add = fadd fast <8 x half> %4, %acc.coerce
101 define dso_local <8 x half> @test7(<8 x half> %acc.coerce, <8 x half> %lhs.coerce.conj, <8 x half> %rhs.coerce) local_unnamed_addr #0 {
102 ; CHECK-LABEL: test7:
103 ; CHECK: # %bb.0: # %entry
104 ; CHECK-NEXT: vfcmaddcph %xmm1, %xmm2, %xmm0
107 %0 = bitcast <8 x half> %lhs.coerce.conj to <4 x i32>
108 %xor.i.i = xor <4 x i32> <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>, %0
109 %1 = bitcast <4 x i32> %xor.i.i to <4 x float>
110 %2 = bitcast <8 x half> %rhs.coerce to <4 x float>
111 %3 = tail call fast <4 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.128(<4 x float> %1, <4 x float> %2, <4 x float> zeroinitializer, i8 -1) #2
112 %4 = bitcast <4 x float> %3 to <8 x half>
113 %add = fadd fast <8 x half> %acc.coerce, %4
117 define dso_local <8 x half> @test8(<8 x half> %acc.coerce, <4 x float> %lhs.coerce.conj, <8 x half> %rhs.coerce) local_unnamed_addr #0 {
118 ; CHECK-LABEL: test8:
119 ; CHECK: # %bb.0: # %entry
120 ; CHECK-NEXT: vfcmaddcph %xmm1, %xmm2, %xmm0
123 %0 = bitcast <4 x float> %lhs.coerce.conj to <4 x i32>
124 %xor.i.i = xor <4 x i32> <i32 -2147483648, i32 -2147483648, i32 -2147483648, i32 -2147483648>, %0
125 %1 = bitcast <4 x i32> %xor.i.i to <4 x float>
126 %2 = bitcast <8 x half> %rhs.coerce to <4 x float>
127 %3 = tail call fast <4 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.128(<4 x float> %1, <4 x float> %2, <4 x float> zeroinitializer, i8 -1) #2
128 %4 = bitcast <4 x float> %3 to <8 x half>
129 %add = fadd fast <8 x half> %acc.coerce, %4
133 define dso_local <32 x half> @test9(<32 x half> %acc.coerce, <8 x i64> %lhs.coerce.conj, <32 x half> %rhs.coerce) local_unnamed_addr #0 {
134 ; CHECK-LABEL: test9:
135 ; CHECK: # %bb.0: # %entry
136 ; CHECK-NEXT: vfcmaddcph %zmm1, %zmm2, %zmm0
139 %xor1.i = xor <8 x i64> %lhs.coerce.conj, <i64 -9223372034707292160, i64 -9223372034707292160, i64 -9223372034707292160, i64 -9223372034707292160, i64 -9223372034707292160, i64 -9223372034707292160, i64 -9223372034707292160, i64 -9223372034707292160>
140 %0 = bitcast <8 x i64> %xor1.i to <16 x float>
141 %1 = bitcast <32 x half> %rhs.coerce to <16 x float>
142 %2 = tail call fast <16 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.512(<16 x float> %0, <16 x float> %1, <16 x float> zeroinitializer, i16 -1, i32 4) #2
143 %3 = bitcast <16 x float> %2 to <32 x half>
144 %add = fadd fast <32 x half> %3, %acc.coerce
148 declare <16 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.512(<16 x float>, <16 x float>, <16 x float>, i16, i32 immarg)
149 declare <8 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.256(<8 x float>, <8 x float>, <8 x float>, i8)
150 declare <4 x float> @llvm.x86.avx512fp16.mask.vfmul.cph.128(<4 x float>, <4 x float>, <4 x float>, i8)