1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
|
; RUN: llc < %s -march=arm -mattr=+neon,+vfp4 | FileCheck %s
; Check generated fused MAC and MLS.
define double @fusedMACTest1(double %d1, double %d2, double %d3) {
;CHECK: fusedMACTest1:
;CHECK: vfma.f64
%1 = fmul double %d1, %d2
%2 = fadd double %1, %d3
ret double %2
}
define float @fusedMACTest2(float %f1, float %f2, float %f3) {
;CHECK: fusedMACTest2:
;CHECK: vfma.f32
%1 = fmul float %f1, %f2
%2 = fadd float %1, %f3
ret float %2
}
define double @fusedMACTest3(double %d1, double %d2, double %d3) {
;CHECK: fusedMACTest3:
;CHECK: vfms.f64
%1 = fmul double %d2, %d3
%2 = fsub double %d1, %1
ret double %2
}
define float @fusedMACTest4(float %f1, float %f2, float %f3) {
;CHECK: fusedMACTest4:
;CHECK: vfms.f32
%1 = fmul float %f2, %f3
%2 = fsub float %f1, %1
ret float %2
}
define double @fusedMACTest5(double %d1, double %d2, double %d3) {
;CHECK: fusedMACTest5:
;CHECK: vfnma.f64
%1 = fmul double %d1, %d2
%2 = fsub double -0.0, %1
%3 = fsub double %2, %d3
ret double %3
}
define float @fusedMACTest6(float %f1, float %f2, float %f3) {
;CHECK: fusedMACTest6:
;CHECK: vfnma.f32
%1 = fmul float %f1, %f2
%2 = fsub float -0.0, %1
%3 = fsub float %2, %f3
ret float %3
}
define double @fusedMACTest7(double %d1, double %d2, double %d3) {
;CHECK: fusedMACTest7:
;CHECK: vfnms.f64
%1 = fmul double %d1, %d2
%2 = fsub double %1, %d3
ret double %2
}
define float @fusedMACTest8(float %f1, float %f2, float %f3) {
;CHECK: fusedMACTest8:
;CHECK: vfnms.f32
%1 = fmul float %f1, %f2
%2 = fsub float %1, %f3
ret float %2
}
define <2 x float> @fusedMACTest9(<2 x float> %a, <2 x float> %b) {
;CHECK: fusedMACTest9:
;CHECK: vfma.f32
%mul = fmul <2 x float> %a, %b
%add = fadd <2 x float> %mul, %a
ret <2 x float> %add
}
define <2 x float> @fusedMACTest10(<2 x float> %a, <2 x float> %b) {
;CHECK: fusedMACTest10:
;CHECK: vfms.f32
%mul = fmul <2 x float> %a, %b
%sub = fsub <2 x float> %a, %mul
ret <2 x float> %sub
}
define <4 x float> @fusedMACTest11(<4 x float> %a, <4 x float> %b) {
;CHECK: fusedMACTest11:
;CHECK: vfma.f32
%mul = fmul <4 x float> %a, %b
%add = fadd <4 x float> %mul, %a
ret <4 x float> %add
}
define <4 x float> @fusedMACTest12(<4 x float> %a, <4 x float> %b) {
;CHECK: fusedMACTest12:
;CHECK: vfms.f32
%mul = fmul <4 x float> %a, %b
%sub = fsub <4 x float> %a, %mul
ret <4 x float> %sub
}
define float @test_fma_f32(float %a, float %b, float %c) nounwind readnone ssp {
entry:
; CHECK: test_fma_f32
; CHECK: vfma.f32
%call = tail call float @llvm.fma.f32(float %a, float %b, float %c) nounwind readnone
ret float %call
}
define double @test_fma_f64(double %a, double %b, double %c) nounwind readnone ssp {
entry:
; CHECK: test_fma_f64
; CHECK: vfma.f64
%call = tail call double @llvm.fma.f64(double %a, double %b, double %c) nounwind readnone
ret double %call
}
define <2 x float> @test_fma_v2f32(<2 x float> %a, <2 x float> %b, <2 x float> %c) nounwind readnone ssp {
entry:
; CHECK: test_fma_v2f32
; CHECK: vfma.f32
%0 = tail call <2 x float> @llvm.fma.v2f32(<2 x float> %a, <2 x float> %b, <2 x float> %c) nounwind
ret <2 x float> %0
}
define float @test_fnma_f32(float %a, float %b, float %c) nounwind readnone ssp {
entry:
; CHECK: test_fnma_f32
; CHECK: vfnma.f32
%call = tail call float @llvm.fma.f32(float %a, float %b, float %c) nounwind readnone
%tmp1 = fsub float -0.0, %call
%tmp2 = fsub float %tmp1, %c
ret float %tmp2
}
define double @test_fnma_f64(double %a, double %b, double %c) nounwind readnone ssp {
entry:
; CHECK: test_fnma_f64
; CHECK: vfnma.f64
%call = tail call double @llvm.fma.f64(double %a, double %b, double %c) nounwind readnone
%tmp = fsub double -0.0, %call
ret double %tmp
}
declare float @llvm.fma.f32(float, float, float) nounwind readnone
declare double @llvm.fma.f64(double, double, double) nounwind readnone
declare <2 x float> @llvm.fma.v2f32(<2 x float>, <2 x float>, <2 x float>) nounwind readnone
|
