Full Text Bug Listing

Geoffrey Garen

Reported 2013-03-07 16:58:29 PST

REGRESSION (r143759): 40% JSBench regression, 20% Octane/closure regression, 40% Octane/jquery regression, 2% Octane regression

Geoffrey Garen

Comment 1 2013-03-07 16:58:57 PST

Seen on the bot, which runs: run-jsc-benchmarks --outer 2 --inner 2

Geoffrey Garen

Comment 2 2013-03-07 16:59:47 PST

In this configuration, the cache's starting guess of 1MB dominates program execution, before any of the cache's smarts kick in.

Geoffrey Garen

Comment 3 2013-03-07 17:04:59 PST

Created attachment 192115 [details] Patch

Geoffrey Garen

Comment 4 2013-03-07 17:06:13 PST

Tested vs r143758, r143759+this patch is no performance regression: Benchmark report for SunSpider, V8Spider, Octane, Kraken, JSBench, JSRegress, and DSP on garen (MacPro5,1). VMs tested: "BASE" at /Volumes/Big/ggaren/baseline-webkit/WebKitBuild/Release/DumpRenderTree (r143758) "PATCH" at /Volumes/Big/ggaren/webkit/WebKitBuild/Release/DumpRenderTree (r143949) Collected 4 samples per benchmark/VM, with 2 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. BASE PATCH SunSpider: 3d-cube 10.8445+-0.7846 10.8033+-0.9457 3d-morph 9.2441+-0.1685 ? 9.3491+-0.1444 ? might be 1.0113x slower 3d-raytrace 11.7272+-0.8772 11.6065+-1.0136 might be 1.0104x faster access-binary-trees 2.4954+-0.9608 ? 2.5096+-0.9218 ? access-fannkuch 7.5344+-0.0534 ? 7.5380+-0.0824 ? access-nbody 4.7698+-0.1879 4.7188+-0.1338 might be 1.0108x faster access-nsieve 4.8518+-0.0417 4.8140+-0.1543 bitops-3bit-bits-in-byte 1.7240+-0.0130 ? 1.7407+-0.0738 ? bitops-bits-in-byte 6.4092+-0.0801 ? 6.4232+-0.0434 ? bitops-bitwise-and 2.2124+-0.0088 ? 2.3375+-0.3203 ? might be 1.0565x slower bitops-nsieve-bits 4.0410+-0.0372 4.0195+-0.0206 controlflow-recursive 2.8617+-0.0298 ? 2.9075+-0.0377 ? might be 1.0160x slower crypto-aes 8.7312+-1.1281 ? 8.8525+-0.8099 ? might be 1.0139x slower crypto-md5 4.1807+-0.3279 ? 4.1904+-0.0810 ? crypto-sha1 3.5004+-0.0847 ? 3.5198+-0.1011 ? date-format-tofte 17.6857+-3.1110 ? 17.7252+-2.9684 ? date-format-xparb 16.7502+-2.1209 ? 16.8943+-2.2214 ? math-cordic 3.7443+-0.0577 3.7332+-0.0387 math-partial-sums 11.9683+-0.0915 ? 12.0475+-0.0688 ? math-spectral-norm 3.1191+-0.0279 ? 3.1193+-0.0128 ? regexp-dna 10.7065+-1.4032 ? 10.7498+-1.6171 ? string-base64 6.5704+-1.8249 6.4070+-1.5885 might be 1.0255x faster string-fasta 11.1532+-0.5612 11.0920+-0.5646 string-tagcloud 15.0109+-0.4949 ? 15.0544+-0.3297 ? string-unpack-code 28.1624+-1.2837 ? 28.3555+-1.1793 ? string-validate-input 9.9772+-0.3993 ? 10.0818+-0.4584 ? might be 1.0105x slower <arithmetic> * 8.4606+-0.3793 ? 8.4842+-0.4918 ? might be 1.0028x slower <geometric> 6.6440+-0.1988 ? 