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
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
|
#!/usr/bin/env python2.7
# Copyright 2017, Google Inc.
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
#
# * Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
# * Redistributions in binary form must reproduce the above
# copyright notice, this list of conditions and the following disclaimer
# in the documentation and/or other materials provided with the
# distribution.
# * Neither the name of Google Inc. nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import cgi
import multiprocessing
import os
import subprocess
import sys
import argparse
import python_utils.jobset as jobset
import python_utils.start_port_server as start_port_server
flamegraph_dir = os.path.join(os.path.expanduser('~'), 'FlameGraph')
os.chdir(os.path.join(os.path.dirname(sys.argv[0]), '../..'))
if not os.path.exists('reports'):
os.makedirs('reports')
start_port_server.start_port_server()
def fnize(s):
out = ''
for c in s:
if c in '<>, /':
if len(out) and out[-1] == '_': continue
out += '_'
else:
out += c
return out
# index html
index_html = """
<html>
<head>
<title>Microbenchmark Results</title>
</head>
<body>
"""
def heading(name):
global index_html
index_html += "<h1>%s</h1>\n" % name
def link(txt, tgt):
global index_html
index_html += "<p><a href=\"%s\">%s</a></p>\n" % (
cgi.escape(tgt, quote=True), cgi.escape(txt))
def text(txt):
global index_html
index_html += "<p><pre>%s</pre></p>\n" % cgi.escape(txt)
def collect_latency(bm_name, args):
"""generate latency profiles"""
benchmarks = []
profile_analysis = []
cleanup = []
heading('Latency Profiles: %s' % bm_name)
subprocess.check_call(
['make', bm_name,
'CONFIG=basicprof', '-j', '%d' % multiprocessing.cpu_count()])
for line in subprocess.check_output(['bins/basicprof/%s' % bm_name,
'--benchmark_list_tests']).splitlines():
link(line, '%s.txt' % fnize(line))
benchmarks.append(
jobset.JobSpec(['bins/basicprof/%s' % bm_name,
'--benchmark_filter=^%s$' % line,
'--benchmark_min_time=0.05'],
environ={'LATENCY_TRACE': '%s.trace' % fnize(line)}))
profile_analysis.append(
jobset.JobSpec([sys.executable,
'tools/profiling/latency_profile/profile_analyzer.py',
'--source', '%s.trace' % fnize(line), '--fmt', 'simple',
'--out', 'reports/%s.txt' % fnize(line)], timeout_seconds=None))
cleanup.append(jobset.JobSpec(['rm', '%s.trace' % fnize(line)]))
# periodically flush out the list of jobs: profile_analysis jobs at least
# consume upwards of five gigabytes of ram in some cases, and so analysing
# hundreds of them at once is impractical -- but we want at least some
# concurrency or the work takes too long
if len(benchmarks) >= min(16, multiprocessing.cpu_count()):
# run up to half the cpu count: each benchmark can use up to two cores
# (one for the microbenchmark, one for the data flush)
jobset.run(benchmarks, maxjobs=max(1, multiprocessing.cpu_count()/2))
jobset.run(profile_analysis, maxjobs=multiprocessing.cpu_count())
jobset.run(cleanup, maxjobs=multiprocessing.cpu_count())
benchmarks = []
profile_analysis = []
cleanup = []
# run the remaining benchmarks that weren't flushed
if len(benchmarks):
jobset.run(benchmarks, maxjobs=max(1, multiprocessing.cpu_count()/2))
jobset.run(profile_analysis, maxjobs=multiprocessing.cpu_count())
jobset.run(cleanup, maxjobs=multiprocessing.cpu_count())
def collect_perf(bm_name, args):
"""generate flamegraphs"""
heading('Flamegraphs: %s' % bm_name)
subprocess.check_call(
['make', bm_name,
'CONFIG=mutrace', '-j', '%d' % multiprocessing.cpu_count()])
benchmarks = []
profile_analysis = []
cleanup = []
for line in subprocess.check_output(['bins/mutrace/%s' % bm_name,
'--benchmark_list_tests']).splitlines():
link(line, '%s.svg' % fnize(line))
benchmarks.append(
jobset.JobSpec(['perf', 'record', '-o', '%s-perf.data' % fnize(line),
'-g', '-F', '997',
'bins/mutrace/%s' % bm_name,
'--benchmark_filter=^%s$' % line,
'--benchmark_min_time=10']))
profile_analysis.append(
jobset.JobSpec(['tools/run_tests/performance/process_local_perf_flamegraphs.sh'],
environ = {
