1; RUN: llc < %s -O3 -mtriple=arm64-apple-ios -enable-unsafe-fp-math | FileCheck %s
2; RUN: llc < %s -O3 -mtriple=arm64-apple-ios -fp-contract=fast | FileCheck %s
3
4define void @foo_2d(double* %src) {
5entry:
6  %arrayidx1 = getelementptr inbounds double, double* %src, i64 5
7  %arrayidx2 = getelementptr inbounds double, double* %src, i64 11
8  %tmp = bitcast double* %arrayidx1 to <2 x double>*
9  br label %for.body
10
11; CHECK-LABEL: %for.body
12; CHECK: fmls.2d {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}
13; CHECK: fmls.2d {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
14; CHECK: fmls.d {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}[0]
15for.body:                                         ; preds = %for.body, %entry
16  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
17  %indvars.iv.next = sub nuw nsw i64 %indvars.iv, 1
18  %arrayidx3 = getelementptr inbounds double, double* %src, i64 %indvars.iv.next
19  %tmp1 = load double, double* %arrayidx3, align 8
20  %add = fadd fast double %tmp1, %tmp1
21  %mul = fmul fast double %add, %add
22  %e1 = insertelement <2 x double> undef, double %add, i32 0
23  %e2 = insertelement <2 x double> %e1, double %add, i32 1
24  %sub2 = fsub fast <2 x double> %e2, <double 3.000000e+00, double -3.000000e+00>
25  %e3 = insertelement <2 x double> undef, double %mul, i32 0
26  %e4 = insertelement <2 x double> %e3, double %mul, i32 1
27  %mul2 = fmul fast <2 x double> %sub2,<double 3.000000e+00, double -3.000000e+00>
28  %e5 = insertelement <2 x double> undef, double %add, i32 0
29  %e6 = insertelement <2 x double> %e5, double %add, i32 1
30  %sub3 = fsub fast  <2 x double>  <double 3.000000e+00, double -3.000000e+00>, %mul2
31  %mulx = fmul fast <2 x double> %sub2, %e2
32  %subx = fsub fast  <2 x double> %e4, %mulx
33  %e7 = insertelement <2 x double> undef, double %mul, i32 0
34  %e8 = insertelement <2 x double> %e7, double %mul, i32 1
35  %e9 = fmul fast <2 x double>  %subx, %sub3
36  store <2 x double> %e9, <2 x double>* %tmp, align 8
37  %e10 = extractelement <2 x double> %sub3, i32 0
38  %mul3 = fmul fast double %mul, %e10
39  %sub4 = fsub fast double %mul, %mul3
40  store double %sub4, double* %arrayidx2, align 8
41  %exitcond = icmp eq i64 %indvars.iv.next, 25
42  br i1 %exitcond, label %for.end, label %for.body
43
44for.end:                                          ; preds = %for.body
45  ret void
46}
47define void @foo_2s(float* %src) {
48entry:
49  %arrayidx1 = getelementptr inbounds float, float* %src, i64 5
50  %arrayidx2 = getelementptr inbounds float, float* %src, i64 11
51  %tmp = bitcast float* %arrayidx1 to <2 x float>*
52  br label %for.body
53
54; CHECK-LABEL: %for.body
55; CHECK: fmls.2s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}
56; CHECK: fmls.2s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
57; CHECK: fmls.s {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}[0]
58for.body:                                         ; preds = %for.body, %entry
59  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
60  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
61  %arrayidx3 = getelementptr inbounds float, float* %src, i64 %indvars.iv.next
62  %tmp1 = load float, float* %arrayidx3, align 8
63  %add = fadd fast float %tmp1, %tmp1
64  %mul = fmul fast float %add, %add
65  %e1 = insertelement <2 x float> undef, float %add, i32 0
66  %e2 = insertelement <2 x float> %e1, float %add, i32 1
67  %add2 = fsub fast <2 x float> %e2, <float 3.000000e+00, float -3.000000e+00>
68  %e3 = insertelement <2 x float> undef, float %mul, i32 0
69  %e4 = insertelement <2 x float> %e3, float %mul, i32 1
70  %mul2 = fmul fast <2 x float> %add2,<float 3.000000e+00, float -3.000000e+00>
