blob: 48edae8c7d13700797743e5a20ea4b3986a5bad4 (
plain)
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
236
237
238
239
240
241
242
243
244
|
; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py
; RUN: opt < %s -mtriple=armv8a-linux-gnueabihf -mattr=+fp64 -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefix=CHECK-V8
; RUN: opt < %s -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve.fp -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefix=CHECK-MVEFP
; RUN: opt < %s -mtriple=thumbv8.1m.main-none-eabi -mattr=+mve -passes="print<cost-model>" 2>&1 -disable-output | FileCheck %s --check-prefix=CHECK-MVEI
target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
define void @fmin_strict() {
; CHECK-V8-LABEL: 'fmin_strict'
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmin_v2f16 = call half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %fmin_v4f16 = call half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %fmin_v8f16 = call half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 91 for instruction: %fmin_v16f16 = call half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmin_v2f32 = call float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmin_v4f32 = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmin_v8f32 = call float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmin_v2f64 = call double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmin_v4f64 = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %fmin_v4f128 = call fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fmin_strict'
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmin_v2f16 = call half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmin_v4f16 = call half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmin_v8f16 = call half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmin_v16f16 = call half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmin_v2f32 = call float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmin_v4f32 = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmin_v8f32 = call float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f64 = call double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmin_v4f64 = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmin_v4f128 = call fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fmin_strict'
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f16 = call half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmin_v4f16 = call half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 409 for instruction: %fmin_v8f16 = call half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 545 for instruction: %fmin_v16f16 = call half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f32 = call float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmin_v4f32 = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 205 for instruction: %fmin_v8f32 = call float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f64 = call double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmin_v4f64 = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmin_v4f128 = call fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
%fmin_v2f16 = call half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
%fmin_v4f16 = call half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
%fmin_v8f16 = call half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
%fmin_v16f16 = call half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
%fmin_v2f32 = call float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
%fmin_v4f32 = call float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
%fmin_v8f32 = call float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
%fmin_v2f64 = call double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
%fmin_v4f64 = call double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
%fmin_v4f128 = call fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
ret void
}
define void @fmin_unordered() {
; CHECK-V8-LABEL: 'fmin_unordered'
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmin_v2f16 = call reassoc half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %fmin_v4f16 = call reassoc half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %fmin_v8f16 = call reassoc half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 91 for instruction: %fmin_v16f16 = call reassoc half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmin_v2f32 = call reassoc float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmin_v4f32 = call reassoc float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmin_v8f32 = call reassoc float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmin_v2f64 = call reassoc double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmin_v4f64 = call reassoc double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %fmin_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fmin_unordered'
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmin_v2f16 = call reassoc half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmin_v4f16 = call reassoc half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmin_v8f16 = call reassoc half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmin_v16f16 = call reassoc half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmin_v2f32 = call reassoc float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmin_v4f32 = call reassoc float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmin_v8f32 = call reassoc float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f64 = call reassoc double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmin_v4f64 = call reassoc double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmin_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fmin_unordered'
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f16 = call reassoc half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmin_v4f16 = call reassoc half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 409 for instruction: %fmin_v8f16 = call reassoc half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 545 for instruction: %fmin_v16f16 = call reassoc half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f32 = call reassoc float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmin_v4f32 = call reassoc float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 205 for instruction: %fmin_v8f32 = call reassoc float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmin_v2f64 = call reassoc double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmin_v4f64 = call reassoc double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmin_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
%fmin_v2f16 = call reassoc half @llvm.vector.reduce.fmin.v2f16(<2 x half> undef)
%fmin_v4f16 = call reassoc half @llvm.vector.reduce.fmin.v4f16(<4 x half> undef)
%fmin_v8f16 = call reassoc half @llvm.vector.reduce.fmin.v8f16(<8 x half> undef)
%fmin_v16f16 = call reassoc half @llvm.vector.reduce.fmin.v16f16(<16 x half> undef)
%fmin_v2f32 = call reassoc float @llvm.vector.reduce.fmin.v2f32(<2 x float> undef)
%fmin_v4f32 = call reassoc float @llvm.vector.reduce.fmin.v4f32(<4 x float> undef)
%fmin_v8f32 = call reassoc float @llvm.vector.reduce.fmin.v8f32(<8 x float> undef)
%fmin_v2f64 = call reassoc double @llvm.vector.reduce.fmin.v2f64(<2 x double> undef)
%fmin_v4f64 = call reassoc double @llvm.vector.reduce.fmin.v4f64(<4 x double> undef)
%fmin_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128> undef)
ret void
}
define void @fmax_strict() {
; CHECK-V8-LABEL: 'fmax_strict'
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmax_v2f16 = call half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %fmax_v4f16 = call half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %fmax_v8f16 = call half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 91 for instruction: %fmax_v16f16 = call half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmax_v2f32 = call float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmax_v4f32 = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmax_v8f32 = call float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmax_v2f64 = call double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmax_v4f64 = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %fmax_v4f128 = call fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fmax_strict'
