648db22b |
1 | #! /usr/bin/env python3 |
2 | # THIS BENCHMARK IS BEING REPLACED BY automated-bencmarking.py |
3 | |
4 | # ################################################################ |
5 | # Copyright (c) Meta Platforms, Inc. and affiliates. |
6 | # All rights reserved. |
7 | # |
8 | # This source code is licensed under both the BSD-style license (found in the |
9 | # LICENSE file in the root directory of this source tree) and the GPLv2 (found |
10 | # in the COPYING file in the root directory of this source tree). |
11 | # You may select, at your option, one of the above-listed licenses. |
12 | # ########################################################################## |
13 | |
14 | # Limitations: |
15 | # - doesn't support filenames with spaces |
16 | # - dir1/zstd and dir2/zstd will be merged in a single results file |
17 | |
18 | import argparse |
19 | import os # getloadavg |
20 | import string |
21 | import subprocess |
22 | import time # strftime |
23 | import traceback |
24 | import hashlib |
25 | import platform # system |
26 | |
27 | script_version = 'v1.1.2 (2017-03-26)' |
28 | default_repo_url = 'https://github.com/facebook/zstd.git' |
29 | working_dir_name = 'speedTest' |
30 | working_path = os.getcwd() + '/' + working_dir_name # /path/to/zstd/tests/speedTest |
31 | clone_path = working_path + '/' + 'zstd' # /path/to/zstd/tests/speedTest/zstd |
32 | email_header = 'ZSTD_speedTest' |
33 | pid = str(os.getpid()) |
34 | verbose = False |
35 | clang_version = "unknown" |
36 | gcc_version = "unknown" |
37 | args = None |
38 | |
39 | |
40 | def hashfile(hasher, fname, blocksize=65536): |
41 | with open(fname, "rb") as f: |
42 | for chunk in iter(lambda: f.read(blocksize), b""): |
43 | hasher.update(chunk) |
44 | return hasher.hexdigest() |
45 | |
46 | |
47 | def log(text): |
48 | print(time.strftime("%Y/%m/%d %H:%M:%S") + ' - ' + text) |
49 | |
50 | |
51 | def execute(command, print_command=True, print_output=False, print_error=True, param_shell=True): |
52 | if print_command: |
53 | log("> " + command) |
54 | popen = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=param_shell, cwd=execute.cwd) |
55 | stdout_lines, stderr_lines = popen.communicate(timeout=args.timeout) |
56 | stderr_lines = stderr_lines.decode("utf-8") |
57 | stdout_lines = stdout_lines.decode("utf-8") |
58 | if print_output: |
59 | if stdout_lines: |
60 | print(stdout_lines) |
61 | if stderr_lines: |
62 | print(stderr_lines) |
63 | if popen.returncode is not None and popen.returncode != 0: |
64 | if stderr_lines and not print_output and print_error: |
65 | print(stderr_lines) |
66 | raise RuntimeError(stdout_lines + stderr_lines) |
67 | return (stdout_lines + stderr_lines).splitlines() |
68 | execute.cwd = None |
69 | |
70 | |
71 | def does_command_exist(command): |
72 | try: |
73 | execute(command, verbose, False, False) |
74 | except Exception: |
75 | return False |
76 | return True |
77 | |
78 | |
79 | def send_email(emails, topic, text, have_mutt, have_mail): |
80 | logFileName = working_path + '/' + 'tmpEmailContent' |
81 | with open(logFileName, "w") as myfile: |
82 | myfile.writelines(text) |
83 | myfile.close() |
84 | if have_mutt: |
85 | execute('mutt -s "' + topic + '" ' + emails + ' < ' + logFileName, verbose) |
86 | elif have_mail: |
87 | execute('mail -s "' + topic + '" ' + emails + ' < ' + logFileName, verbose) |
88 | else: |
89 | log("e-mail cannot be sent (mail or mutt not found)") |
90 | |
91 | |
92 | def send_email_with_attachments(branch, commit, last_commit, args, text, results_files, |
93 | logFileName, have_mutt, have_mail): |
94 | with open(logFileName, "w") as myfile: |
95 | myfile.writelines(text) |
96 | myfile.close() |
