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
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mtriple=aarch64-unknown-linux-gnu < %s | FileCheck %s
define <4 x i16> @fold_urem_vec_1(<4 x i16> %x) {
; CHECK-LABEL: fold_urem_vec_1:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: umov w8, v0.h[0]
; CHECK-NEXT: mov w9, #55879
; CHECK-NEXT: movk w9, #689, lsl #16
; CHECK-NEXT: umov w10, v0.h[1]
; CHECK-NEXT: mov w11, #33826
; CHECK-NEXT: mov w12, #95
; CHECK-NEXT: movk w11, #528, lsl #16
; CHECK-NEXT: umov w13, v0.h[2]
; CHECK-NEXT: umull x9, w8, w9
; CHECK-NEXT: umull x11, w10, w11
; CHECK-NEXT: lsr x9, x9, #32
; CHECK-NEXT: lsr x11, x11, #32
; CHECK-NEXT: msub w8, w9, w12, w8
; CHECK-NEXT: mov w9, #48149
; CHECK-NEXT: movk w9, #668, lsl #16
; CHECK-NEXT: mov w12, #124
; CHECK-NEXT: umull x9, w13, w9
; CHECK-NEXT: msub w10, w11, w12, w10
; CHECK-NEXT: umov w11, v0.h[3]
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov w12, #22281
; CHECK-NEXT: lsr x8, x9, #32
; CHECK-NEXT: mov w9, #98
; CHECK-NEXT: movk w12, #65, lsl #16
; CHECK-NEXT: msub w8, w8, w9, w13
; CHECK-NEXT: mov v0.h[1], w10
; CHECK-NEXT: umull x9, w11, w12
; CHECK-NEXT: mov w10, #1003
; CHECK-NEXT: lsr x9, x9, #32
; CHECK-NEXT: mov v0.h[2], w8
; CHECK-NEXT: msub w8, w9, w10, w11
; CHECK-NEXT: mov v0.h[3], w8
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%1 = urem <4 x i16> %x, <i16 95, i16 124, i16 98, i16 1003>
ret <4 x i16> %1
}
define <4 x i16> @fold_urem_vec_2(<4 x i16> %x) {
; CHECK-LABEL: fold_urem_vec_2:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: umov w8, v0.h[0]
; CHECK-NEXT: mov w9, #55879
; CHECK-NEXT: movk w9, #689, lsl #16
; CHECK-NEXT: umov w10, v0.h[1]
; CHECK-NEXT: mov w12, #95
; CHECK-NEXT: umov w13, v0.h[2]
; CHECK-NEXT: umull x11, w8, w9
; CHECK-NEXT: umull x14, w10, w9
; CHECK-NEXT: lsr x11, x11, #32
; CHECK-NEXT: msub w8, w11, w12, w8
; CHECK-NEXT: lsr x11, x14, #32
; CHECK-NEXT: umull x14, w13, w9
; CHECK-NEXT: msub w10, w11, w12, w10
; CHECK-NEXT: umov w11, v0.h[3]
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: lsr x8, x14, #32
; CHECK-NEXT: msub w8, w8, w12, w13
; CHECK-NEXT: mov v0.h[1], w10
; CHECK-NEXT: umull x9, w11, w9
; CHECK-NEXT: lsr x9, x9, #32
; CHECK-NEXT: mov v0.h[2], w8
; CHECK-NEXT: msub w8, w9, w12, w11
; CHECK-NEXT: mov v0.h[3], w8
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: ret
%1 = urem <4 x i16> %x, <i16 95, i16 95, i16 95, i16 95>
ret <4 x i16> %1
}
; Don't fold if we can combine urem with udiv.