6.6702+-0.3672 ? might be 1.0039x slower <harmonic> 5.1917+-0.1548 ? 5.2317+-0.3022 ? might be 1.0077x slower BASE PATCH V8Spider: crypto 83.2711+-0.1185 ? 83.6557+-1.1847 ? deltablue 130.6743+-1.1040 130.4390+-0.8337 earley-boyer 84.7060+-1.5650 ? 84.9540+-2.0482 ? raytrace 64.6742+-0.7971 ? 64.8422+-0.7115 ? regexp 99.7057+-0.6723 99.4838+-1.8482 richards 113.2200+-1.4973 112.7955+-0.5902 splay 68.0616+-8.2055 ? 69.2809+-8.0491 ? might be 1.0179x slower <arithmetic> 92.0447+-1.5513 ? 92.2073+-1.6689 ? might be 1.0018x slower <geometric> * 89.4217+-1.9567 ? 89.6802+-2.0215 ? might be 1.0029x slower <harmonic> 86.9216+-2.3685 ? 87.2713+-2.3732 ? might be 1.0040x slower BASE PATCH Octane and V8v7: encrypt 0.44408+-0.00092 ? 0.44435+-0.00098 ? decrypt 7.93859+-0.08746 7.89670+-0.05854 deltablue x2 0.59119+-0.02061 ? 0.59236+-0.01795 ? earley 0.86965+-0.02265 0.86194+-0.01804 boyer 12.55236+-0.13026 12.47342+-0.07120 raytrace x2 4.79423+-1.06122 ? 4.88771+-1.14686 ? might be 1.0195x slower regexp x2 30.25632+-0.53173 ? 30.38146+-1.05853 ? richards x2 0.28592+-0.00116 0.28522+-0.00079 splay x2 0.69100+-0.05211 0.68795+-0.01002 navier-stokes x2 11.54579+-0.07162 11.53851+-0.09948 closure 0.36877+-0.01537 0.31873+-0.08159 might be 1.1570x faster jquery 4.90589+-0.84145 4.75864+-1.10108 might be 1.0309x faster gbemu x2 257.86023+-7.74264 255.38346+-7.76150 mandreel x2 173.86414+-0.98129 173.86199+-0.65114 pdfjs x2 126.44341+-0.77217 ? 126.47745+-0.40573 ? box2d x2 33.79110+-0.50082 ? 33.93483+-0.48910 ? V8v7: <arithmetic> 7.38335+-0.20504 ? 7.40143+-0.17180 ? might be 1.0024x slower <geometric> * 2.42741+-0.07452 ? 2.42974+-0.06818 ? might be 1.0010x slower <harmonic> 0.91187+-0.01510 0.91072+-0.00498 might be 1.0013x faster Octane including V8v7: <arithmetic> 50.28177+-0.51132 50.10829+-0.52922 might be 1.0035x faster <geometric> * 7.65436+-0.12320 7.60030+-0.10402 might be 1.0071x faster <harmonic> 1.26463+-0.01875 1.23414+-0.04579 might be 1.0247x faster BASE PATCH Kraken: ai-astar 502.670+-2.658 501.996+-1.897 audio-beat-detection 238.697+-1.598 ? 244.784+-8.832 ? might be 1.0255x slower audio-dft 300.669+-0.783 298.920+-1.627 audio-fft 139.287+-0.227 139.242+-0.060 audio-oscillator 239.892+-0.754 239.858+-1.058 imaging-darkroom 286.448+-1.849 ? 293.626+-5.750 ? might be 1.0251x slower imaging-desaturate 149.738+-0.544 149.587+-0.394 imaging-gaussian-blur 409.483+-0.650 ? 409.964+-0.764 ? json-parse-financial 76.690+-0.298 ? 77.636+-0.723 ? might be 1.0123x slower json-stringify-tinderbox 97.542+-0.335 ^ 96.687+-0.272 ^ definitely 1.0088x faster stanford-crypto-aes 121.043+-0.905 119.419+-0.784 might be 1.0136x faster stanford-crypto-ccm 138.379+-0.500 ! 