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
245
246
247
248
249
250
251
|
# Copyright 2016 Advanced Micro Devices, Inc.
#
# Permission is hereby granted, free of charge, to any person obtaining a
# copy of this software and associated documentation files (the "Software"),
# to deal in the Software without restriction, including without limitation
# the rights to use, copy, modify, merge, publish, distribute, sublicense,
# and/or sell copies of the Software, and to permit persons to whom the
# Software is furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice (including the next
# paragraph) shall be included in all copies or substantial portions of the
# Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
from __future__ import print_function, division, absolute_import
import os
import random
import textwrap
from six.moves import range
from modules import utils
TYPES = ['char', 'uchar', 'short', 'ushort', 'int', 'uint', 'long', 'ulong', 'half', 'float', 'double']
VEC_SIZES = ['2', '3', '4', '8', '16']
DIR_NAME = os.path.join("cl", "vstore")
def gen_array(size):
random.seed(size)
return ' '.join([str(random.randint(0, 255)) for i in range(size)])
def ext_req(type_name):
if type_name[:6] == "double":
return "require_device_extensions: cl_khr_fp64"
if type_name[:4] == "half":
return "require_device_extensions: cl_khr_fp16"
return ""
def begin_test(suffix, type_name, mem_type, vec_sizes, addr_space, aligned):
file_name = os.path.join(DIR_NAME, "vstore{}-{}-{}.cl".format(suffix, type_name, addr_space))
print(file_name)
f = open(file_name, 'w')
f.write(textwrap.dedent(("""\
/*!
[config]
name: Vector store{suffix} {addr_space} {type_name}2,3,4,8,16
clc_version_min: 11
dimensions: 1
global_size: 1 0 0
""" + ext_req(type_name))
.format(type_name=type_name, addr_space=addr_space, suffix=suffix)))
for s in vec_sizes:
size = int(s) if s != '' else 1
modsize = 4 if size == 3 and aligned else size
offset = modsize if aligned else 1
canary= '0xdeadp1' if type_name in ('float', 'double') else '0xdead'
ty_name = type_name + s
f.write(textwrap.dedent("""
[test]
name: vector store{suffix} {addr_space} {type_name}
kernel_name: vstore{suffix}{n}_{addr_space}
arg_out: 0 buffer {mem_type}[{size}] {offset_zeros}{gen_array} {canary}
arg_in: 0 buffer {mem_type}[{size}] {offset_size_zeros} {canary}
arg_in: 1 buffer {type_name}[1] {gen_array}
[test]
name: vector store{suffix} {addr_space} offset {type_name}
kernel_name: vstore{suffix}{n}_{addr_space}_offset
arg_out: 0 buffer {mem_type}[{offset_size}] {offset_zeros} {gen_array} {padd_zeros} {gen_array} {canary}
arg_in: 0 buffer {mem_type}[{offset_size}] {offset_modsize_size_zeros} {canary}
arg_in: 1 buffer {type_name}[1] {gen_array}
""".format(type_name=ty_name, mem_type=mem_type, size=size + offset + 1,
offset_zeros = ("0 " * offset),
offset_size_zeros = ("0 " * (offset + size)),
padd_zeros = ("0 " * (modsize - size)),
offset_modsize_size_zeros = ("0 " * (modsize + size + offset)),
offset_size=modsize + size + offset + 1, n=s,
gen_array=gen_array(size),
suffix=suffix, addr_space=addr_space,
canary=canary)))
f.write(textwrap.dedent("""
!*/
"""))
if type_name == "double":
f.write(textwrap.dedent("""
#pragma OPENCL EXTENSION cl_khr_fp64: enable
"""))
if type_name == "half":
f.write(textwrap.dedent("""
#pragma OPENCL EXTENSION cl_khr_fp16: enable
"""))
return f
def gen_test_global(suffix, t, mem_type, vec_sizes, aligned):
f = begin_test(suffix, t, mem_type, vec_sizes, 'global', aligned)
for s in vec_sizes:
offset = int(s) if aligned else 1
offset = 4 if offset == 3 else offset
type_name = t + s
f.write(textwrap.dedent("""
kernel void vstore{suffix}{n}_global(global {mem_type} *out,
global {type_name} *in) {{
{type_name} tmp = in[0];
vstore{suffix}{n}(tmp, 0, out + {offset});
}}
kernel void vstore{suffix}{n}_global_offset(global {mem_type} *out,
global {type_name} *in) {{
{type_name} tmp = in[0];
vstore{suffix}{n}(tmp, 0, out + {offset});