define <4 x i16> @combine_urem_udiv(<4 x i16> %x) {
; CHECK-LABEL: combine_urem_udiv:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: umov w8, v0.h[0]
; CHECK-NEXT: mov w9, #55879
; CHECK-NEXT: movk w9, #689, lsl #16
; CHECK-NEXT: umov w10, v0.h[1]
; CHECK-NEXT: mov w12, #95
; CHECK-NEXT: umov w14, v0.h[2]
; CHECK-NEXT: umov w15, v0.h[3]
; CHECK-NEXT: umull x11, w8, w9
; CHECK-NEXT: umull x13, w10, w9
; CHECK-NEXT: lsr x11, x11, #32
; CHECK-NEXT: lsr x13, x13, #32
; CHECK-NEXT: msub w8, w11, w12, w8
; CHECK-NEXT: msub w10, w13, w12, w10
; CHECK-NEXT: fmov s1, w11
; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: umull x8, w14, w9
; CHECK-NEXT: umull x9, w15, w9
; CHECK-NEXT: lsr x8, x8, #32
; CHECK-NEXT: mov v0.h[1], w10
; CHECK-NEXT: lsr x9, x9, #32
; CHECK-NEXT: msub w10, w8, w12, w14
; CHECK-NEXT: mov v1.h[1], w13
; CHECK-NEXT: msub w11, w9, w12, w15
; CHECK-NEXT: mov v0.h[2], w10
; CHECK-NEXT: mov v1.h[2], w8
; CHECK-NEXT: mov v0.h[3], w11
; CHECK-NEXT: mov v1.h[3], w9
; CHECK-NEXT: add v0.4h, v0.4h, v1.4h
; CHECK-NEXT: ret
%1 = urem <4 x i16> %x, <i16 95, i16 95, i16 95, i16 95>
%2 = udiv <4 x i16> %x, <i16 95, i16 95, i16 95, i16 95>
%3 = add <4 x i16> %1, %2
ret <4 x i16> %3
}
; Don't fold for divisors that are a power of two.
define <4 x i16> @dont_fold_urem_power_of_two(<4 x i16> %x) {
; CHECK-LABEL: dont_fold_urem_power_of_two:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: umov w9, v0.h[0]
; CHECK-NEXT: umov w11, v0.h[1]
; CHECK-NEXT: umov w10, v0.h[3]
; CHECK-NEXT: mov w8, #55879
; CHECK-NEXT: movk w8, #689, lsl #16
; CHECK-NEXT: and w9, w9, #0x3f
; CHECK-NEXT: umull x8, w10, w8
; CHECK-NEXT: fmov s1, w9
; CHECK-NEXT: and w9, w11, #0x1f
; CHECK-NEXT: umov w11, v0.h[2]
; CHECK-NEXT: lsr x8, x8, #32
; CHECK-NEXT: mov v1.h[1], w9
; CHECK-NEXT: mov w9, #95
; CHECK-NEXT: and w11, w11, #0x7
; CHECK-NEXT: msub w8, w8, w9, w10
; CHECK-NEXT: mov v1.h[2], w11
; CHECK-NEXT: mov v1.h[3], w8
; CHECK-NEXT: fmov d0, d1
; CHECK-NEXT: ret
%1 = urem <4 x i16> %x, <i16 64, i16 32, i16 8, i16 95>
ret <4 x i16> %1
}
; Don't fold if the divisor is one.