71  %e5 = insertelement <2 x float> undef, float %add, i32 0
72  %e6 = insertelement <2 x float> %e5, float %add, i32 1
73  %add3 = fsub fast  <2 x float>  <float 3.000000e+00, float -3.000000e+00>, %mul2
74  %mulx = fmul fast <2 x float> %add2, %e2
75  %addx = fsub fast  <2 x float> %e4, %mulx
76  %e7 = insertelement <2 x float> undef, float %mul, i32 0
77  %e8 = insertelement <2 x float> %e7, float %mul, i32 1
78  %e9 = fmul fast <2 x float>  %addx, %add3
79  store <2 x float> %e9, <2 x float>* %tmp, align 8
80  %e10 = extractelement <2 x float> %add3, i32 0
81  %mul3 = fmul fast float %mul, %e10
82  %add4 = fsub fast float %mul, %mul3
83  store float %add4, float* %arrayidx2, align 8
84  %exitcond = icmp eq i64 %indvars.iv.next, 25
85  br i1 %exitcond, label %for.end, label %for.body
86
87for.end:                                          ; preds = %for.body
88  ret void
89}
90define void @foo_4s(float* %src) {
91entry:
92  %arrayidx1 = getelementptr inbounds float, float* %src, i64 5
93  %arrayidx2 = getelementptr inbounds float, float* %src, i64 11
94  %tmp = bitcast float* %arrayidx1 to <4 x float>*
95  br label %for.body
96
97; CHECK-LABEL: %for.body
98; CHECK: fmls.4s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}
99; CHECK: fmls.4s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
100for.body:                                         ; preds = %for.body, %entry
101  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
102  %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
103  %arrayidx3 = getelementptr inbounds float, float* %src, i64 %indvars.iv.next
104  %tmp1 = load float, float* %arrayidx3, align 8
105  %add = fadd fast float %tmp1, %tmp1
106  %mul = fmul fast float %add, %add
107  %e1 = insertelement <4 x float> undef, float %add, i32 0
108  %e2 = insertelement <4 x float> %e1, float %add, i32 1
109  %add2 = fadd fast <4 x float> %e2, <float 3.000000e+00, float -3.000000e+00, float 5.000000e+00, float 7.000000e+00>
110  %e3 = insertelement <4 x float> undef, float %mul, i32 0
111  %e4 = insertelement <4 x float> %e3, float %mul, i32 1
112  %mul2 = fmul fast <4 x float> %add2,<float 3.000000e+00, float -3.000000e+00, float 5.000000e+00, float 7.000000e+00>
113  %e5 = insertelement <4 x float> undef, float %add, i32 0
114  %e6 = insertelement <4 x float> %e5, float %add, i32 1
115  %add3 = fsub fast  <4 x float> <float 3.000000e+00, float -3.000000e+00, float 5.000000e+00, float 7.000000e+00> , %mul2
116  %mulx = fmul fast <4 x float> %add2, %e2
117  %addx = fsub fast  <4 x float> %e4, %mulx
118  %e7 = insertelement <4 x float> undef, float %mul, i32 0
119  %e8 = insertelement <4 x float> %e7, float %mul, i32 1
120  %e9 = fmul fast <4 x float>  %addx, %add3
121  store <4 x float> %e9, <4 x float>* %tmp, align 8
122  %e10 = extractelement <4 x float> %add3, i32 0
123  %mul3 = fmul fast float %mul, %e10
124  store float %mul3, float* %arrayidx2, align 8
125  %exitcond = icmp eq i64 %indvars.iv.next, 25
126  br i1 %exitcond, label %for.end, label %for.body
127
128for.end:                                          ; preds = %for.body
129  ret void
130}
131
132; CHECK-LABEL: test1:
133; CHECK: fnmadd s0, s0, s1, s2
134define float @test1(float %a, float %b, float %c) {
135entry:
136  %0 = fmul float %a, %b
137  %mul = fsub float -0.000000e+00, %0
138  %sub1 = fsub float %mul, %c
139  ret float %sub1
140}
141
142; CHECK-LABEL: test2:
143; CHECK: fnmadd d0, d0, d1, d2
144define double @test2(double %a, double %b, double %c) {
145entry:
146  %0 = fmul double %a, %b
147  %mul = fsub double -0.000000e+00, %0
148  %sub1 = fsub double %mul, %c
149  ret double %sub1
150}
151