Mercurial > hg
view contrib/revsetbenchmarks.py @ 42044:bb271ec2fbfb
compression: introduce a `storage.revlog.zstd.level` configuration
This option control the zstd compression level used when compressing revlog
chunk. The usage of zstd for revlog compression has not graduated from
experimental yet, but we intend to fix that soon.
The option name for the compression level is more straight forward to pick, so
this changesets comes first. Having a dedicated option for each compression
engine is useful because they don't support the same range of values.
I ran the same measurement as for the zlib compression level (in the parent
changesets). The variation in repository size is stay mostly in the same (small)
range. The "read/write" performance see smallish variation, but are overall much
better than zlib. Write performance show the same tend of having better write
performance for when reaching high-end compression.
Again, we don't intend to change the default zstd compression level (currently:
3) in this series. However this is worth investigating in the future.
The Performance comparison of zlib vs zstd is quite impressive. The repository
size stay in the same range, but the performance are much better in all
situations.
Comparison summary
==================
We are looking at:
- performance range for zlib
- performance range for zstd
- comparison of default zstd (level-3) to default zlib (level 6)
- comparison of the slowest zstd time to the fastest zlib time
Read performance:
-----------------
| zlib | zstd | cmp | f2s
mercurial | 0.170159 - 0.189219 | 0.144127 - 0.149624 | 80% | 88%
pypy | 2.679217 - 2.768691 | 1.532317 - 1.705044 | 60% | 63%
netbeans | 122.477027 - 141.620281 | 72.996346 - 89.731560 | 58% | 73%
mozilla | 147.867662 - 170.572118 | 91.700995 - 105.853099 | 56% | 71%
Write performance:
------------------
| zlib | zstd | cmp | f2s
mercurial | 53.250304 - 56.2936129 | 40.877025 - 45.677286 | 75% | 86%
pypy | 460.721984 - 476.589918 | 270.545409 - 301.002219 | 63% | 65%
netbeans | 520.560316 - 715.930400 | 370.356311 - 428.329652 | 55% | 82%
mozilla | 739.803002 - 987.056093 | 505.152906 - 591.930683 | 57% | 80%
Raw data
--------
repo alg lvl .hg/store size 00manifest.d read write
mercurial zlib 1 49,402,813 5,963,475 0.170159 53.250304
mercurial zlib 6 47,197,397 5,875,730 0.182820 56.264320
mercurial zlib 9 47,121,596 5,849,781 0.189219 56.293612
mercurial zstd 1 49,737,084 5,966,355 0.144127 40.877025
mercurial zstd 3 48,961,867 5,895,208 0.146376 42.268142
mercurial zstd 5 48,200,592 5,938,676 0.149624 43.162875
mercurial zstd 10 47,833,520 5,913,353 0.145185 44.012489
mercurial zstd 15 47,314,604 5,728,679 0.147686 45.677286
mercurial zstd 20 47,330,502 5,830,539 0.145789 45.025407
mercurial zstd 22 47,330,076 5,830,539 0.143996 44.690460
pypy zlib 1 370,830,572 28,462,425 2.679217 460.721984
pypy zlib 6 340,112,317 27,648,747 2.768691 467.537158
pypy zlib 9 338,360,736 27,639,003 2.763495 476.589918
pypy zstd 1 362,377,479 27,916,214 1.532317 270.545409
pypy zstd 3 354,137,693 27,905,988 1.686718 294.951509
pypy zstd 5 342,640,043 27,655,774 1.705044 301.002219
pypy zstd 10 334,224,327 27,164,493 1.567287 285.186239
pypy zstd 15 329,000,363 26,645,965 1.637729 299.561332
pypy zstd 20 324,534,039 26,199,547 1.526813 302.149827
pypy zstd 22 324,530,595 26,198,932 1.525718 307.821218
netbeans zlib 1 1,281,847,810 165,495,457 122.477027 520.560316
netbeans zlib 6 1,205,284,353 159,161,207 139.876147 715.930400
netbeans zlib 9 1,197,135,671 155,034,586 141.620281 678.297064
netbeans zstd 1 1,259,581,737 160,840,613 72.996346 370.356311
netbeans zstd 3 1,232,978,122 157,691,551 81.622317 396.733087
netbeans zstd 5 1,208,034,075 160,246,880 83.080549 364.342626
netbeans zstd 10 1,188,624,176 156,083,417 79.323935 403.594602
netbeans zstd 15 1,176,973,589 153,859,477 89.731560 428.329652
netbeans zstd 20 1,162,958,258 151,147,535 82.842667 392.335349
netbeans zstd 22 1,162,707,029 151,150,220 82.565695 402.840655
mozilla zlib 1 2,775,497,186 298,527,987 147.867662 751.263721
mozilla zlib 6 2,596,856,420 286,597,671 170.572118 987.056093
mozilla zlib 9 2,587,542,494 287,018,264 163.622338 739.803002
mozilla zstd 1 2,723,159,348 286,617,532 91.700995 570.042751
mozilla zstd 3 2,665,055,001 286,152,013 95.240155 561.412805
mozilla zstd 5 2,607,819,817 288,060,030 101.978048 505.152906
mozilla zstd 10 2,558,761,085 283,967,648 104.113481 497.771202
mozilla zstd 15 2,526,216,060 275,581,300 105.853099 591.930683