define <4 x i16> @dont_fold_srem_one(<4 x i16> %x) {
; CHECK-LABEL: dont_fold_srem_one:
; CHECK: // %bb.0:
; CHECK-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NEXT: umov w8, v0.h[1]
; CHECK-NEXT: mov w9, #13629
; CHECK-NEXT: movk w9, #100, lsl #16
; CHECK-NEXT: umov w10, v0.h[2]
; CHECK-NEXT: mov w11, #25645
; CHECK-NEXT: mov w12, #654
; CHECK-NEXT: movk w11, #2849, lsl #16
; CHECK-NEXT: movi d1, #0000000000000000
; CHECK-NEXT: umull x9, w8, w9
; CHECK-NEXT: mov w13, #5560
; CHECK-NEXT: umull x11, w10, w11
; CHECK-NEXT: movk w13, #12, lsl #16
; CHECK-NEXT: lsr x9, x9, #32
; CHECK-NEXT: lsr x11, x11, #32
; CHECK-NEXT: msub w8, w9, w12, w8
; CHECK-NEXT: umov w9, v0.h[3]
; CHECK-NEXT: mov w12, #23
; CHECK-NEXT: msub w10, w11, w12, w10
; CHECK-NEXT: mov w11, #5423
; CHECK-NEXT: mov v1.h[1], w8
; CHECK-NEXT: umull x8, w9, w13
; CHECK-NEXT: lsr x8, x8, #32
; CHECK-NEXT: mov v1.h[2], w10
; CHECK-NEXT: msub w8, w8, w11, w9
; CHECK-NEXT: mov v1.h[3], w8
; CHECK-NEXT: fmov d0, d1
; CHECK-NEXT: ret
%1 = urem <4 x i16> %x, <i16 1, i16 654, i16 23, i16 5423>
ret <4 x i16> %1
}
; Don't fold if the divisor is 2^16.
define <4 x i16> @dont_fold_urem_i16_smax(<4 x i16> %x) {
; CHECK-LABEL: dont_fold_urem_i16_smax:
; CHECK: // %bb.0:
; CHECK-NEXT: ret
%1 = urem <4 x i16> %x, <i16 1, i16 65536, i16 23, i16 5423>
ret <4 x i16> %1
}
; Don't fold i64 urem.
define <4 x i64> @dont_fold_urem_i64(<4 x i64> %x) {
; CHECK-LABEL: dont_fold_urem_i64:
; CHECK: // %bb.0:
; CHECK-NEXT: mov x8, #17097
; CHECK-NEXT: fmov x9, d1
; CHECK-NEXT: movk x8, #45590, lsl #16
; CHECK-NEXT: mov x13, #21445
; CHECK-NEXT: movk x8, #34192, lsl #32
; CHECK-NEXT: movk x13, #1603, lsl #16
; CHECK-NEXT: movk x8, #25644, lsl #48
; CHECK-NEXT: movk x13, #15432, lsl #32
; CHECK-NEXT: mov x10, v0.d[1]
; CHECK-NEXT: movk x13, #25653, lsl #48
; CHECK-NEXT: umulh x8, x9, x8
; CHECK-NEXT: mov x11, v1.d[1]
; CHECK-NEXT: sub x12, x9, x8
; CHECK-NEXT: lsr x14, x10, #1
; CHECK-NEXT: add x8, x8, x12, lsr #1
; CHECK-NEXT: mov x12, #12109
; CHECK-NEXT: movk x12, #52170, lsl #16
; CHECK-NEXT: umulh x13, x14, x13
; CHECK-NEXT: movk x12, #28749, lsl #32
; CHECK-NEXT: mov w14, #23
; CHECK-NEXT: movk x12, #49499, lsl #48
; CHECK-NEXT: lsr x8, x8, #4
; CHECK-NEXT: lsr x13, x13, #7
; CHECK-NEXT: umulh x12, x11, x12
; CHECK-NEXT: msub x8, x8, x14, x9
; CHECK-NEXT: mov w9, #5423
; CHECK-NEXT: lsr x12, x12, #12
; CHECK-NEXT: mov w14, #654
; CHECK-NEXT: movi v0.2d, #0000000000000000
; CHECK-NEXT: msub x9, x12, x9, x11
; CHECK-NEXT: msub x10, x13, x14, x10
; CHECK-NEXT: fmov d1, x8
; CHECK-NEXT: mov v1.d[1], x9
; CHECK-NEXT: mov v0.d[1], x10
; CHECK-NEXT: ret
%1 = urem <4 x i64> %x, <i64 1, i64 654, i64 23, i64 5423>
ret <4 x i64> %1
}
|
