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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown -mattr=+mmx,+sse2 | FileCheck %s --check-prefix=X86
; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+mmx,+sse2 | FileCheck %s --check-prefix=X64
define i32 @test0(ptr %v4) nounwind {
; X86-LABEL: test0:
; X86: # %bb.0: # %entry
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: pshufw $238, (%eax), %mm0 # mm0 = mem[2,3,2,3]
; X86-NEXT: movd %mm0, %eax
; X86-NEXT: addl $32, %eax
; X86-NEXT: retl
;
; X64-LABEL: test0:
; X64: # %bb.0: # %entry
; X64-NEXT: pshufw $238, (%rdi), %mm0 # mm0 = mem[2,3,2,3]
; X64-NEXT: movd %mm0, %eax
; X64-NEXT: addl $32, %eax
; X64-NEXT: retq
entry:
%v5 = load <1 x i64>, ptr %v4, align 8
%v12 = bitcast <1 x i64> %v5 to <4 x i16>
%v13 = bitcast <4 x i16> %v12 to <1 x i64>
%v14 = tail call <1 x i64> @llvm.x86.sse.pshuf.w(<1 x i64> %v13, i8 -18)
%v15 = bitcast <1 x i64> %v14 to <4 x i16>
%v16 = bitcast <4 x i16> %v15 to <1 x i64>
%v17 = extractelement <1 x i64> %v16, i32 0
%v18 = bitcast i64 %v17 to <2 x i32>
%v19 = extractelement <2 x i32> %v18, i32 0
%v20 = add i32 %v19, 32
ret i32 %v20
}
define i32 @test1(ptr nocapture readonly %ptr) nounwind {
; X86-LABEL: test1:
; X86: # %bb.0: # %entry
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: movd (%eax), %mm0
; X86-NEXT: pshufw $232, %mm0, %mm0 # mm0 = mm0[0,2,2,3]
; X86-NEXT: movd %mm0, %eax
; X86-NEXT: emms
; X86-NEXT: retl
;
; X64-LABEL: test1:
; X64: # %bb.0: # %entry
; X64-NEXT: movd (%rdi), %mm0
; X64-NEXT: pshufw $232, %mm0, %mm0 # mm0 = mm0[0,2,2,3]
; X64-NEXT: movd %mm0, %eax
; X64-NEXT: emms
; X64-NEXT: retq
entry:
%0 = load i32, ptr %ptr, align 4
%1 = insertelement <2 x i32> undef, i32 %0, i32 0
%2 = insertelement <2 x i32> %1, i32 0, i32 1
%3 = bitcast <2 x i32> %2 to <1 x i64>
%4 = bitcast <1 x i64> %3 to i64
%5 = bitcast i64 %4 to <4 x i16>
%6 = bitcast <4 x i16> %5 to <1 x i64>
%7 = tail call <1 x i64> @llvm.x86.sse.pshuf.w(<1 x i64> %6, i8 -24)
%8 = bitcast <1 x i64> %7 to <4 x i16>
%9 = bitcast <4 x i16> %8 to <1 x i64>
%10 = extractelement <1 x i64> %9, i32 0
%11 = bitcast i64 %10 to <2 x i32>
%12 = extractelement <2 x i32> %11, i32 0
tail call void @llvm.x86.mmx.emms()
ret i32 %12
}
define i32 @test2(ptr nocapture readonly %ptr) nounwind {
; X86-LABEL: test2:
; X86: # %bb.0: # %entry
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: pshufw $232, (%eax), %mm0 # mm0 = mem[0,2,2,3]
; X86-NEXT: movd %mm0, %eax
; X86-NEXT: emms
; X86-NEXT: retl
;
; X64-LABEL: test2:
; X64: # %bb.0: # %entry
; X64-NEXT: pshufw $232, (%rdi), %mm0 # mm0 = mem[0,2,2,3]
; X64-NEXT: movd %mm0, %eax
; X64-NEXT: emms
; X64-NEXT: retq
entry:
%0 = load <1 x i64>, ptr %ptr, align 8
%1 = tail call <1 x i64> @llvm.x86.sse.pshuf.w(<1 x i64> %0, i8 -24)
%2 = bitcast <1 x i64> %1 to <4 x i16>
%3 = bitcast <4 x i16> %2 to <1 x i64>
%4 = extractelement <1 x i64> %3, i32 0
%5 = bitcast i64 %4 to <2 x i32>
%6 = extractelement <2 x i32> %5, i32 0
tail call void @llvm.x86.mmx.emms()
ret i32 %6
}
define i32 @test3(<1 x i64> %a) nounwind {
; X86-LABEL: test3:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: retl
;
; X64-LABEL: test3:
; X64: # %bb.0:
; X64-NEXT: movq %rdi, %rax
; X64-NEXT: # kill: def $eax killed $eax killed $rax
; X64-NEXT: retq
%tmp0 = bitcast <1 x i64> %a to <2 x i32>
%tmp1 = extractelement <2 x i32> %tmp0, i32 0
ret i32 %tmp1
}
; Verify we don't muck with extractelts from the upper lane.
define i32 @test4(<1 x i64> %a) nounwind {
; X86-LABEL: test4:
; X86: # %bb.0:
; X86-NEXT: movl {{[0-9]+}}(%esp), %eax
; X86-NEXT: retl
;
; X64-LABEL: test4:
; X64: # %bb.0:
; X64-NEXT: movq %rdi, %xmm0
; X64-NEXT: pshufd {{.*#+}} xmm0 = xmm0[1,1,1,1]
; X64-NEXT: movd %xmm0, %eax
; X64-NEXT: retq
%tmp0 = bitcast <1 x i64> %a to <2 x i32>
%tmp1 = extractelement <2 x i32> %tmp0, i32 1
ret i32 %tmp1
}
declare <1 x i64> @llvm.x86.sse.pshuf.w(<1 x i64>, i8)
declare void @llvm.x86.mmx.emms()
|
