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
| ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-pc-linux -mcpu=corei7 | FileCheck %s --check-prefix=SSE
; RUN: llc < %s -mtriple=x86_64-pc-linux -mcpu=core-avx-i | FileCheck %s --check-prefix=AVX
define <2 x double> @v2f2d_ext_vec(<2 x float> %v1) nounwind {
; SSE-LABEL: v2f2d_ext_vec:
; SSE: # %bb.0: # %entry
; SSE-NEXT: cvtps2pd %xmm0, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: v2f2d_ext_vec:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vcvtps2pd %xmm0, %xmm0
; AVX-NEXT: retq
entry:
%f1 = fpext <2 x float> %v1 to <2 x double>
ret <2 x double> %f1
}
define <3 x double> @v3f2d_ext_vec(<3 x float> %v1) nounwind {
; SSE-LABEL: v3f2d_ext_vec:
; SSE: # %bb.0: # %entry
; SSE-NEXT: cvtps2pd %xmm0, %xmm2
; SSE-NEXT: movhlps {{.*#+}} xmm0 = xmm0[1,1]
; SSE-NEXT: cvtps2pd %xmm0, %xmm0
; SSE-NEXT: movlps %xmm0, -{{[0-9]+}}(%rsp)
; SSE-NEXT: movaps %xmm2, %xmm1
; SSE-NEXT: unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm2[1]
; SSE-NEXT: fldl -{{[0-9]+}}(%rsp)
; SSE-NEXT: movaps %xmm2, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: v3f2d_ext_vec:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vcvtps2pd %xmm0, %ymm0
; AVX-NEXT: retq
entry:
%f1 = fpext <3 x float> %v1 to <3 x double>
ret <3 x double> %f1
}
define <4 x double> @v4f2d_ext_vec(<4 x float> %v1) nounwind {
; SSE-LABEL: v4f2d_ext_vec:
; SSE: # %bb.0: # %entry
; SSE-NEXT: cvtps2pd %xmm0, %xmm2
; SSE-NEXT: movhlps {{.*#+}} xmm0 = xmm0[1,1]
; SSE-NEXT: cvtps2pd %xmm0, %xmm1
; SSE-NEXT: movaps %xmm2, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: v4f2d_ext_vec:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vcvtps2pd %xmm0, %ymm0
; AVX-NEXT: retq
entry:
%f1 = fpext <4 x float> %v1 to <4 x double>
ret <4 x double> %f1
}
define <8 x double> @v8f2d_ext_vec(<8 x float> %v1) nounwind {
; SSE-LABEL: v8f2d_ext_vec:
; SSE: # %bb.0: # %entry
; SSE-NEXT: cvtps2pd %xmm0, %xmm5
; SSE-NEXT: cvtps2pd %xmm1, %xmm2
; SSE-NEXT: movhlps {{.*#+}} xmm0 = xmm0[1,1]
; SSE-NEXT: cvtps2pd %xmm0, %xmm4
; SSE-NEXT: movhlps {{.*#+}} xmm1 = xmm1[1,1]
; SSE-NEXT: cvtps2pd %xmm1, %xmm3
; SSE-NEXT: movaps %xmm5, %xmm0
; SSE-NEXT: movaps %xmm4, %xmm1
; SSE-NEXT: retq
;
; AVX-LABEL: v8f2d_ext_vec:
; AVX: # %bb.0: # %entry
; AVX-NEXT: vcvtps2pd %xmm0, %ymm2
; AVX-NEXT: vextractf128 $1, %ymm0, %xmm0
; AVX-NEXT: vcvtps2pd %xmm0, %ymm1
; AVX-NEXT: vmovaps %ymm2, %ymm0
; AVX-NEXT: retq
entry:
%f1 = fpext <8 x float> %v1 to <8 x double>
ret <8 x double> %f1
}
define void @test_vector_creation() nounwind {
; SSE-LABEL: test_vector_creation:
; SSE: # %bb.0:
; SSE-NEXT: xorps %xmm0, %xmm0
; SSE-NEXT: movhps {{.*#+}} xmm0 = xmm0[0,1],mem[0,1]
; SSE-NEXT: movaps %xmm0, (%rax)
; SSE-NEXT: retq
;
; AVX-LABEL: test_vector_creation:
; AVX: # %bb.0:
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
; AVX-NEXT: vmovhps {{.*#+}} xmm0 = xmm0[0,1],mem[0,1]
; AVX-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0
; AVX-NEXT: vmovaps %ymm0, (%rax)
; AVX-NEXT: vzeroupper
; AVX-NEXT: retq
%1 = insertelement <4 x double> undef, double 0.000000e+00, i32 2
%2 = load double, double addrspace(1)* null
%3 = insertelement <4 x double> %1, double %2, i32 3
store <4 x double> %3, <4 x double>* undef
ret void
}
|