97 | email_topic = '[%s:%s] Warning for %s:%s last_commit=%s speed<%s ratio<%s' \ |
98 | % (email_header, pid, branch, commit, last_commit, |
99 | args.lowerLimit, args.ratioLimit) |
100 | if have_mutt: |
101 | execute('mutt -s "' + email_topic + '" ' + args.emails + ' -a ' + results_files |
102 | + ' < ' + logFileName) |
103 | elif have_mail: |
104 | execute('mail -s "' + email_topic + '" ' + args.emails + ' < ' + logFileName) |
105 | else: |
106 | log("e-mail cannot be sent (mail or mutt not found)") |
107 | |
108 | |
109 | def git_get_branches(): |
110 | execute('git fetch -p', verbose) |
111 | branches = execute('git branch -rl', verbose) |
112 | output = [] |
113 | for line in branches: |
114 | if ("HEAD" not in line) and ("coverity_scan" not in line) and ("gh-pages" not in line): |
115 | output.append(line.strip()) |
116 | return output |
117 | |
118 | |
119 | def git_get_changes(branch, commit, last_commit): |
120 | fmt = '--format="%h: (%an) %s, %ar"' |
121 | if last_commit is None: |
122 | commits = execute('git log -n 10 %s %s' % (fmt, commit)) |
123 | else: |
124 | commits = execute('git --no-pager log %s %s..%s' % (fmt, last_commit, commit)) |
125 | return str('Changes in %s since %s:\n' % (branch, last_commit)) + '\n'.join(commits) |
126 | |
127 | |
128 | def get_last_results(resultsFileName): |
129 | if not os.path.isfile(resultsFileName): |
130 | return None, None, None, None |
131 | commit = None |
132 | csize = [] |
133 | cspeed = [] |
134 | dspeed = [] |
135 | with open(resultsFileName, 'r') as f: |
136 | for line in f: |
137 | words = line.split() |
138 | if len(words) <= 4: # branch + commit + compilerVer + md5 |
139 | commit = words[1] |
140 | csize = [] |
141 | cspeed = [] |
142 | dspeed = [] |
143 | if (len(words) == 8) or (len(words) == 9): # results: "filename" or "XX files" |
144 | csize.append(int(words[1])) |
145 | cspeed.append(float(words[3])) |
146 | dspeed.append(float(words[5])) |
147 | return commit, csize, cspeed, dspeed |
148 | |
149 | |
150 | def benchmark_and_compare(branch, commit, last_commit, args, executableName, md5sum, compilerVersion, resultsFileName, |
151 | testFilePath, fileName, last_csize, last_cspeed, last_dspeed): |
152 | sleepTime = 30 |
153 | while os.getloadavg()[0] > args.maxLoadAvg: |
154 | log("WARNING: bench loadavg=%.2f is higher than %s, sleeping for %s seconds" |
155 | % (os.getloadavg()[0], args.maxLoadAvg, sleepTime)) |
156 | time.sleep(sleepTime) |
157 | start_load = str(os.getloadavg()) |
158 | osType = platform.system() |
159 | if osType == 'Linux': |
160 | cpuSelector = "taskset --cpu-list 0" |
161 | else: |
162 | cpuSelector = "" |
163 | if args.dictionary: |
164 | result = execute('%s programs/%s -rqi5b1e%s -D %s %s' % (cpuSelector, executableName, args.lastCLevel, args.dictionary, testFilePath), print_output=True) |
165 | else: |
166 | result = execute('%s programs/%s -rqi5b1e%s %s' % (cpuSelector, executableName, args.lastCLevel, testFilePath), print_output=True) |
167 | end_load = str(os.getloadavg()) |
168 | linesExpected = args.lastCLevel + 1 |
169 | if len(result) != linesExpected: |
170 | raise RuntimeError("ERROR: number of result lines=%d is different that expected %d\n%s" % (len(result), linesExpected, '\n'.join(result))) |
171 | with open(resultsFileName, "a") as myfile: |
172 | myfile.write('%s %s %s md5=%s\n' % (branch, commit, compilerVersion, md5sum)) |
173 | myfile.write('\n'.join(result) + '\n') |
174 | myfile.close() |
175 | if (last_cspeed == None): |
176 | log("WARNING: No data for comparison for branch=%s file=%s " % (branch, fileName)) |
177 | return "" |
178 | commit, csize, cspeed, dspeed = get_last_results(resultsFileName) |
179 | text = "" |
180 | for i in range(0, min(len(cspeed), len(last_cspeed))): |