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmax_v2f16 = call half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmax_v4f16 = call half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmax_v8f16 = call half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmax_v16f16 = call half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmax_v2f32 = call float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmax_v4f32 = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmax_v8f32 = call float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f64 = call double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmax_v4f64 = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmax_v4f128 = call fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fmax_strict'
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f16 = call half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmax_v4f16 = call half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 409 for instruction: %fmax_v8f16 = call half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 545 for instruction: %fmax_v16f16 = call half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f32 = call float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmax_v4f32 = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 205 for instruction: %fmax_v8f32 = call float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f64 = call double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmax_v4f64 = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmax_v4f128 = call fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
%fmax_v2f16 = call half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
%fmax_v4f16 = call half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
%fmax_v8f16 = call half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
%fmax_v16f16 = call half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
%fmax_v2f32 = call float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
%fmax_v4f32 = call float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
%fmax_v8f32 = call float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
%fmax_v2f64 = call double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
%fmax_v4f64 = call double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
%fmax_v4f128 = call fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
ret void
}
define void @fmax_unordered() {
; CHECK-V8-LABEL: 'fmax_unordered'
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmax_v2f16 = call reassoc half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 19 for instruction: %fmax_v4f16 = call reassoc half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 43 for instruction: %fmax_v8f16 = call reassoc half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 91 for instruction: %fmax_v16f16 = call reassoc half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmax_v2f32 = call reassoc float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmax_v4f32 = call reassoc float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmax_v8f32 = call reassoc float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmax_v2f64 = call reassoc double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmax_v4f64 = call reassoc double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 94 for instruction: %fmax_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
; CHECK-V8-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEFP-LABEL: 'fmax_unordered'
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 2 for instruction: %fmax_v2f16 = call reassoc half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmax_v4f16 = call reassoc half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 7 for instruction: %fmax_v8f16 = call reassoc half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 9 for instruction: %fmax_v16f16 = call reassoc half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 1 for instruction: %fmax_v2f32 = call reassoc float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 3 for instruction: %fmax_v4f32 = call reassoc float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 5 for instruction: %fmax_v8f32 = call reassoc float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f64 = call reassoc double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmax_v4f64 = call reassoc double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmax_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
; CHECK-MVEFP-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
; CHECK-MVEI-LABEL: 'fmax_unordered'
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f16 = call reassoc half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmax_v4f16 = call reassoc half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 409 for instruction: %fmax_v8f16 = call reassoc half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 545 for instruction: %fmax_v16f16 = call reassoc half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f32 = call reassoc float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 137 for instruction: %fmax_v4f32 = call reassoc float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 205 for instruction: %fmax_v8f32 = call reassoc float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 35 for instruction: %fmax_v2f64 = call reassoc double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 69 for instruction: %fmax_v4f64 = call reassoc double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 118 for instruction: %fmax_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
; CHECK-MVEI-NEXT: Cost Model: Found an estimated cost of 0 for instruction: ret void
;
%fmax_v2f16 = call reassoc half @llvm.vector.reduce.fmax.v2f16(<2 x half> undef)
%fmax_v4f16 = call reassoc half @llvm.vector.reduce.fmax.v4f16(<4 x half> undef)
%fmax_v8f16 = call reassoc half @llvm.vector.reduce.fmax.v8f16(<8 x half> undef)
%fmax_v16f16 = call reassoc half @llvm.vector.reduce.fmax.v16f16(<16 x half> undef)
%fmax_v2f32 = call reassoc float @llvm.vector.reduce.fmax.v2f32(<2 x float> undef)
%fmax_v4f32 = call reassoc float @llvm.vector.reduce.fmax.v4f32(<4 x float> undef)
%fmax_v8f32 = call reassoc float @llvm.vector.reduce.fmax.v8f32(<8 x float> undef)
%fmax_v2f64 = call reassoc double @llvm.vector.reduce.fmax.v2f64(<2 x double> undef)
%fmax_v4f64 = call reassoc double @llvm.vector.reduce.fmax.v4f64(<4 x double> undef)
%fmax_v4f128 = call reassoc fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128> undef)
ret void
}
declare half @llvm.vector.reduce.fmin.v2f16(<2 x half>)
declare half @llvm.vector.reduce.fmin.v4f16(<4 x half>)
declare half @llvm.vector.reduce.fmin.v8f16(<8 x half>)
declare half @llvm.vector.reduce.fmin.v16f16(<16 x half>)
declare float @llvm.vector.reduce.fmin.v2f32(<2 x float>)
declare float @llvm.vector.reduce.fmin.v4f32(<4 x float>)
declare float @llvm.vector.reduce.fmin.v8f32(<8 x float>)
declare double @llvm.vector.reduce.fmin.v2f64(<2 x double>)
declare double @llvm.vector.reduce.fmin.v4f64(<4 x double>)
declare fp128 @llvm.vector.reduce.fmin.v4f128(<4 x fp128>)
declare half @llvm.vector.reduce.fmax.v2f16(<2 x half>)
declare half @llvm.vector.reduce.fmax.v4f16(<4 x half>)
declare half @llvm.vector.reduce.fmax.v8f16(<8 x half>)
declare half @llvm.vector.reduce.fmax.v16f16(<16 x half>)
declare float @llvm.vector.reduce.fmax.v2f32(<2 x float>)
declare float @llvm.vector.reduce.fmax.v4f32(<4 x float>)
declare float @llvm.vector.reduce.fmax.v8f32(<8 x float>)
declare double @llvm.vector.reduce.fmax.v2f64(<2 x double>)
declare double @llvm.vector.reduce.fmax.v4f64(<4 x double>)
declare fp128 @llvm.vector.reduce.fmax.v4f128(<4 x fp128>)
|