139.402+-0.437 ! definitely 1.0074x slower stanford-crypto-pbkdf2 264.232+-2.113 262.253+-2.080 stanford-crypto-sha256-iterative 121.310+-0.742 ? 121.636+-1.425 ? <arithmetic> * 220.434+-0.409 ? 221.072+-0.803 ? might be 1.0029x slower <geometric> 190.846+-0.279 ? 191.319+-0.633 ? might be 1.0025x slower <harmonic> 166.129+-0.267 ? 166.446+-0.509 ? might be 1.0019x slower BASE PATCH JSBench: amazon 10.7500+-0.7956 ? 11.0000+-0.0000 ? might be 1.0233x slower facebook 41.5000+-2.7561 41.5000+-2.7561 google 89.0000+-1.8374 88.5000+-2.7561 twitter 10.5000+-0.9187 10.2500+-0.7956 might be 1.0244x faster yahoo 3.2500+-0.7956 ? 3.5000+-0.9187 ? might be 1.0769x slower <arithmetic> * 31.0000+-1.1326 30.9500+-1.2013 might be 1.0016x faster <geometric> 16.8152+-1.0642 ? 17.0363+-0.8991 ? might be 1.0131x slower <harmonic> 9.3654+-1.2465 ? 9.7523+-1.3733 ? might be 1.0413x slower BASE PATCH JSRegress: adapt-to-double-divide 21.3252+-0.0940 21.3044+-0.0361 aliased-arguments-getbyval 1.0646+-0.0209 1.0378+-0.0238 might be 1.0259x faster allocate-big-object 3.5673+-3.5418 ? 4.6005+-3.9606 ? might be 1.2896x slower arity-mismatch-inlining 0.7907+-0.0134 ? 0.7963+-0.0366 ? array-access-polymorphic-structure 9.9472+-6.0593 ? 9.9867+-6.0416 ? array-with-double-add 5.2810+-0.3585 5.2633+-0.4279 array-with-double-increment 3.8212+-0.0423 3.7784+-0.0571 might be 1.0113x faster array-with-double-mul-add 7.1237+-0.7616 ? 7.1602+-0.7909 ? array-with-double-sum 7.4321+-0.1943 7.3661+-0.0133 array-with-int32-add-sub 9.3740+-1.0129 9.3699+-0.9902 array-with-int32-or-double-sum 7.5025+-0.0903 7.4628+-0.0113 big-int-mul 4.6415+-0.0966 4.6177+-0.0556 boolean-test 4.0421+-0.0470 4.0408+-0.0361 cast-int-to-double 13.5284+-0.0299 13.5047+-0.0150 cell-argument 13.6459+-0.0447 13.6141+-0.0132 cfg-simplify 3.5595+-0.0212 ? 3.5778+-0.0759 ? cmpeq-obj-to-obj-other 11.4730+-1.1690 ? 11.9147+-0.2717 ? might be 1.0385x slower constant-test 8.0853+-0.0931 8.0473+-0.2754 direct-arguments-getbyval 0.9675+-0.0602 0.9573+-0.0338 might be 1.0106x faster double-pollution-getbyval 10.1666+-0.0619 10.1593+-0.0192 double-pollution-putbyoffset 5.7504+-1.8409 ? 5.7628+-1.9206 ? external-arguments-getbyval 2.6129+-0.5319 2.6120+-0.5266 external-arguments-putbyval 3.8735+-0.7932 3.8421+-0.8527 Float32Array-matrix-mult 14.3423+-1.1474 ? 14.6804+-1.4607 ? might be 1.0236x slower fold-double-to-int 20.9918+-0.8626 20.8351+-0.4499 function-dot-apply 2.9362+-0.1133 ? 2.9719+-0.0213 ? might be 1.0122x slower function-test 4.7047+-0.1645 4.6480+-0.1301 might be 1.0122x faster get-by-id-chain-from-try-block 7.1092+-0.1136 7.0693+-0.0321 HashMap-put-get-iterate-keys 86.5285+-1.0078 84.2521+-1.6288 might be 1.0270x faster HashMap-put-get-iterate 87.5262+-2.5336 ? 