'PERF_BASE_NAME': fnize(line),
'OUTPUT_DIR': 'reports',
'OUTPUT_FILENAME': fnize(line),
}))
cleanup.append(jobset.JobSpec(['rm', '%s-perf.data' % fnize(line)]))
cleanup.append(jobset.JobSpec(['rm', '%s-out.perf' % fnize(line)]))
# periodically flush out the list of jobs: temporary space required for this
# processing is large
if len(benchmarks) >= 20:
# run up to half the cpu count: each benchmark can use up to two cores
# (one for the microbenchmark, one for the data flush)
jobset.run(benchmarks, maxjobs=1)
jobset.run(profile_analysis, maxjobs=multiprocessing.cpu_count())
jobset.run(cleanup, maxjobs=multiprocessing.cpu_count())
benchmarks = []
profile_analysis = []
cleanup = []
# run the remaining benchmarks that weren't flushed
if len(benchmarks):
jobset.run(benchmarks, maxjobs=1)
jobset.run(profile_analysis, maxjobs=multiprocessing.cpu_count())
jobset.run(cleanup, maxjobs=multiprocessing.cpu_count())
def run_summary(bm_name, cfg, base_json_name):
subprocess.check_call(
['make', bm_name,
'CONFIG=%s' % cfg, '-j', '%d' % multiprocessing.cpu_count()])
cmd = ['bins/%s/%s' % (cfg, bm_name),
'--benchmark_out=%s.%s.json' % (base_json_name, cfg),
'--benchmark_out_format=json']
if args.summary_time is not None:
cmd += ['--benchmark_min_time=%d' % args.summary_time]
return subprocess.check_output(cmd)
def collect_summary(bm_name, args):
heading('Summary: %s [no counters]' % bm_name)
text(run_summary(bm_name, 'opt', bm_name))
heading('Summary: %s [with counters]' % bm_name)
text(run_summary(bm_name, 'counters', bm_name))
if args.bigquery_upload:
with open('%s.csv' % bm_name, 'w') as f:
f.write(subprocess.check_output(['tools/profiling/microbenchmarks/bm2bq.py',
'%s.counters.json' % bm_name,
'%s.opt.json' % bm_name]))
subprocess.check_call(['bq', 'load', 'microbenchmarks.microbenchmarks', '%s.csv' % bm_name])
collectors = {
'latency': collect_latency,
'perf': collect_perf,
'summary': collect_summary,
}
argp = argparse.ArgumentParser(description='Collect data from microbenchmarks')
argp.add_argument('-c', '--collect',
choices=sorted(collectors.keys()),
nargs='*',
default=sorted(collectors.keys()),
help='Which collectors should be run against each benchmark')
argp.add_argument('-b', '--benchmarks',
default=['bm_fullstack_unary_ping_pong',
'bm_fullstack_streaming_ping_pong',
'bm_fullstack_streaming_pump',
'bm_closure',
'bm_cq',
'bm_call_create',
'bm_error',
'bm_chttp2_hpack',
'bm_metadata',
'bm_fullstack_trickle',
],
nargs='+',
type=str,
help='Which microbenchmarks should be run')
argp.add_argument('--diff_perf',
default=None,
type=str,
help='Diff microbenchmarks against this git revision')
argp.add_argument('--bigquery_upload',
default=False,
action='store_const',
const=True,
help='Upload results from summary collection to bigquery')
argp.add_argument('--summary_time',
default=None,
type=int,
help='Minimum time to run benchmarks for the summary collection')
args = argp.parse_args()
try:
for collect in args.collect:
for bm_name in args.benchmarks:
collectors[collect](bm_name, args)
if args.diff_perf:
if 'summary' not in args.collect:
for bm_name in args.benchmarks:
run_summary(bm_name, 'opt', bm_name)
run_summary(bm_name, 'counters', bm_name)
where_am_i = subprocess.check_output(['git', 'rev-parse', '--abbrev-ref', 'HEAD']).strip()
subprocess.check_call(['git', 'checkout', args.diff_perf])
comparables = []
subprocess.check_call(['make', 'clean'])
try:
for bm_name in args.benchmarks:
try:
run_summary(bm_name, 'opt', '%s.old' % bm_name)
run_summary(bm_name, 'counters', '%s.old' % bm_name)
comparables.append(bm_name)
except subprocess.CalledProcessError, e:
pass
finally:
subprocess.check_call(['git', 'checkout', where_am_i])
for bm_name in comparables:
diff = subprocess.check_output(['tools/profiling/microbenchmarks/bm_diff.py',
'%s.counters.json' % bm_name,
'%s.opt.json' % bm_name,
'%s.old.counters.json' % bm_name,
'%s.old.opt.json' % bm_name]).strip()
if diff:
heading('Performance diff: %s' % bm_name)
text(diff)
finally:
index_html += "</body>\n</html>\n"
with open('reports/index.html', 'w') as f:
f.write(index_html)
|