1 ; RUN: llc < %s -O=3 -mtriple=arm64-apple-ios -mcpu=cyclone -enable-unsafe-fp-math | FileCheck %s
2 define void @foo_2d(double* %src) {
4 ; CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{d[0-9]+}}
5 ; CHECK: fmadd {{d[0-9]+}}, {{d[0-9]+}}, {{d[0-9]+}}, {{d[0-9]+}}
7 %arrayidx1 = getelementptr inbounds double, double* %src, i64 5
8 %arrayidx2 = getelementptr inbounds double, double* %src, i64 11
9 %tmp = bitcast double* %arrayidx1 to <2 x double>*
10 %tmp1 = load double, double* %arrayidx2, align 8
11 %tmp2 = load double, double* %arrayidx1, align 8
12 %fmul = fmul fast double %tmp1, %tmp1
13 %fmul2 = fmul fast double %tmp2, 0x3F94AFD6A052BF5B
14 %fadd = fadd fast double %fmul, %fmul2
17 ; CHECK-LABEL: %for.body
18 ; CHECK: fmla.2d {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}
19 ; CHECK: fmla.2d {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
20 ; CHECK: fmla.d {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}[0]
21 for.body: ; preds = %for.body, %entry
22 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
23 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
24 %arrayidx3 = getelementptr inbounds double, double* %src, i64 %indvars.iv.next
25 %tmp3 = load double, double* %arrayidx3, align 8
26 %add = fadd fast double %tmp3, %tmp3
27 %mul = fmul fast double %add, %fadd
28 %e1 = insertelement <2 x double> undef, double %add, i32 0
29 %e2 = insertelement <2 x double> %e1, double %add, i32 1
30 %add2 = fadd fast <2 x double> %e2, <double 3.000000e+00, double -3.000000e+00>
31 %e3 = insertelement <2 x double> undef, double %mul, i32 0
32 %e4 = insertelement <2 x double> %e3, double %mul, i32 1
33 %mul2 = fmul fast <2 x double> %add2,<double 3.000000e+00, double -3.000000e+00>
34 %e5 = insertelement <2 x double> undef, double %add, i32 0
35 %e6 = insertelement <2 x double> %e5, double %add, i32 1
36 %add3 = fadd fast <2 x double> %mul2, <double 3.000000e+00, double -3.000000e+00>
37 %mulx = fmul fast <2 x double> %add2, %e2
38 %addx = fadd fast <2 x double> %mulx, %e4
39 %e7 = insertelement <2 x double> undef, double %mul, i32 0
40 %e8 = insertelement <2 x double> %e7, double %mul, i32 1
41 %e9 = fmul fast <2 x double> %addx, %add3
42 store <2 x double> %e9, <2 x double>* %tmp, align 8
43 %e10 = extractelement <2 x double> %add3, i32 0
44 %mul3 = fmul fast double %mul, %e10
45 %add4 = fadd fast double %mul3, %mul
46 store double %add4, double* %arrayidx2, align 8
47 %exitcond = icmp eq i64 %indvars.iv.next, 25
48 br i1 %exitcond, label %for.end, label %for.body
50 for.end: ; preds = %for.body
53 define void @foo_2s(float* %src) {
55 %arrayidx1 = getelementptr inbounds float, float* %src, i64 5
56 %arrayidx2 = getelementptr inbounds float, float* %src, i64 11
57 %tmp = bitcast float* %arrayidx1 to <2 x float>*
60 ; CHECK-LABEL: %for.body
61 ; CHECK: fmla.2s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}
62 ; CHECK: fmla.2s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
63 ; CHECK: fmla.s {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}[0]
64 for.body: ; preds = %for.body, %entry
65 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
66 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
67 %arrayidx3 = getelementptr inbounds float, float* %src, i64 %indvars.iv.next
68 %tmp1 = load float, float* %arrayidx3, align 8
