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
|
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=x86-64 -mattr=+f16c | FileCheck %s --check-prefix=CHECK --check-prefix=GENERIC
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=ivybridge | FileCheck %s --check-prefix=CHECK --check-prefix=IVY
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=haswell | FileCheck %s --check-prefix=CHECK --check-prefix=HASWELL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=broadwell | FileCheck %s --check-prefix=CHECK --check-prefix=BROADWELL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=skylake | FileCheck %s --check-prefix=CHECK --check-prefix=SKYLAKE
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=btver2 | FileCheck %s --check-prefix=CHECK --check-prefix=BTVER2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=znver1 | FileCheck %s --check-prefix=CHECK --check-prefix=ZNVER1
define <4 x float> @test_vcvtph2ps_128(<8 x i16> %a0, <8 x i16> *%a1) {
; GENERIC-LABEL: test_vcvtph2ps_128:
; GENERIC: # %bb.0:
; GENERIC-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [8:1.00]
; GENERIC-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT: retq # sched: [1:1.00]
;
; IVY-LABEL: test_vcvtph2ps_128:
; IVY: # %bb.0:
; IVY-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [8:1.00]
; IVY-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [3:1.00]
; IVY-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; IVY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: test_vcvtph2ps_128:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [6:1.00]
; HASWELL-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [2:1.00]
; HASWELL-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; BROADWELL-LABEL: test_vcvtph2ps_128:
; BROADWELL: # %bb.0:
; BROADWELL-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [6:1.00]
; BROADWELL-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [2:1.00]
; BROADWELL-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; BROADWELL-NEXT: retq # sched: [7:1.00]
;
; SKYLAKE-LABEL: test_vcvtph2ps_128:
; SKYLAKE: # %bb.0:
; SKYLAKE-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [9:0.50]
; SKYLAKE-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [5:1.00]
; SKYLAKE-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [4:0.50]
; SKYLAKE-NEXT: retq # sched: [7:1.00]
;
; BTVER2-LABEL: test_vcvtph2ps_128:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [8:1.00]
; BTVER2-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [3:1.00]
; BTVER2-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; ZNVER1-LABEL: test_vcvtph2ps_128:
; ZNVER1: # %bb.0:
; ZNVER1-NEXT: vcvtph2ps (%rdi), %xmm1 # sched: [100:0.25]
; ZNVER1-NEXT: vcvtph2ps %xmm0, %xmm0 # sched: [100:0.25]
; ZNVER1-NEXT: vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; ZNVER1-NEXT: retq # sched: [1:0.50]
%1 = load <8 x i16>, <8 x i16> *%a1
%2 = call <4 x float> @llvm.x86.vcvtph2ps.128(<8 x i16> %1)
%3 = call <4 x float> @llvm.x86.vcvtph2ps.128(<8 x i16> %a0)
%4 = fadd <4 x float> %2, %3
ret <4 x float> %4
}
declare <4 x float> @llvm.x86.vcvtph2ps.128(<8 x i16>)
define <8 x float> @test_vcvtph2ps_256(<8 x i16> %a0, <8 x i16> *%a1) {
; GENERIC-LABEL: test_vcvtph2ps_256:
; GENERIC: # %bb.0:
; GENERIC-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [8:1.00]
; GENERIC-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [3:1.00]
; GENERIC-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
; GENERIC-NEXT: retq # sched: [1:1.00]
;
; IVY-LABEL: test_vcvtph2ps_256:
; IVY: # %bb.0:
; IVY-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [8:1.00]
; IVY-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [3:1.00]
; IVY-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
; IVY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: test_vcvtph2ps_256:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [7:1.00]
; HASWELL-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [2:1.00]
; HASWELL-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; BROADWELL-LABEL: test_vcvtph2ps_256:
; BROADWELL: # %bb.0:
; BROADWELL-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [6:1.00]
; BROADWELL-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [2:1.00]
; BROADWELL-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
; BROADWELL-NEXT: retq # sched: [7:1.00]
;
; SKYLAKE-LABEL: test_vcvtph2ps_256:
; SKYLAKE: # %bb.0:
; SKYLAKE-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [10:0.50]
; SKYLAKE-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [7:1.00]
; SKYLAKE-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [4:0.50]
; SKYLAKE-NEXT: retq # sched: [7:1.00]
;
; BTVER2-LABEL: test_vcvtph2ps_256:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [8:2.00]
; BTVER2-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [3:2.00]