181 | print("%s:%s -%d cSpeed=%6.2f cLast=%6.2f cDiff=%1.4f dSpeed=%6.2f dLast=%6.2f dDiff=%1.4f ratioDiff=%1.4f %s" % (branch, commit, i+1, cspeed[i], last_cspeed[i], cspeed[i]/last_cspeed[i], dspeed[i], last_dspeed[i], dspeed[i]/last_dspeed[i], float(last_csize[i])/csize[i], fileName)) |
182 | if (cspeed[i]/last_cspeed[i] < args.lowerLimit): |
183 | text += "WARNING: %s -%d cSpeed=%.2f cLast=%.2f cDiff=%.4f %s\n" % (executableName, i+1, cspeed[i], last_cspeed[i], cspeed[i]/last_cspeed[i], fileName) |
184 | if (dspeed[i]/last_dspeed[i] < args.lowerLimit): |
185 | text += "WARNING: %s -%d dSpeed=%.2f dLast=%.2f dDiff=%.4f %s\n" % (executableName, i+1, dspeed[i], last_dspeed[i], dspeed[i]/last_dspeed[i], fileName) |
186 | if (float(last_csize[i])/csize[i] < args.ratioLimit): |
187 | text += "WARNING: %s -%d cSize=%d last_cSize=%d diff=%.4f %s\n" % (executableName, i+1, csize[i], last_csize[i], float(last_csize[i])/csize[i], fileName) |
188 | if text: |
189 | text = args.message + ("\nmaxLoadAvg=%s load average at start=%s end=%s\n%s last_commit=%s md5=%s\n" % (args.maxLoadAvg, start_load, end_load, compilerVersion, last_commit, md5sum)) + text |
190 | return text |
191 | |
192 | |
193 | def update_config_file(branch, commit): |
194 | last_commit = None |
195 | commitFileName = working_path + "/commit_" + branch.replace("/", "_") + ".txt" |
196 | if os.path.isfile(commitFileName): |
197 | with open(commitFileName, 'r') as infile: |
198 | last_commit = infile.read() |
199 | with open(commitFileName, 'w') as outfile: |
200 | outfile.write(commit) |
201 | return last_commit |
202 | |
203 | |
204 | def double_check(branch, commit, args, executableName, md5sum, compilerVersion, resultsFileName, filePath, fileName): |
205 | last_commit, csize, cspeed, dspeed = get_last_results(resultsFileName) |
206 | if not args.dry_run: |
207 | text = benchmark_and_compare(branch, commit, last_commit, args, executableName, md5sum, compilerVersion, resultsFileName, filePath, fileName, csize, cspeed, dspeed) |
208 | if text: |
209 | log("WARNING: redoing tests for branch %s: commit %s" % (branch, commit)) |
210 | text = benchmark_and_compare(branch, commit, last_commit, args, executableName, md5sum, compilerVersion, resultsFileName, filePath, fileName, csize, cspeed, dspeed) |
211 | return text |
212 | |
213 | |
214 | def test_commit(branch, commit, last_commit, args, testFilePaths, have_mutt, have_mail): |
215 | local_branch = branch.split('/')[1] |
216 | version = local_branch.rpartition('-')[2] + '_' + commit |
217 | if not args.dry_run: |
218 | execute('make -C programs clean zstd CC=clang MOREFLAGS="-Werror -Wconversion -Wno-sign-conversion -DZSTD_GIT_COMMIT=%s" && ' % version + |
219 | 'mv programs/zstd programs/zstd_clang && ' + |
220 | 'make -C programs clean zstd zstd32 MOREFLAGS="-DZSTD_GIT_COMMIT=%s"' % version) |
221 | md5_zstd = hashfile(hashlib.md5(), clone_path + '/programs/zstd') |
222 | md5_zstd32 = hashfile(hashlib.md5(), clone_path + '/programs/zstd32') |
223 | md5_zstd_clang = hashfile(hashlib.md5(), clone_path + '/programs/zstd_clang') |
224 | print("md5(zstd)=%s\nmd5(zstd32)=%s\nmd5(zstd_clang)=%s" % (md5_zstd, md5_zstd32, md5_zstd_clang)) |
225 | print("gcc_version=%s clang_version=%s" % (gcc_version, clang_version)) |
226 | |
227 | logFileName = working_path + "/log_" + branch.replace("/", "_") + ".txt" |
228 | text_to_send = [] |
229 | results_files = "" |
230 | if args.dictionary: |
231 | dictName = args.dictionary.rpartition('/')[2] |
232 | else: |
233 | dictName = None |
234 | |
235 | for filePath in testFilePaths: |
236 | fileName = filePath.rpartition('/')[2] |
237 | if dictName: |
