[Webkit-unassigned] [Bug 68329] DFG should support continuous optimization
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Sun Sep 18 22:23:52 PDT 2011
https://bugs.webkit.org/show_bug.cgi?id=68329
--- Comment #2 from Filip Pizlo <fpizlo at apple.com> 2011-09-18 22:23:52 PST ---
Looks like adding a speculative success counter to the code that DFG emits is *almost* performance neutral. It's a slight loss, but hopefully I can make up for it by the increased robustness.
Benchmark report for SunSpider, V8, and Kraken.
VMs tested:
"TipOfTree" at /Volumes/Data/pizlo/quinary/OpenSource/WebKitBuild/Release/jsc
"JettisonCB" at /Volumes/Data/pizlo/octonary/OpenSource/WebKitBuild/Release/jsc
Collected 12 samples per benchmark/VM, with 4 VM invocations per benchmark. 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 JettisonCB
SunSpider:
3d-cube 7.7189+-0.1412 7.7112+-0.1051
3d-morph 7.4322+-0.1846 ? 7.4622+-0.1422 ?
3d-raytrace 7.6861+-0.0888 7.5094+-0.0946 might be 1.0235x faster
access-binary-trees 2.3022+-0.0572 ? 2.3973+-0.0883 ? might be 1.0413x slower
access-fannkuch 11.5342+-0.1283 ? 11.5562+-0.1893 ?
access-nbody 4.1764+-0.1071 ^ 3.8803+-0.0972 ^ definitely 1.0763x faster
access-nsieve 2.5951+-0.0377 ? 2.6183+-0.0797 ?
bitops-3bit-bits-in-byte 1.6526+-0.0440 ! 1.7839+-0.0379 ! definitely 1.0794x slower
bitops-bits-in-byte 2.7086+-0.0479 2.6894+-0.0520
bitops-bitwise-and 3.6113+-0.0852 ? 3.6227+-0.0851 ?
bitops-nsieve-bits 5.3445+-0.0918 ? 5.3558+-0.1016 ?
controlflow-recursive 1.9275+-0.0509 ! 2.1499+-0.0869 ! definitely 1.1154x slower
crypto-aes 6.9565+-0.2272 6.8406+-0.3221 might be 1.0169x faster
crypto-md5 2.7340+-0.0502 ? 2.7975+-0.0356 ? might be 1.0232x slower
crypto-sha1 2.2154+-0.0499 ? 2.3144+-0.0540 ? might be 1.0447x slower
date-format-tofte 9.7958+-0.1211 ? 9.9403+-0.1487 ? might be 1.0147x slower
date-format-xparb 8.7927+-0.1389 ? 8.8095+-0.1998 ?
math-cordic 6.2006+-0.1075 6.1583+-0.0908
math-partial-sums 7.2989+-0.1076 7.2459+-0.0633
math-spectral-norm 2.6274+-0.0671 2.5728+-0.0299 might be 1.0212x faster
regexp-dna 10.9608+-0.1526 10.7005+-0.1278 might be 1.0243x faster
string-base64 5.7498+-0.1633 5.7256+-0.1321
string-fasta 6.9099+-0.1548 ? 7.0445+-0.1264 ? might be 1.0195x slower
string-tagcloud 11.5898+-0.1934 ? 11.8337+-0.1978 ? might be 1.0210x slower
string-unpack-code 18.4722+-0.1787 ? 18.8110+-0.5315 ? might be 1.0183x slower
string-validate-input 6.4533+-0.1169 ? 6.4611+-0.1352 ?
<arithmetic> 6.3633+-0.0157 ? 6.3843+-0.0220 ?
<geometric> 5.2546+-0.0131 ! 5.2953+-0.0155 ! definitely 1.0077x slower
<harmonic> 4.2874+-0.0205 ! 4.3699+-0.0179 ! definitely 1.0193x slower
TipOfTree JettisonCB
V8:
crypto 82.8718+-0.5337 82.4999+-0.4342
deltablue 241.8472+-2.3797 ! 245.6910+-1.1746 ! definitely 1.0159x slower
earley-boyer 98.3761+-2.6772 96.7450+-0.2958 might be 1.0169x faster
raytrace 69.0379+-0.2578 ! 70.1706+-0.4662 ! definitely 1.0164x slower
regexp 106.5903+-1.2884 105.7796+-0.9258
richards 217.2081+-0.8096 ! 225.6307+-1.1982 ! definitely 1.0388x slower
splay 98.6183+-0.5517 98.4457+-0.6877
<arithmetic> 130.6500+-0.3123 ! 132.1375+-0.3785 ! definitely 1.0114x slower
<geometric> 117.5667+-0.3564 ? 118.2467+-0.3255 ?
<harmonic> 107.6964+-0.4021 ? 107.9803+-0.3229 ?
TipOfTree JettisonCB
Kraken:
ai-astar 630.8458+-4.8120 628.8718+-4.5694
audio-beat-detection 465.4545+-2.0257 ! 473.6872+-0.9470 ! definitely 1.0177x slower
audio-dft 424.4798+-5.0051 418.3906+-2.0215 might be 1.0146x faster
audio-fft 362.5508+-1.6881 ? 367.1407+-2.9210 ? might be 1.0127x slower
audio-oscillator 315.8901+-0.7736 ^ 312.8390+-1.2768 ^ definitely 1.0098x faster
imaging-darkroom 413.5906+-2.0065 412.9257+-1.1696
imaging-desaturate 217.8714+-0.7518 212.2578+-5.0701 might be 1.0264x faster
imaging-gaussian-blur 591.4198+-1.2417 ! 598.7139+-2.5048 ! definitely 1.0123x slower
json-parse-financial 49.7857+-0.5397 49.6653+-0.6754
json-stringify-tinderbox 67.6122+-0.3064 67.4863+-0.2974
stanford-crypto-aes 144.6965+-1.1703 ^ 141.7946+-0.6062 ^ definitely 1.0205x faster
stanford-crypto-ccm 111.3790+-0.7516 ? 111.6174+-0.6448 ?
stanford-crypto-pbkdf2 393.2011+-2.0680 391.7593+-2.1244
stanford-crypto-sha256-iterative 147.4451+-0.6313 ? 148.8170+-0.9358 ?
<arithmetic> 309.7302+-0.4355 309.7119+-0.6984
<geometric> 242.2064+-0.3787 241.6892+-0.5748
<harmonic> 173.2442+-0.5841 172.6992+-0.6739
TipOfTree JettisonCB
All benchmarks:
<arithmetic> 115.2387+-0.1474 ? 115.4664+-0.2303 ?
<geometric> 26.1288+-0.0404 ! 26.2463+-0.0382 ! definitely 1.0045x slower
<harmonic> 7.5683+-0.0355 ! 7.7104+-0.0309 ! definitely 1.0188x slower
--
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