; BTVER2-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; ZNVER1-LABEL: test_vcvtph2ps_256:
; ZNVER1: # %bb.0:
; ZNVER1-NEXT: vcvtph2ps (%rdi), %ymm1 # sched: [100:0.25]
; ZNVER1-NEXT: vcvtph2ps %xmm0, %ymm0 # sched: [100:0.25]
; ZNVER1-NEXT: vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
; ZNVER1-NEXT: retq # sched: [1:0.50]
%1 = load <8 x i16>, <8 x i16> *%a1
%2 = call <8 x float> @llvm.x86.vcvtph2ps.256(<8 x i16> %1)
%3 = call <8 x float> @llvm.x86.vcvtph2ps.256(<8 x i16> %a0)
%4 = fadd <8 x float> %2, %3
ret <8 x float> %4
}
declare <8 x float> @llvm.x86.vcvtph2ps.256(<8 x i16>)
define <8 x i16> @test_vcvtps2ph_128(<4 x float> %a0, <4 x float> %a1, <4 x i16> *%a2) {
; GENERIC-LABEL: test_vcvtps2ph_128:
; GENERIC: # %bb.0:
; GENERIC-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [4:1.00]
; GENERIC-NEXT: retq # sched: [1:1.00]
;
; IVY-LABEL: test_vcvtps2ph_128:
; IVY: # %bb.0:
; IVY-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [3:1.00]
; IVY-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [4:1.00]
; IVY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: test_vcvtps2ph_128:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [4:1.00]
; HASWELL-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [5:1.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; BROADWELL-LABEL: test_vcvtps2ph_128:
; BROADWELL: # %bb.0:
; BROADWELL-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [4:1.00]
; BROADWELL-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [5:1.00]
; BROADWELL-NEXT: retq # sched: [7:1.00]
;
; SKYLAKE-LABEL: test_vcvtps2ph_128:
; SKYLAKE: # %bb.0:
; SKYLAKE-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [5:1.00]
; SKYLAKE-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [6:1.00]
; SKYLAKE-NEXT: retq # sched: [7:1.00]
;
; BTVER2-LABEL: test_vcvtps2ph_128:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [3:1.00]
; BTVER2-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [4:1.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; ZNVER1-LABEL: test_vcvtps2ph_128:
; ZNVER1: # %bb.0:
; ZNVER1-NEXT: vcvtps2ph $0, %xmm0, %xmm0 # sched: [100:0.25]
; ZNVER1-NEXT: vcvtps2ph $0, %xmm1, (%rdi) # sched: [100:0.25]
; ZNVER1-NEXT: retq # sched: [1:0.50]
%1 = call <8 x i16> @llvm.x86.vcvtps2ph.128(<4 x float> %a0, i32 0)
%2 = call <8 x i16> @llvm.x86.vcvtps2ph.128(<4 x float> %a1, i32 0)
%3 = shufflevector <8 x i16> %2, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x i16> %3, <4 x i16> *%a2
ret <8 x i16> %1
}
declare <8 x i16> @llvm.x86.vcvtps2ph.128(<4 x float>, i32)
define <8 x i16> @test_vcvtps2ph_256(<8 x float> %a0, <8 x float> %a1, <8 x i16> *%a2) {
; GENERIC-LABEL: test_vcvtps2ph_256:
; GENERIC: # %bb.0:
; GENERIC-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [4:1.00]
; GENERIC-NEXT: vzeroupper # sched: [100:0.33]
; GENERIC-NEXT: retq # sched: [1:1.00]
;
; IVY-LABEL: test_vcvtps2ph_256:
; IVY: # %bb.0:
; IVY-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [3:1.00]
; IVY-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [4:1.00]
; IVY-NEXT: vzeroupper # sched: [100:0.33]
; IVY-NEXT: retq # sched: [1:1.00]
;
; HASWELL-LABEL: test_vcvtps2ph_256:
; HASWELL: # %bb.0:
; HASWELL-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [6:1.00]
; HASWELL-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [7:1.00]
; HASWELL-NEXT: vzeroupper # sched: [4:1.00]
; HASWELL-NEXT: retq # sched: [7:1.00]
;
; BROADWELL-LABEL: test_vcvtps2ph_256:
; BROADWELL: # %bb.0:
; BROADWELL-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [6:1.00]
; BROADWELL-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [7:1.00]
; BROADWELL-NEXT: vzeroupper # sched: [4:1.00]
; BROADWELL-NEXT: retq # sched: [7:1.00]
;
; SKYLAKE-LABEL: test_vcvtps2ph_256:
; SKYLAKE: # %bb.0:
; SKYLAKE-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [7:1.00]
; SKYLAKE-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [8:1.00]
; SKYLAKE-NEXT: vzeroupper # sched: [4:1.00]
; SKYLAKE-NEXT: retq # sched: [7:1.00]
;
; BTVER2-LABEL: test_vcvtps2ph_256:
; BTVER2: # %bb.0:
; BTVER2-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [6:2.00]
; BTVER2-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [7:2.00]
; BTVER2-NEXT: retq # sched: [4:1.00]
;
; ZNVER1-LABEL: test_vcvtps2ph_256:
; ZNVER1: # %bb.0:
; ZNVER1-NEXT: vcvtps2ph $0, %ymm0, %xmm0 # sched: [100:0.25]
; ZNVER1-NEXT: vcvtps2ph $0, %ymm1, (%rdi) # sched: [100:0.25]
; ZNVER1-NEXT: vzeroupper # sched: [100:0.25]
; ZNVER1-NEXT: retq # sched: [1:0.50]
%1 = call <8 x i16> @llvm.x86.vcvtps2ph.256(<8 x float> %a0, i32 0)
%2 = call <8 x i16> @llvm.x86.vcvtps2ph.256(<8 x float> %a1, i32 0)
store <8 x i16> %2, <8 x i16> *%a2
ret <8 x i16> %1
}
declare <8 x i16> @llvm.x86.vcvtps2ph.256(<8 x float>, i32)
|