87.6030+-2.8559 ? HashMap-string-put-get-iterate 78.8381+-2.0048 ? 78.8828+-2.0635 ? indexed-properties-in-objects 4.1038+-0.0188 ? 4.1235+-0.0663 ? inline-arguments-access 1.3083+-0.0171 ? 1.3184+-0.0350 ? inline-arguments-local-escape 24.2919+-0.0986 ? 24.4677+-0.0854 ? inline-get-scoped-var 6.1459+-0.0062 ? 6.1943+-0.1248 ? inlined-put-by-id-transition 14.2435+-0.2155 ? 14.4020+-0.2386 ? might be 1.0111x slower int-or-other-abs-then-get-by-val 8.0425+-0.0306 8.0302+-0.0199 int-or-other-abs-zero-then-get-by-val 35.8246+-0.7639 35.3892+-0.3974 might be 1.0123x faster int-or-other-add-then-get-by-val 10.0778+-0.0253 10.0604+-0.0092 int-or-other-add 10.2256+-0.2170 10.1357+-0.0193 int-or-other-div-then-get-by-val 7.3948+-0.0205 ? 7.3955+-0.0068 ? int-or-other-max-then-get-by-val 9.3279+-0.5347 ? 9.4453+-0.5487 ? might be 1.0126x slower int-or-other-min-then-get-by-val 7.6627+-0.0234 7.6189+-0.0310 int-or-other-mod-then-get-by-val 7.3554+-0.0268 7.3547+-0.0421 int-or-other-mul-then-get-by-val 6.6501+-0.0152 6.6296+-0.0130 int-or-other-neg-then-get-by-val 7.9337+-0.2913 7.9233+-0.2668 int-or-other-neg-zero-then-get-by-val 34.7742+-0.1800 ? 35.0142+-0.4512 ? int-or-other-sub-then-get-by-val 10.0842+-0.0407 10.0565+-0.0036 int-or-other-sub 8.1322+-0.0147 ? 8.3122+-0.3412 ? might be 1.0221x slower int-overflow-local 16.1440+-0.6392 15.8088+-0.0405 might be 1.0212x faster Int16Array-bubble-sort 74.5720+-7.3809 72.5685+-0.8320 might be 1.0276x faster Int16Array-load-int-mul 1.7955+-0.0383 1.7850+-0.0086 Int8Array-load 5.1475+-0.0475 5.1412+-0.0625 integer-divide 14.0141+-0.0611 14.0137+-0.0662 integer-modulo 2.1102+-0.0654 ? 2.1296+-0.1021 ? make-indexed-storage 4.7345+-1.9224 4.7283+-1.9563 method-on-number 22.1796+-0.8321 ? 22.5731+-1.6184 ? might be 1.0177x slower nested-function-parsing 375.9600+-19.9780 ? 384.5302+-28.2400 ? might be 1.0228x slower new-array-buffer-dead 3.5644+-0.4021 ? 3.6366+-0.3225 ? might be 1.0203x slower new-array-buffer-push 14.3543+-7.7096 ? 14.5480+-8.0693 ? might be 1.0135x slower new-array-dead 26.8387+-0.0145 ? 26.9267+-0.2253 ? new-array-push 9.5856+-0.5040 ? 9.6016+-0.7151 ? number-test 4.0155+-0.0843 4.0056+-0.0865 object-closure-call 8.3179+-0.4477 8.2891+-0.6222 object-test 4.5630+-0.0333 4.5261+-0.0813 poly-stricteq 86.6387+-0.5544 ? 87.7812+-3.4300 ? might be 1.0132x slower polymorphic-structure 19.0568+-0.0242 ? 19.0883+-0.0595 ? polyvariant-monomorphic-get-by-id 11.8458+-0.0417 ? 11.8679+-0.1097 ? rare-osr-exit-on-local 19.5688+-0.3128 19.4901+-0.0836 register-pressure-from-osr 30.2239+-0.2969 30.0742+-0.0206 simple-activation-demo 33.4773+-0.4188 ? 