[Webkit-unassigned] [Bug 74908] Temporary GPR should not be lazily allocated in DFG JIT on X86
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Mon Dec 19 22:18:22 PST 2011
https://bugs.webkit.org/show_bug.cgi?id=74908
--- Comment #3 from Yuqiang Xian <yuqiang.xian at intel.com> 2011-12-19 22:18:22 PST ---
Seems some cases (especially, Kraken) are impacted by this change, on Linux IA32 (Core i7 Nehalem), possibly caused by some instruction ordering changes, which may need deeper investigation.
The performance result:
Benchmark report for SunSpider, V8, and Kraken on \c.
VMs tested:
"ToT" at /home/yxian/WebKit_orig/WebKitBuild/Release/Source/JavaScriptCore/shell/jsc_efl
"74908" at /mnt/supplement/WebKit/WebKitBuild/Release_efl/Source/JavaScriptCore/shell/jsc_efl
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.
ToT 74908
SunSpider:
3d-cube 8.1819+-1.1747 7.4230+-0.0596 might be 1.1022x faster
3d-morph 10.3083+-0.0440 ? 10.3085+-0.0402 ?
3d-raytrace 9.5854+-0.0422 ? 9.6253+-0.0472 ?
access-binary-trees 2.0577+-0.4597 1.8671+-0.0448 might be 1.1021x faster
access-fannkuch 10.5143+-0.0494 ? 10.5700+-0.0385 ?
access-nbody 5.4011+-0.0447 5.3716+-0.0419
access-nsieve 3.7123+-0.0302 3.6787+-0.0179
bitops-3bit-bits-in-byte 1.1956+-0.0357 1.1731+-0.0326 might be 1.0192x faster
bitops-bits-in-byte 4.6542+-0.0578 4.5883+-0.0299 might be 1.0144x faster
bitops-bitwise-and 4.2350+-0.0416 4.2235+-0.0216
bitops-nsieve-bits 6.7435+-0.1197 6.7103+-0.0849
controlflow-recursive 2.8677+-0.0536 2.8480+-0.0474
crypto-aes 10.5288+-0.0560 ! 10.7119+-0.0832 ! definitely 1.0174x slower
crypto-md5 3.6444+-0.0481 3.6160+-0.0533
crypto-sha1 3.0002+-0.0659 2.9547+-0.0457 might be 1.0154x faster
date-format-tofte 11.7048+-0.0871 ? 12.4378+-1.5467 ? might be 1.0626x slower
date-format-xparb 11.1803+-0.0719 ? 11.1967+-0.0848 ?
math-cordic 9.1706+-0.0320 ? 9.1716+-0.0408 ?
math-partial-sums 14.5324+-1.4646 13.8742+-0.0552 might be 1.0474x faster
math-spectral-norm 2.5975+-0.0438 2.5914+-0.0397
regexp-dna 9.1554+-0.0716 9.1394+-0.0621
string-base64 5.3683+-0.0487 ? 5.3807+-0.0720 ?
string-fasta 9.6004+-0.0515 9.5828+-0.0477
string-tagcloud 15.6370+-0.0537 ? 15.6695+-0.0588 ?
string-unpack-code 24.9733+-0.2835 ^ 24.5256+-0.0843 ^ definitely 1.0183x faster
string-validate-input 7.1826+-0.0475 ? 7.1969+-0.0606 ?
<arithmetic> * 7.9897+-0.0725 7.9399+-0.0647 might be 1.0063x faster
<geometric> 6.4304+-0.0505 6.3797+-0.0265 might be 1.0079x faster
<harmonic> 4.9402+-0.0567 4.8836+-0.0293 might be 1.0116x faster
ToT 74908
V8:
crypto 97.0162+-0.4236 ? 97.1926+-0.3836 ?
deltablue 166.9306+-0.6094 ? 168.1061+-0.6851 ?
earley-boyer 107.9100+-0.2276 ? 108.3725+-0.6809 ?
raytrace 52.6538+-0.4061 52.5592+-0.4340
regexp 128.7707+-0.7285 128.2945+-0.2053
richards 173.4722+-0.6191 ! 176.2666+-1.7734 ! definitely 1.0161x slower
splay 126.8781+-0.8428 ? 127.1935+-1.4026 ?
<arithmetic> 121.9474+-0.2155 ! 122.5693+-0.2931 ! definitely 1.0051x slower
<geometric> * 114.6731+-0.2402 ? 115.0977+-0.2633 ? might be 1.0037x slower
<harmonic> 106.1049+-0.3105 ? 106.3385+-0.2966 ? might be 1.0022x slower
ToT 74908
Kraken:
ai-astar 785.4775+-1.8174 784.4535+-5.3564
audio-beat-detection 352.6068+-2.4209 ! 362.4537+-1.0279 ! definitely 1.0279x slower
audio-dft 344.7694+-1.9229 ! 371.6174+-3.1400 ! definitely 1.0779x slower
audio-fft 215.2775+-0.1940 ! 224.9574+-0.7567 ! definitely 1.0450x slower
audio-oscillator 340.3142+-4.0461 ? 343.4859+-2.6913 ?
imaging-darkroom 386.6637+-10.3607 ? 387.8107+-10.2604 ?
imaging-desaturate 316.4435+-0.3136 ^ 289.5850+-0.1891 ^ definitely 1.0927x faster
imaging-gaussian-blur 628.6579+-1.5163 ^ 626.8492+-0.2748 ^ definitely 1.0029x faster
json-parse-financial 68.8907+-0.6046 ? 69.0699+-0.5954 ?
json-stringify-tinderbox 99.8946+-0.3831 ! 107.0620+-2.1414 ! definitely 1.0717x slower
stanford-crypto-aes 132.6336+-0.5289 ? 134.1565+-1.3846 ? might be 1.0115x slower
stanford-crypto-ccm 128.3499+-0.9291 ? 128.4467+-0.4705 ?
stanford-crypto-pbkdf2 282.3136+-2.1761 282.2923+-0.8014
stanford-crypto-sha256-iterative 108.4356+-0.2734 ! 109.9840+-0.5452 ! definitely 1.0143x slower
<arithmetic> * 299.3377+-0.9721 ! 301.5874+-1.0977 ! definitely 1.0075x slower
<geometric> 238.5068+-0.7272 ! 241.3118+-0.8654 ! definitely 1.0118x slower
<harmonic> 187.7669+-0.5586 ! 190.6914+-0.7903 ! definitely 1.0156x slower
ToT 74908
All benchmarks:
<arithmetic> 111.7467+-0.3133 ! 112.4818+-0.3473 ! definitely 1.0066x slower
<geometric> 28.9750+-0.1359 28.9653+-0.0887 might be 1.0003x faster
<harmonic> 8.6980+-0.0973 8.6030+-0.0503 might be 1.0110x faster
ToT 74908
Geomean of preferred means:
<scaled-result> 64.9695+-0.2435 ? 65.0765+-0.2367 ? might be 1.0016x 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