mozilla zstd 20 2,485,114,806 266,478,859 95.268795 576.515389
mozilla zstd 22 2,484,869,080 266,456,505 94.429282 572.785537
author | Pierre-Yves David <pierre-yves.david@octobus.net> |
---|---|
date | Wed, 27 Mar 2019 18:35:59 +0100 |
parents | fbb43514f342 |
children | 2372284d9457 |
line wrap: on
line source
#!/usr/bin/env python # Measure the performance of a list of revsets against multiple revisions # defined by parameter. Checkout one by one and run perfrevset with every # revset in the list to benchmark its performance. # # You should run this from the root of your mercurial repository. # # call with --help for details from __future__ import absolute_import, print_function import math import optparse # cannot use argparse, python 2.7 only import os import re import subprocess import sys DEFAULTVARIANTS = ['plain', 'min', 'max', 'first', 'last', 'reverse', 'reverse+first', 'reverse+last', 'sort', 'sort+first', 'sort+last'] def check_output(*args, **kwargs): kwargs.setdefault('stderr', subprocess.PIPE) kwargs.setdefault('stdout', subprocess.PIPE) proc = subprocess.Popen(*args, **kwargs) output, error = proc.communicate() if proc.returncode != 0: raise subprocess.CalledProcessError(proc.returncode, ' '.join(args[0])) return output def update(rev): """update the repo to a revision""" try: subprocess.check_call(['hg', 'update', '--quiet', '--check', str(rev)]) check_output(['make', 'local'], stderr=None) # suppress output except for error/warning except subprocess.CalledProcessError as exc: print('update to revision %s failed, aborting'%rev, file=sys.stderr) sys.exit(exc.returncode) def hg(cmd, repo=None): """run a mercurial command <cmd> is the list of command + argument, <repo> is an optional repository path to run this command in.""" fullcmd = ['./hg'] if repo is not None: fullcmd += ['-R', repo] fullcmd += ['--config', 'extensions.perf=' + os.path.join(contribdir, 'perf.py')] fullcmd += cmd return check_output(fullcmd, stderr=subprocess.STDOUT) def perf(revset, target=None, contexts=False): """run benchmark for this very revset""" try: args = ['perfrevset'] if contexts: args.append('--contexts') args.append('--') args.append(revset) output = hg(args, repo=target) return parseoutput(output) except subprocess.CalledProcessError as exc: print('abort: cannot run revset benchmark: %s'%exc.cmd, file=sys.stderr) if getattr(exc, 'output', None) is None: # no output before 2.7 print('(no output)', file=sys.stderr) else: print(exc.output, file=sys.stderr) return None outputre = re.compile(br'! wall (\d+.\d+) comb (\d+.\d+) user (\d+.\d+) ' br'sys (\d+.\d+) \(best of (\d+)\)') def parseoutput(output): """parse a textual output into a dict We cannot just use json because we want to compare with old versions of Mercurial that may not support json output. """ match = outputre.search(output) if not match: print('abort: invalid output:', file=sys.stderr) print(output, file=sys.stderr) sys.exit(1) return {'comb': float(match.group(2)), 'count': int(match.group(5)), 'sys': float(match.group(3)), 'user': float(match.group(4)), 'wall': float(match.group(1)), } def printrevision(rev): """print data about a revision""" sys.stdout.write("Revision ") sys.stdout.flush() subprocess.check_call(['hg', 'log', '--rev', str(rev), '--template', '{if(tags, " ({tags})")} ' '{rev}:{node|short}: {desc|firstline}\n']) def idxwidth(nbidx): """return the max width of number used for index This is similar to log10(nbidx), but we use custom code here because we start with zero and we'd rather not deal with all the extra rounding business that log10 would imply. """ nbidx -= 1 # starts at 0 idxwidth = 0 while nbidx: idxwidth += 1 nbidx //= 10 if not idxwidth: idxwidth = 1 return idxwidth def getfactor(main, other, field, sensitivity=0.05): """return the relative factor between values for 'field' in main and other Return None if the factor is insignificant (less than <sensitivity> variation).""" factor = 1 if main is not None: factor = other[field] / main[field] low, high = 1 - sensitivity, 1 + sensitivity if (low < factor < high): return None return factor def formatfactor(factor): """format a factor into a 4 char string 22% 156% x2.4 x23 x789 x1e4 x5x7 """ if factor is None: return ' ' elif factor < 2: return '%3i%%' % (factor * 100) elif factor < 10: return 'x%3.1f' % factor elif factor < 1000: return '%4s' % ('x%i' % factor) else: order = int(math.log(factor)) + 1 while math.log(factor) > 1: factor //= 0 return 'x%ix%i' % (factor, order) def formattiming(value): """format a value to strictly 8 char, dropping