238 | resultsFileName = working_path + "/" + dictName.replace(".", "_") + "_" + branch.replace("/", "_") + "_" + fileName.replace(".", "_") + ".txt" |
239 | else: |
240 | resultsFileName = working_path + "/results_" + branch.replace("/", "_") + "_" + fileName.replace(".", "_") + ".txt" |
241 | text = double_check(branch, commit, args, 'zstd', md5_zstd, 'gcc_version='+gcc_version, resultsFileName, filePath, fileName) |
242 | if text: |
243 | text_to_send.append(text) |
244 | results_files += resultsFileName + " " |
245 | resultsFileName = working_path + "/results32_" + branch.replace("/", "_") + "_" + fileName.replace(".", "_") + ".txt" |
246 | text = double_check(branch, commit, args, 'zstd32', md5_zstd32, 'gcc_version='+gcc_version, resultsFileName, filePath, fileName) |
247 | if text: |
248 | text_to_send.append(text) |
249 | results_files += resultsFileName + " " |
250 | resultsFileName = working_path + "/resultsClang_" + branch.replace("/", "_") + "_" + fileName.replace(".", "_") + ".txt" |
251 | text = double_check(branch, commit, args, 'zstd_clang', md5_zstd_clang, 'clang_version='+clang_version, resultsFileName, filePath, fileName) |
252 | if text: |
253 | text_to_send.append(text) |
254 | results_files += resultsFileName + " " |
255 | if text_to_send: |
256 | send_email_with_attachments(branch, commit, last_commit, args, text_to_send, results_files, logFileName, have_mutt, have_mail) |
257 | |
258 | |
259 | if __name__ == '__main__': |
260 | parser = argparse.ArgumentParser() |
261 | parser.add_argument('testFileNames', help='file or directory names list for speed benchmark') |
262 | parser.add_argument('emails', help='list of e-mail addresses to send warnings') |
263 | parser.add_argument('--dictionary', '-D', help='path to the dictionary') |
264 | parser.add_argument('--message', '-m', help='attach an additional message to e-mail', default="") |
265 | parser.add_argument('--repoURL', help='changes default repository URL', default=default_repo_url) |
266 | parser.add_argument('--lowerLimit', '-l', type=float, help='send email if speed is lower than given limit', default=0.98) |
267 | parser.add_argument('--ratioLimit', '-r', type=float, help='send email if ratio is lower than given limit', default=0.999) |
268 | parser.add_argument('--maxLoadAvg', type=float, help='maximum load average to start testing', default=0.75) |
269 | parser.add_argument('--lastCLevel', type=int, help='last compression level for testing', default=5) |
270 | parser.add_argument('--sleepTime', '-s', type=int, help='frequency of repository checking in seconds', default=300) |
271 | parser.add_argument('--timeout', '-t', type=int, help='timeout for executing shell commands', default=1800) |
272 | parser.add_argument('--dry-run', dest='dry_run', action='store_true', help='not build', default=False) |
273 | parser.add_argument('--verbose', '-v', action='store_true', help='more verbose logs', default=False) |
274 | args = parser.parse_args() |
275 | verbose = args.verbose |
276 | |
277 | # check if test files are accessible |
278 | testFileNames = args.testFileNames.split() |
279 | testFilePaths = [] |
280 | for fileName in testFileNames: |
281 | fileName = os.path.expanduser(fileName) |
282 | if os.path.isfile(fileName) or os.path.isdir(fileName): |
283 | testFilePaths.append(os.path.abspath(fileName)) |
284 | else: |
285 | log("ERROR: File/directory not found: " + fileName) |
286 | exit(1) |
287 | |
288 | # check if dictionary is accessible |
289 | if args.dictionary: |
290 | args.dictionary = os.path.abspath(os.path.expanduser(args.dictionary)) |
291 | if not os.path.isfile(args.dictionary): |
292 | log("ERROR: Dictionary not found: " + args.dictionary) |
293 | exit(1) |
294 | |