vstore{suffix}{n}(tmp, 1, out + {offset});
}}
""".format(type_name=type_name, mem_type=mem_type, n=s, suffix=suffix,
offset=offset)))
f.close()
def gen_test_local_private(suffix, t, mem_type, vec_sizes, addr_space, aligned):
f = begin_test(suffix, t, mem_type, vec_sizes, addr_space, aligned)
for s in vec_sizes:
size = int(s) if s != '' else 1
modsize = 4 if size == 3 and aligned else size
offset = modsize if aligned else 1
type_name = t + s
f.write(textwrap.dedent("""
kernel void vstore{suffix}{n}_{addr_space}(global {mem_type} *out,
global {type_name} *in) {{
{type_name} tmp = in[0];
volatile {addr_space} {mem_type} loc[{size}];
for (int i = 0; i < {size}; ++i)
loc[i] = ({mem_type})0;
vstore{suffix}{n}(tmp, 0, ({addr_space} {mem_type}*)loc + {offset});
for (int i = 0; i < {size}; ++i)
out[i] = loc[i];
}}
kernel void vstore{suffix}{n}_{addr_space}_offset(global {mem_type} *out,
global {type_name} *in) {{
{type_name} tmp = in[0];
volatile {addr_space} {mem_type} loc[{offset_size}];
for (int i = 0; i < {offset_size}; ++i)
loc[i] = ({mem_type})0;
vstore{suffix}{n}(tmp, 0, ({addr_space} {mem_type}*)loc + {offset});
vstore{suffix}{n}(tmp, 1, ({addr_space} {mem_type}*)loc + {offset});
for (int i = 0; i < {offset_size}; ++i)
out[i] = loc[i];
}}
""".format(type_name=type_name, mem_type=mem_type, n=s, suffix=suffix,
offset_size=size + modsize + offset, size=size + offset,
addr_space=addr_space, offset=offset)))
f.close()
# vstore_half is special, because CLC won't allow us to use half type without
# cl_khr_fp16
def gen_test_local_private_half(suffix, t, vec_sizes, addr_space, aligned):
f = begin_test(suffix, t, 'half', vec_sizes, addr_space, aligned)
for s in vec_sizes:
size = int(s) if s != '' else 1
modsize = 4 if size == 3 and aligned else size
offset = modsize if aligned else 1
type_name = t + s
f.write(textwrap.dedent("""
kernel void vstore{suffix}{n}_{addr_space}(global half *out,
global {type_name} *in) {{
{type_name} tmp = in[0];
volatile {addr_space} short loc[{size}];
for (int i = 0; i < {size}; ++i)
loc[i] = 0;
vstore{suffix}{n}(tmp, 0, ({addr_space} half*)loc + {offset});
for (int i = 0; i < {size}; ++i)
((global short *)out)[i] = loc[i];
}}
kernel void vstore{suffix}{n}_{addr_space}_offset(global half *out,
global {type_name} *in) {{
{type_name} tmp = in[0];
volatile {addr_space} short loc[{offset_size}];
for (int i = 0; i < {offset_size}; ++i)
loc[i] = 0;
vstore{suffix}{n}(tmp, 0, ({addr_space} half*)loc + {offset});
vstore{suffix}{n}(tmp, 1, ({addr_space} half*)loc + {offset});
for (int i = 0; i < {offset_size}; ++i)
((global short *)out)[i] = loc[i];
}}
""".format(type_name=type_name, n=s, suffix=suffix,
offset_size=size + modsize + offset, size=size + offset,
addr_space=addr_space, offset=offset)))
def gen_test_local(suffix, t, mem_type, vec_sizes, aligned):
if mem_type == 'half':
gen_test_local_private_half(suffix, t, vec_sizes, 'local', aligned)
else:
gen_test_local_private(suffix, t, mem_type, vec_sizes, 'local', aligned)
def gen_test_private(suffix, t, mem_type, vec_sizes, aligned):
if mem_type == 'half':
gen_test_local_private_half(suffix, t, vec_sizes, 'private', aligned)
else:
gen_test_local_private(suffix, t, mem_type, vec_sizes, 'private', aligned)
def main():
utils.safe_makedirs(DIR_NAME)
for t in TYPES:
gen_test_global('', t, t, VEC_SIZES, False);
gen_test_local('', t, t, VEC_SIZES, False);
gen_test_private('', t, t, VEC_SIZES, False);
for aligned in False, True:
suffix = "a_half" if aligned else "_half"
vec_sizes = VEC_SIZES if aligned else [''] + VEC_SIZES
gen_test_global(suffix, 'float', 'half', vec_sizes, aligned);
gen_test_global(suffix, 'double', 'half', vec_sizes, aligned);
gen_test_local(suffix, 'float', 'half', vec_sizes, aligned);
gen_test_local(suffix, 'double', 'half', vec_sizes, aligned);
gen_test_private(suffix, 'float', 'half', vec_sizes, aligned);
gen_test_private(suffix, 'double', 'half', vec_sizes, aligned);
if __name__ == '__main__':
main()
|
