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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefix=CHECK --check-prefix=SSE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=CHECK --check-prefix=AVX --check-prefix=AVX2
; fold (srem x, 1) -> 0
define i32 @combine_srem_by_one(i32 %x) {
; CHECK-LABEL: combine_srem_by_one:
; CHECK: # %bb.0:
; CHECK-NEXT: xorl %eax, %eax
; CHECK-NEXT: retq
%1 = srem i32 %x, 1
ret i32 %1
}
define <4 x i32> @combine_vec_srem_by_one(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_by_one:
; SSE: # %bb.0:
; SSE-NEXT: xorps %xmm0, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_srem_by_one:
; AVX: # %bb.0:
; AVX-NEXT: vxorps %xmm0, %xmm0, %xmm0
; AVX-NEXT: retq
%1 = srem <4 x i32> %x, <i32 1, i32 1, i32 1, i32 1>
ret <4 x i32> %1
}
; TODO fold (srem x, x) -> 0
define i32 @combine_srem_dupe(i32 %x) {
; CHECK-LABEL: combine_srem_dupe:
; CHECK: # %bb.0:
; CHECK-NEXT: movl %edi, %eax
; CHECK-NEXT: cltd
; CHECK-NEXT: idivl %edi
; CHECK-NEXT: movl %edx, %eax
; CHECK-NEXT: retq
%1 = srem i32 %x, %x
ret i32 %1
}
define <4 x i32> @combine_vec_srem_dupe(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_dupe:
; SSE: # %bb.0:
; SSE-NEXT: pextrd $1, %xmm0, %ecx
; SSE-NEXT: movl %ecx, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %ecx
; SSE-NEXT: movl %edx, %ecx
; SSE-NEXT: movd %xmm0, %esi
; SSE-NEXT: movl %esi, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %esi
; SSE-NEXT: movd %edx, %xmm1
; SSE-NEXT: pinsrd $1, %ecx, %xmm1
; SSE-NEXT: pextrd $2, %xmm0, %ecx
; SSE-NEXT: movl %ecx, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %ecx
; SSE-NEXT: pinsrd $2, %edx, %xmm1
; SSE-NEXT: pextrd $3, %xmm0, %ecx
; SSE-NEXT: movl %ecx, %eax
; SSE-NEXT: cltd
; SSE-NEXT: idivl %ecx
; SSE-NEXT: pinsrd $3, %edx, %xmm1
; SSE-NEXT: movdqa %xmm1, %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_srem_dupe:
; AVX: # %bb.0:
; AVX-NEXT: vpextrd $1, %xmm0, %ecx
; AVX-NEXT: movl %ecx, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %ecx
; AVX-NEXT: movl %edx, %ecx
; AVX-NEXT: vmovd %xmm0, %esi
; AVX-NEXT: movl %esi, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %esi
; AVX-NEXT: vmovd %edx, %xmm1
; AVX-NEXT: vpinsrd $1, %ecx, %xmm1, %xmm1
; AVX-NEXT: vpextrd $2, %xmm0, %ecx
; AVX-NEXT: movl %ecx, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %ecx
; AVX-NEXT: vpinsrd $2, %edx, %xmm1, %xmm1
; AVX-NEXT: vpextrd $3, %xmm0, %ecx
; AVX-NEXT: movl %ecx, %eax
; AVX-NEXT: cltd
; AVX-NEXT: idivl %ecx
; AVX-NEXT: vpinsrd $3, %edx, %xmm1, %xmm0
; AVX-NEXT: retq
%1 = srem <4 x i32> %x, %x
ret <4 x i32> %1
}
; fold (srem x, y) -> (urem x, y) iff x and y are positive
define <4 x i32> @combine_vec_srem_by_pos0(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_by_pos0:
; SSE: # %bb.0:
; SSE-NEXT: andps {{.*}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX1-LABEL: combine_vec_srem_by_pos0:
; AVX1: # %bb.0:
; AVX1-NEXT: vandps {{.*}}(%rip), %xmm0, %xmm0
; AVX1-NEXT: retq
;
; AVX2-LABEL: combine_vec_srem_by_pos0:
; AVX2: # %bb.0:
; AVX2-NEXT: vbroadcastss {{.*#+}} xmm1 = [3,3,3,3]
; AVX2-NEXT: vandps %xmm1, %xmm0, %xmm0
; AVX2-NEXT: retq
%1 = and <4 x i32> %x, <i32 255, i32 255, i32 255, i32 255>
%2 = srem <4 x i32> %1, <i32 4, i32 4, i32 4, i32 4>
ret <4 x i32> %2
}
define <4 x i32> @combine_vec_srem_by_pos1(<4 x i32> %x) {
; SSE-LABEL: combine_vec_srem_by_pos1:
; SSE: # %bb.0:
; SSE-NEXT: andps {{.*}}(%rip), %xmm0
; SSE-NEXT: retq
;
; AVX-LABEL: combine_vec_srem_by_pos1:
; AVX: # %bb.0:
; AVX-NEXT: vandps {{.*}}(%rip), %xmm0, %xmm0
; AVX-NEXT: retq
%1 = and <4 x i32> %x, <i32 255, i32 255, i32 255, i32 255>
%2 = srem <4 x i32> %1, <i32 1, i32 4, i32 8, i32 16>
ret <4 x i32> %2
}
|
