REOPENED 156126
[JSC] Initialize SSA's live values at tail lazily 
https://bugs.webkit.org/show_bug.cgi?id=156126
Summary [JSC] Initialize SSA's live values at tail lazily
Benjamin Poulain
Reported 2016-04-01 20:15:25 PDT
[JSC] Initialize SSA's live values at tail lazily
Attachments
Patch (3.15 KB, patch)
2016-04-01 20:21 PDT, Benjamin Poulain 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Benjamin Poulain
Comment 1 2016-04-01 20:21:11 PDT
Created attachment 275458 [details] Patch
Benjamin Poulain
Comment 2 2016-04-01 20:21:27 PDT
Conf#1 Conf#2 SunSpider: 3d-cube 4.9843+-0.0989 4.8228+-0.0860 might be 1.0335x faster 3d-morph 5.2516+-0.0529 5.2458+-0.0405 3d-raytrace 5.6559+-0.0602 ? 5.6577+-0.0714 ? access-binary-trees 2.1776+-0.0390 2.1644+-0.0467 access-fannkuch 6.0087+-0.1125 5.9731+-0.0698 access-nbody 2.5784+-0.0412 2.5704+-0.0401 access-nsieve 3.2382+-0.0472 ? 3.2700+-0.0455 ? bitops-3bit-bits-in-byte 1.1550+-0.0165 1.1506+-0.0173 bitops-bits-in-byte 2.7831+-0.0301 ? 2.7858+-0.0165 ? bitops-bitwise-and 2.0909+-0.0651 ? 2.1227+-0.0546 ? might be 1.0152x slower bitops-nsieve-bits 3.1348+-0.0582 ? 3.1656+-0.0538 ? controlflow-recursive 2.3729+-0.0364 2.3697+-0.0395 crypto-aes 4.0952+-0.0824 4.0890+-0.0548 crypto-md5 2.5103+-0.0377 2.5018+-0.0358 crypto-sha1 2.3167+-0.0245 ? 2.3370+-0.0369 ? date-format-tofte 6.4573+-0.0751 ? 6.5397+-0.0460 ? might be 1.0128x slower date-format-xparb 4.9208+-0.0881 ? 5.0084+-0.1151 ? might be 1.0178x slower math-cordic 2.8579+-0.0303 2.8406+-0.0208 math-partial-sums 4.8531+-0.0779 ? 4.8812+-0.0663 ? math-spectral-norm 2.0302+-0.0161 ? 2.0481+-0.0342 ? regexp-dna 6.3161+-0.1355 6.3114+-0.1183 string-base64 4.4776+-0.0821 ? 4.5047+-0.0941 ? string-fasta 5.9144+-0.0399 ? 6.0104+-0.1151 ? might be 1.0162x slower string-tagcloud 8.1667+-0.0703 ? 8.1715+-0.0397 ? string-unpack-code 19.2938+-0.3697 19.2308+-0.3947 string-validate-input 4.3740+-0.0298 ? 4.4175+-0.0338 ? <arithmetic> 4.6160+-0.0185 ? 4.6227+-0.0216 ? might be 1.0015x slower Conf#1 Conf#2 Octane: encrypt 0.16431+-0.00084 ? 0.16481+-0.00093 ? decrypt 2.84765+-0.00555 2.84736+-0.00436 deltablue x2 0.14187+-0.00207 0.13995+-0.00154 might be 1.0137x faster earley 0.28669+-0.00104 0.28559+-0.00086 boyer 5.06090+-0.01125 5.01881+-0.04993 navier-stokes x2 5.00413+-0.01770 5.00265+-0.01709 raytrace x2 0.90057+-0.00276 0.89915+-0.00193 richards x2 0.08212+-0.00028 ! 0.08299+-0.00054 ! definitely 1.0106x slower splay x2 0.35457+-0.00193 0.35295+-0.00186 regexp x2 18.89440+-0.08488 ? 18.94250+-0.07855 ? pdfjs x2 39.29117+-0.27624 39.23886+-0.31162 mandreel x2 42.62469+-0.10164 ? 42.71183+-0.10411 ? gbemu x2 24.08632+-0.05836 24.06376+-0.07484 closure 0.53854+-0.00102 ? 