some precision if needed""" if value < 10**7: return ('%.6f' % value)[:8] else: # value is HUGE very unlikely to happen (4+ month run) return '%i' % value _marker = object() def printresult(variants, idx, data, maxidx, verbose=False, reference=_marker): """print a line of result to stdout""" mask = '%%0%ii) %%s' % idxwidth(maxidx) out = [] for var in variants: if data[var] is None: out.append('error ') out.append(' ' * 4) continue out.append(formattiming(data[var]['wall'])) if reference is not _marker: factor = None if reference is not None: factor = getfactor(reference[var], data[var], 'wall') out.append(formatfactor(factor)) if verbose: out.append(formattiming(data[var]['comb'])) out.append(formattiming(data[var]['user'])) out.append(formattiming(data[var]['sys'])) out.append('%6d' % data[var]['count']) print(mask % (idx, ' '.join(out))) def printheader(variants, maxidx, verbose=False, relative=False): header = [' ' * (idxwidth(maxidx) + 1)] for var in variants: if not var: var = 'iter' if len(var) > 8: var = var[:3] + '..' + var[-3:] header.append('%-8s' % var) if relative: header.append(' ') if verbose: header.append('%-8s' % 'comb') header.append('%-8s' % 'user') header.append('%-8s' % 'sys') header.append('%6s' % 'count') print(' '.join(header)) def getrevs(spec): """get the list of rev matched by a revset""" try: out = check_output(['hg', 'log', '--template={rev}\n', '--rev', spec]) except subprocess.CalledProcessError as exc: print("abort, can't get revision from %s"%spec, file=sys.stderr) sys.exit(exc.returncode) return [r for r in out.split() if r] def applyvariants(revset, variant): if variant == 'plain': return revset for var in variant.split('+'): revset = '%s(%s)' % (var, revset) return revset helptext="""This script will run multiple variants of provided revsets using different revisions in your mercurial repository. After the benchmark are run summary output is provided. Use it to demonstrate speed improvements or pin point regressions. Revsets to run are specified in a file (or from stdin), one revsets per line. Line starting with '#' will be ignored, allowing insertion of comments.""" parser = optparse.OptionParser(usage="usage: %prog [options] <revs>", description=helptext) parser.add_option("-f", "--file", help="read revset from FILE (stdin if omitted)", metavar="FILE") parser.add_option("-R", "--repo", help="run benchmark on REPO", metavar="REPO") parser.add_option("-v", "--verbose", action='store_true', help="display all timing data (not just best total time)") parser.add_option("", "--variants", default=','.join(DEFAULTVARIANTS), help="comma separated list of variant to test " "(eg: plain,min,sorted) (plain = no modification)") parser.add_option('', '--contexts', action='store_true', help='obtain changectx from results instead of integer revs') (options, args) = parser.parse_args() if not args: parser.print_help() sys.exit(255) # the directory where both this script and the perf.py extension live. contribdir = os.path.dirname(__file__) revsetsfile = sys.stdin if options.file: revsetsfile = open(options.file) revsets = [l.strip() for l in revsetsfile if not l.startswith('#')] revsets = [l for l in revsets if l] print("Revsets to benchmark") print("----------------------------") for idx, rset in enumerate(revsets): print("%i) %s" % (idx, rset)) print("----------------------------") print() revs = [] for a in args: revs.extend(getrevs(a)) variants = options.variants.split(',') results = [] for r in revs: print("----------------------------") printrevision(r) print("----------------------------") update(r) res = [] results.append(res) printheader(variants, len(revsets), verbose=options.verbose) for idx, rset in enumerate(revsets): varres = {} for var in variants: varrset = applyvariants(rset, var) data = perf(varrset, target=options.repo, contexts=options.contexts) varres[var] = data res.append(varres) printresult(variants, idx, varres, len(revsets), verbose=options.verbose) sys.stdout.flush() print("----------------------------") print(""" Result by revset ================ """) print('Revision:') for idx, rev in enumerate(revs): sys.stdout.write('%i) ' % idx) sys.stdout.flush() printrevision(rev) print() print() for ridx, rset in enumerate(revsets): print("revset #%i: %s" % (ridx, rset)) printheader(variants, len(results), verbose=options.verbose, relative=True) ref = None for idx, data in enumerate(results): printresult(variants, idx, data[ridx], len(results), verbose=options.verbose, reference=ref) ref = data[ridx] print()