295 | # check availability of e-mail senders |
296 | have_mutt = does_command_exist("mutt -h") |
297 | have_mail = does_command_exist("mail -V") |
298 | if not have_mutt and not have_mail: |
299 | log("ERROR: e-mail senders 'mail' or 'mutt' not found") |
300 | exit(1) |
301 | |
302 | clang_version = execute("clang -v 2>&1 | grep ' version ' | sed -e 's:.*version \\([0-9.]*\\).*:\\1:' -e 's:\\.\\([0-9][0-9]\\):\\1:g'", verbose)[0]; |
303 | gcc_version = execute("gcc -dumpversion", verbose)[0]; |
304 | |
305 | if verbose: |
306 | print("PARAMETERS:\nrepoURL=%s" % args.repoURL) |
307 | print("working_path=%s" % working_path) |
308 | print("clone_path=%s" % clone_path) |
309 | print("testFilePath(%s)=%s" % (len(testFilePaths), testFilePaths)) |
310 | print("message=%s" % args.message) |
311 | print("emails=%s" % args.emails) |
312 | print("dictionary=%s" % args.dictionary) |
313 | print("maxLoadAvg=%s" % args.maxLoadAvg) |
314 | print("lowerLimit=%s" % args.lowerLimit) |
315 | print("ratioLimit=%s" % args.ratioLimit) |
316 | print("lastCLevel=%s" % args.lastCLevel) |
317 | print("sleepTime=%s" % args.sleepTime) |
318 | print("timeout=%s" % args.timeout) |
319 | print("dry_run=%s" % args.dry_run) |
320 | print("verbose=%s" % args.verbose) |
321 | print("have_mutt=%s have_mail=%s" % (have_mutt, have_mail)) |
322 | |
323 | # clone ZSTD repo if needed |
324 | if not os.path.isdir(working_path): |
325 | os.mkdir(working_path) |
326 | if not os.path.isdir(clone_path): |
327 | execute.cwd = working_path |
328 | execute('git clone ' + args.repoURL) |
329 | if not os.path.isdir(clone_path): |
330 | log("ERROR: ZSTD clone not found: " + clone_path) |
331 | exit(1) |
332 | execute.cwd = clone_path |
333 | |
334 | # check if speedTest.pid already exists |
335 | pidfile = "./speedTest.pid" |
336 | if os.path.isfile(pidfile): |
337 | log("ERROR: %s already exists, exiting" % pidfile) |
338 | exit(1) |
339 | |
340 | send_email(args.emails, '[%s:%s] test-zstd-speed.py %s has been started' % (email_header, pid, script_version), args.message, have_mutt, have_mail) |
341 | with open(pidfile, 'w') as the_file: |
342 | the_file.write(pid) |
343 | |
344 | branch = "" |
345 | commit = "" |
346 | first_time = True |
347 | while True: |
348 | try: |
349 | if first_time: |
350 | first_time = False |
351 | else: |
352 | time.sleep(args.sleepTime) |
353 | loadavg = os.getloadavg()[0] |
354 | if (loadavg <= args.maxLoadAvg): |
355 | branches = git_get_branches() |
356 | for branch in branches: |
357 | commit = execute('git show -s --format=%h ' + branch, verbose)[0] |
358 | last_commit = update_config_file(branch, commit) |
359 | if commit == last_commit: |
360 | log("skipping branch %s: head %s already processed" % (branch, commit)) |
361 | else: |
362 | log("build branch %s: head %s is different from prev %s" % (branch, commit, last_commit)) |
363 | execute('git checkout -- . && git checkout ' + branch) |
364 | print(git_get_changes(branch, commit, last_commit)) |
365 | test_commit(branch, commit, last_commit, args, testFilePaths, have_mutt, have_mail) |
366 | else: |
367 | log("WARNING: main loadavg=%.2f is higher than %s" % (loadavg, args.maxLoadAvg)) |
368 | if verbose: |
369 | log("sleep for %s seconds" % args.sleepTime) |
370 | except Exception as e: |
371 | stack = traceback.format_exc() |
372 | email_topic = '[%s:%s] ERROR in %s:%s' % (email_header, pid, branch, commit) |
373 | send_email(args.emails, email_topic, stack, have_mutt, have_mail) |
374 | print(stack) |
375 | except KeyboardInterrupt: |
376 | os.unlink(pidfile) |
377 | send_email(args.emails, '[%s:%s] test-zstd-speed.py %s has been stopped' % (email_header, pid, script_version), args.message, have_mutt, have_mail) |
378 | exit(0) |