0.53937+-0.00112 ? jquery 6.83369+-0.02125 6.82240+-0.02126 box2d x2 9.23317+-0.03794 ? 9.23962+-0.02871 ? zlib x2 361.63952+-2.46268 ? 362.88772+-2.89056 ? typescript x2 638.23696+-2.81661 637.70034+-3.10117 <geometric> 5.15144+-0.00750 5.14849+-0.00682 might be 1.0006x faster Conf#1 Conf#2 Kraken: ai-astar 89.627+-1.339 88.192+-0.705 might be 1.0163x faster audio-beat-detection 42.220+-0.045 ! 42.348+-0.081 ! definitely 1.0030x slower audio-dft 99.650+-0.929 98.802+-0.630 audio-fft 32.812+-0.056 ? 32.830+-0.041 ? audio-oscillator 48.572+-0.224 48.525+-0.134 imaging-darkroom 60.223+-0.047 ? 60.354+-0.358 ? imaging-desaturate 45.380+-0.344 45.272+-0.121 imaging-gaussian-blur 62.366+-0.508 62.326+-0.818 json-parse-financial 38.094+-0.182 ^ 37.386+-0.058 ^ definitely 1.0189x faster json-stringify-tinderbox 24.562+-0.339 ^ 23.067+-0.551 ^ definitely 1.0648x faster stanford-crypto-aes 39.445+-0.095 ? 39.814+-0.493 ? stanford-crypto-ccm 34.991+-0.704 ? 35.738+-0.851 ? might be 1.0214x slower stanford-crypto-pbkdf2 100.122+-0.421 ? 100.201+-0.288 ? stanford-crypto-sha256-iterative 39.073+-0.385 38.966+-0.201 <arithmetic> 54.081+-0.122 53.844+-0.145 might be 1.0044x faster Conf#1 Conf#2 AsmBench: bigfib.cpp 443.6288+-2.6904 442.8142+-3.0500 cray.c 356.0264+-0.9260 355.9545+-1.0191 dry.c 444.0648+-20.5465 ? 449.9112+-21.0750 ? might be 1.0132x slower FloatMM.c 733.8145+-3.0473 730.8134+-1.3369 gcc-loops.cpp 3717.9139+-9.2703 3710.3877+-4.1137 n-body.c 810.5781+-2.1601 ? 813.0082+-4.4421 ? Quicksort.c 397.1493+-0.7200 ? 398.3641+-1.2945 ? stepanov_container.cpp 3315.8753+-10.5031 ? 3316.9209+-8.6043 ? Towers.c 273.7492+-1.0351 ^ 272.1418+-0.4725 ^ definitely 1.0059x faster <geometric> 725.4884+-3.2349 ? 725.9056+-3.4389 ? might be 1.0006x slower Conf#1 Conf#2 Geomean of preferred means: <scaled-result> 31.0784+-0.0461 31.0555+-0.0514 might be 1.0007x faster
Mark Lam
Comment 3 2016-04-01 22:13:41 PDT
Comment on attachment 275458 [details] Patch r=me
WebKit Commit Bot
Comment 4 2016-04-02 14:15:57 PDT
Comment on attachment 275458 [details] Patch Clearing flags on attachment: 275458 Committed r198976: <http://trac.webkit.org/changeset/198976>
WebKit Commit Bot
Comment 5 2016-04-02 14:16:01 PDT
All reviewed patches have been landed. Closing bug.
David Kilzer (:ddkilzer)
Comment 6 2016-04-02 18:51:31 PDT
This caused js/regress/array-nonarray-polymorhpic-access.html to crash on most of the bots: <http://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=js%2Fregress%2Farray-nonarray-polymorhpic-access.html> Prepping rollout via #webkit/webkitbot.
WebKit Commit Bot
Comment 7 2016-04-02 18:51:56 PDT
Re-opened since this is blocked by bug 156140
Note You need to log in before you can comment on or make changes to this bug.