33.6417+-1.3324 ? slow-array-profile-convergence 4.5443+-0.7483 4.5164+-0.7040 slow-convergence 3.7814+-0.0202 3.7690+-0.0075 sparse-conditional 1.2935+-0.0098 1.2775+-0.0380 might be 1.0125x faster splice-to-remove 47.7029+-0.2172 ? 47.9695+-0.0679 ? string-hash 2.5328+-0.0310 2.5207+-0.0092 string-repeat-arith 42.1205+-0.5051 41.6890+-0.3246 might be 1.0104x faster string-sub 81.1405+-0.9768 ? 83.3143+-6.5093 ? might be 1.0268x slower string-test 3.9695+-0.0523 ? 3.9813+-0.0882 ? structure-hoist-over-transitions 4.1459+-1.8641 4.1345+-1.8902 tear-off-arguments-simple 2.0060+-0.0253 2.0049+-0.0099 tear-off-arguments 3.4210+-0.0305 ? 3.4270+-0.0153 ? temporal-structure 19.9715+-0.4221 19.8500+-0.0515 to-int32-boolean 26.9224+-0.1199 ? 26.9274+-0.0376 ? undefined-test 4.1865+-0.0296 ? 4.1982+-0.0528 ? <arithmetic> 20.1480+-0.3979 ? 20.2580+-0.6098 ? might be 1.0055x slower <geometric> * 9.0410+-0.2951 ? 9.0697+-0.3690 ? might be 1.0032x slower <harmonic> 5.1275+-0.1346 ? 5.1383+-0.1480 ? might be 1.0021x slower BASE PATCH DSP: filtrr-posterize-tint 50.9678+-3.7282 50.8928+-3.4149 filtrr-tint-contrast-sat-bright 71.6526+-1.8568 ? 71.8545+-1.0988 ? filtrr-tint-sat-adj-contr-mult 86.7718+-7.5529 85.7352+-6.0466 might be 1.0121x faster filtrr-blur-overlay-sat-contr 227.6051+-4.7696 ? 230.4676+-11.7565 ? might be 1.0126x slower filtrr-sat-blur-mult-sharpen-contr 259.4548+-17.7710 ? 261.1887+-15.3500 ? filtrr-sepia-bias 34.7023+-1.9761 34.5577+-2.0879 route9-vp8 x5 1198.1892+-14.6222 1186.0369+-12.7893 might be 1.0102x faster starfield x5 1166.9689+-10.9665 1161.9007+-14.5436 zynaps-quake3 x5 1364.4535+-50.7654 1319.2670+-105.8500 might be 1.0343x faster zynaps-mandelbrot x5 1144.3880+-12.4426 1140.5120+-17.3210 <arithmetic> 965.4289+-10.1073 952.8184+-27.3303 might be 1.0132x faster <geometric> * 671.6801+-5.3838 665.0883+-11.6797 might be 1.0099x faster <harmonic> 263.4108+-10.5501 262.4038+-8.9971 might be 1.0038x faster BASE PATCH All benchmarks: <arithmetic> 170.5102+-1.3270 168.8562+-3.6162 might be 1.0098x faster <geometric> 21.2481+-0.4398 ? 21.2524+-0.4938 ? might be 1.0002x slower <harmonic> 4.4008+-0.0773 4.3572+-0.0079 might be 1.0100x faster BASE PATCH Geomean of preferred means: <scaled-result> 42.2464+-0.5918 42.2031+-0.5879 might be 1.0010x faster

Geoffrey Garen

Comment 5 2013-03-07 19:03:22 PST

Committed r145171: <http://trac.webkit.org/changeset/145171>

Attachments
Patch (4.50 KB, patch) 2013-03-07 17:04 PST, Geoffrey Garen	oliver: review+	Details Formatted Diff Diff
View All Add attachment proposed patch, testcase, etc.

Bug 111797