[Webkit-unassigned] [Bug 70995] The GC should be parallel
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Thu Oct 27 13:54:55 PDT 2011
https://bugs.webkit.org/show_bug.cgi?id=70995
--- Comment #6 from Filip Pizlo <fpizlo at apple.com> 2011-10-27 13:54:55 PST ---
Even when running with parallel GC enabled but with only 1 worker - so the load balancer is doing completely pointless work - this patch is basically performance-neutral. It's a slow-down on v8-splay (4%, ish) but this does not show up in the grand averages.
Benchmark report for SunSpider, V8, and Kraken.
VMs tested:
"TipOfTree" at /Volumes/Data/pizlo/quinary/OpenSource/WebKitBuild/Release/jsc
"ParGC_1worker" at /Volumes/Data/pizlo/senary/OpenSource/WebKitBuild/Release/jsc
Collected 12 samples per benchmark/VM, with 4 VM invocations per benchmark. Emitted a call to gc() between sample
measurements. Used 1 benchmark iteration per VM invocation for warm-up. Used the jsc-specific preciseTime()
function to get microsecond-level timing. Reporting benchmark execution times with 95% confidence intervals in
milliseconds.
TipOfTree ParGC_1worker
SunSpider:
3d-cube 7.2949+-0.1575 ? 7.3727+-0.1966 ? might be 1.0107x slower
3d-morph 7.6382+-0.1124 7.5984+-0.1009
3d-raytrace 7.4934+-0.1674 ? 7.5545+-0.2226 ?
access-binary-trees 1.5628+-0.0328 ! 1.6664+-0.0407 ! definitely 1.0662x slower
access-fannkuch 6.3769+-0.0957 6.3468+-0.1113
access-nbody 3.8433+-0.0674 3.8371+-0.0816
access-nsieve 2.6024+-0.0664 ? 2.6081+-0.0886 ?
bitops-3bit-bits-in-byte 1.3050+-0.0218 1.3031+-0.0404
bitops-bits-in-byte 2.3791+-0.0536 ? 2.4168+-0.0887 ? might be 1.0159x slower
bitops-bitwise-and 3.3935+-0.1110 3.3504+-0.0702 might be 1.0129x faster
bitops-nsieve-bits 5.2880+-0.0830 ? 5.3889+-0.0710 ? might be 1.0191x slower
controlflow-recursive 2.0956+-0.0343 ? 2.1096+-0.0380 ?
crypto-aes 7.4149+-0.1537 ? 7.5127+-0.1670 ? might be 1.0132x slower
crypto-md5 2.6900+-0.0664 ? 2.7824+-0.0428 ? might be 1.0343x slower
crypto-sha1 2.4659+-0.0728 2.4191+-0.0656 might be 1.0194x faster
date-format-tofte 9.9838+-0.3736 ? 10.0864+-0.1818 ? might be 1.0103x slower
date-format-xparb 9.2507+-0.2275 9.2164+-0.2176
math-cordic 6.4024+-0.1527 ? 6.5353+-0.1568 ? might be 1.0208x slower
math-partial-sums 7.5328+-0.1274 7.4789+-0.1229
math-spectral-norm 2.6021+-0.0688 2.5571+-0.0609 might be 1.0176x faster
regexp-dna 11.5334+-0.2701 11.4021+-0.1800 might be 1.0115x faster
string-base64 4.4235+-0.1934 4.4043+-0.1079
string-fasta 6.3310+-0.1557 ? 6.3362+-0.1281 ?
string-tagcloud 11.8947+-0.3594 11.6337+-0.2537 might be 1.0224x faster
string-unpack-code 20.4274+-0.3439 20.1890+-0.3354 might be 1.0118x faster
string-validate-input 5.3787+-0.1406 5.3504+-0.1414
<arithmetic> * 6.1386+-0.0371 6.1330+-0.0328
<geometric> 4.9505+-0.0321 ? 4.9654+-0.0321 ?
<harmonic> 3.9508+-0.0324 ? 3.9787+-0.0356 ?
TipOfTree ParGC_1worker
V8:
crypto 74.6573+-0.9355 ? 74.6671+-1.2695 ?
deltablue 176.7681+-1.9078 ? 178.8428+-1.3773 ? might be 1.0117x slower
earley-boyer 90.6198+-0.4111 ? 91.2420+-0.9776 ?
raytrace 63.4282+-0.7518 62.6713+-0.4769 might be 1.0121x faster
regexp 105.9695+-1.1261 ^ 104.3170+-0.4879 ^ definitely 1.0158x faster
richards 126.5331+-0.4297 ? 126.9162+-0.6510 ?
splay 91.4826+-0.6302 ! 95.4367+-0.4899 ! definitely 1.0432x slower
<arithmetic> 104.2084+-0.3281 ? 104.8705+-0.3554 ?
<geometric> * 98.9672+-0.2982 ? 99.4817+-0.3438 ?
<harmonic> 94.4207+-0.3735 ? 94.7775+-0.3760 ?
TipOfTree ParGC_1worker
Kraken:
ai-astar 497.9819+-1.5014 ? 501.5006+-4.3832 ?
audio-beat-detection 193.8519+-2.2030 191.8244+-1.9925 might be 1.0106x faster
audio-dft 267.2113+-3.2929 ? 271.1404+-4.5041 ? might be 1.0147x slower
audio-fft 125.1287+-1.0431 ? 125.7332+-0.9478 ?
audio-oscillator 250.9315+-1.1048 ? 253.3135+-1.5058 ?
imaging-darkroom 397.6901+-1.3490 ! 402.9572+-1.8083 ! definitely 1.0132x slower
imaging-desaturate 228.5683+-2.8265 226.6920+-1.4972
imaging-gaussian-blur 560.2089+-2.9211 556.4905+-1.8611
json-parse-financial 57.2933+-0.3284 ! 58.8656+-0.6922 ! definitely 1.0274x slower
json-stringify-tinderbox 70.1123+-1.4993 ^ 67.9328+-0.3743 ^ definitely 1.0321x faster
stanford-crypto-aes 132.2364+-1.5831 ? 133.8309+-1.5803 ? might be 1.0121x slower
stanford-crypto-ccm 100.2620+-0.6544 ? 100.6823+-1.2382 ?
stanford-crypto-pbkdf2 198.5007+-3.3325 195.2135+-2.0347 might be 1.0168x faster
stanford-crypto-sha256-iterative 70.9986+-0.3678 70.7538+-0.2947
<arithmetic> * 225.0697+-0.3483 ? 225.4951+-0.7427 ?
<geometric> 177.5370+-0.4508 ? 177.7344+-0.4956 ?
<harmonic> 140.0540+-0.5592 ? 140.1788+-0.4364 ?
TipOfTree ParGC_1worker
All benchmarks:
<arithmetic> 85.9583+-0.1007 ? 86.1804+-0.2358 ?
<geometric> 22.4631+-0.0890 ? 22.5254+-0.0871 ?
<harmonic> 6.9577+-0.0557 ? 7.0059+-0.0611 ?
TipOfTree ParGC_1worker
Geomean of preferred means:
<scaled-result> 51.5176+-0.1332 ? 51.6232+-0.1352 ?
--
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
More information about the webkit-unassigned
mailing list