69 %add = fadd fast float %tmp1, %tmp1
70 %mul = fmul fast float %add, %add
71 %e1 = insertelement <2 x float> undef, float %add, i32 0
72 %e2 = insertelement <2 x float> %e1, float %add, i32 1
73 %add2 = fadd fast <2 x float> %e2, <float 3.000000e+00, float -3.000000e+00>
74 %e3 = insertelement <2 x float> undef, float %mul, i32 0
75 %e4 = insertelement <2 x float> %e3, float %mul, i32 1
76 %mul2 = fmul fast <2 x float> %add2,<float 3.000000e+00, float -3.000000e+00>
77 %e5 = insertelement <2 x float> undef, float %add, i32 0
78 %e6 = insertelement <2 x float> %e5, float %add, i32 1
79 %add3 = fadd fast <2 x float> %mul2, <float 3.000000e+00, float -3.000000e+00>
80 %mulx = fmul fast <2 x float> %add2, %e2
81 %addx = fadd fast <2 x float> %mulx, %e4
82 %e7 = insertelement <2 x float> undef, float %mul, i32 0
83 %e8 = insertelement <2 x float> %e7, float %mul, i32 1
84 %e9 = fmul fast <2 x float> %addx, %add3
85 store <2 x float> %e9, <2 x float>* %tmp, align 8
86 %e10 = extractelement <2 x float> %add3, i32 0
87 %mul3 = fmul fast float %mul, %e10
88 %add4 = fadd fast float %mul3, %mul
89 store float %add4, float* %arrayidx2, align 8
90 %exitcond = icmp eq i64 %indvars.iv.next, 25
91 br i1 %exitcond, label %for.end, label %for.body
93 for.end: ; preds = %for.body
96 define void @foo_4s(float* %src) {
98 %arrayidx1 = getelementptr inbounds float, float* %src, i64 5
99 %arrayidx2 = getelementptr inbounds float, float* %src, i64 11
100 %tmp = bitcast float* %arrayidx1 to <4 x float>*
103 ; CHECK-LABEL: %for.body
104 ; CHECK: fmla.4s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}
105 ; CHECK: fmla.4s {{v[0-9]+}}, {{v[0-9]+}}, {{v[0-9]+}}[0]
106 for.body: ; preds = %for.body, %entry
107 %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
108 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1
109 %arrayidx3 = getelementptr inbounds float, float* %src, i64 %indvars.iv.next
110 %tmp1 = load float, float* %arrayidx3, align 8
111 %add = fadd fast float %tmp1, %tmp1
112 %mul = fmul fast float %add, %add
113 %e1 = insertelement <4 x float> undef, float %add, i32 0
114 %e2 = insertelement <4 x float> %e1, float %add, i32 1
115 %add2 = fadd fast <4 x float> %e2, <float 3.000000e+00, float -3.000000e+00, float 5.000000e+00, float 7.000000e+00>
116 %e3 = insertelement <4 x float> undef, float %mul, i32 0
117 %e4 = insertelement <4 x float> %e3, float %mul, i32 1
118 %mul2 = fmul fast <4 x float> %add2,<float 3.000000e+00, float -3.000000e+00, float 5.000000e+00, float 7.000000e+00>
119 %e5 = insertelement <4 x float> undef, float %add, i32 0
120 %e6 = insertelement <4 x float> %e5, float %add, i32 1
121 %add3 = fadd fast <4 x float> %mul2, <float 3.000000e+00, float -3.000000e+00, float 5.000000e+00, float 7.000000e+00>
122 %mulx = fmul fast <4 x float> %add2, %e2
123 %addx = fadd fast <4 x float> %mulx, %e4
124 %e7 = insertelement <4 x float> undef, float %mul, i32 0
125 %e8 = insertelement <4 x float> %e7, float %mul, i32 1
126 %e9 = fmul fast <4 x float> %addx, %add3
127 store <4 x float> %e9, <4 x float>* %tmp, align 8
128 %e10 = extractelement <4 x float> %add3, i32 0
129 %mul3 = fmul fast float %mul, %e10
130 store float %mul3, float* %arrayidx2, align 8
131 %exitcond = icmp eq i64 %indvars.iv.next, 25
132 br i1 %exitcond, label %for.end, label %for.body
134 for.end: ; preds = %for.body