117576 – The Math object should not be polymorphic

Geoffrey Garen

Reported 2013-06-12 18:55:04 PDT

The Math object should not be polymorphic

Geoffrey Garen

Comment 1 2013-06-12 19:00:23 PDT

Created attachment 204552 [details] Patch

Geoffrey Garen

Comment 2 2013-06-12 19:01:16 PDT

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 (r151529) "PATCH" at /Volumes/Big/ggaren/ftl-webkit/WebKitBuild/Release/DumpRenderTree (r151529) 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. BASE PATCH SunSpider: 3d-cube 6.9991+-0.3032 6.8197+-0.3479 might be 1.0263x faster 3d-morph 8.3527+-0.0712 ^ 7.9616+-0.0550 ^ definitely 1.0491x faster 3d-raytrace 8.8951+-0.1768 8.6686+-0.2393 might be 1.0261x faster access-binary-trees 2.4808+-0.3150 ? 2.6794+-0.3819 ? might be 1.0800x slower access-fannkuch 7.3568+-0.0701 ? 7.4152+-0.0770 ? access-nbody 3.8978+-0.0526 ? 3.9243+-0.0474 ? access-nsieve 4.7610+-0.0628 4.7253+-0.0599 bitops-3bit-bits-in-byte 1.7083+-0.0200 ? 1.7228+-0.0292 ? bitops-bits-in-byte 6.2519+-0.0604 ? 6.2941+-0.0516 ? bitops-bitwise-and 2.6169+-0.0395 ? 2.6352+-0.0414 ? bitops-nsieve-bits 4.4163+-0.0165 ? 4.4756+-0.0501 ? might be 1.0134x slower controlflow-recursive 2.9074+-0.1454 2.8671+-0.0896 might be 1.0141x faster crypto-aes 5.6150+-0.2921 ? 5.6730+-0.2745 ? might be 1.0103x slower crypto-md5 3.0665+-0.0907 ? 3.0793+-0.0702 ? crypto-sha1 2.8525+-0.0569 2.7591+-0.0374 might be 1.0338x faster date-format-tofte 14.9790+-0.9353 ? 15.0234+-1.0384 ? date-format-xparb 8.8826+-0.6562 8.7330+-0.6147 might be 1.0171x faster math-cordic 3.7240+-0.0240 3.7147+-0.0132 math-partial-sums 10.0168+-0.1553 9.8626+-0.0303 might be 1.0156x faster math-spectral-norm 2.7207+-0.0397 2.7201+-0.0405 regexp-dna 12.2831+-0.6124 12.2524+-0.5788 string-base64 6.0654+-0.5023 6.0057+-0.4850 string-fasta 11.0982+-0.2619 11.0202+-0.2278 string-tagcloud 14.4039+-0.2209 14.3952+-0.1995 string-unpack-code 30.2715+-0.5485 ? 30.6999+-0.6442 ? might be 1.0142x slower string-validate-input 8.5635+-0.2414 8.5428+-0.2733 <arithmetic> * 7.5072+-0.1189 7.4873+-0.1091 might be 1.0027x faster <geometric> 5.9255+-0.0863 5.9121+-0.0717 might be 1.0023x faster <harmonic> 4.7884+-0.0716 ? 4.7952+-0.0594 ? might be 1.0014x slower BASE PATCH V8Spider: crypto 78.0422+-0.4432 ? 78.1521+-0.4962 ? deltablue 103.0171+-7.1357 100.8225+-0.7147 might be 1.0218x faster earley-boyer 70.1648+-0.8430 69.5409+-0.7757 raytrace 43.3004+-5.0369 ? 43.3072+-4.8296 ? regexp 97.2794+-0.4370 97.2206+-0.3244 richards 109.0338+-0.9527 108.7302+-2.1540 splay 57.7581+-3.0958 ? 58.0636+-2.9161 ? <arithmetic> 79.7994+-1.5956 79.4053+-0.9631 might be 1.0050x faster <geometric> * 76.0318+-1.6835 75.7832+-1.2647 might be 1.0033x faster <harmonic> 71.9395+-2.0388 71.8198+-1.7505 might be 1.0017x faster BASE PATCH Octane and V8v7: encrypt 0.45834+-0.00028 ? 0.45870+-0.00045 ? decrypt 8.27619+-0.00987 ? 8.28891+-0.01904 ? deltablue x2 0.58152+-0.00935 0.57798+-0.00915 earley 0.86687+-0.00458 0.86103+-0.00574 boyer 12.65223+-0.08544 12.53084+-0.10746 raytrace x2 4.27644+-0.04388 4.27362+-0.06361 regexp x2 30.07173+-0.05846 ? 30.23636+-0.18203 ? richards x2 0.29700+-0.00278 ? 0.29864+-0.00332 ? splay x2 0.62119+-0.00519 ? 0.64181+-0.04272 ? might be 1.0332x slower navier-stokes x2 10.38257+-0.00505 ? 10.40067+-0.01470 ? closure 0.30671+-0.03901 ? 0.30839+-0.03936 ? jquery 4.46597+-0.53022 ? 4.47769+-0.51764 ? gbemu x2 137.87103+-6.44599 ? 138.38188+-7.31860 ? mandreel x2 181.38021+-0.36514 ? 181.38151+-0.71644 ? pdfjs x2 221.85358+-0.81649 220.88466+-0.80310 box2d x2 33.75115+-0.29258 33.64008+-0.16609 V8v7: <arithmetic> 7.16966+-0.01068 ? 7.18735+-0.02703 ? might be 1.0025x slower <geometric> * 2.34789+-0.00678 ? 2.35601+-0.02020 ? might be 1.0035x slower <harmonic> 0.90610+-0.00406 ? 0.91080+-0.01002 ? might be 1.0052x slower Octane including V8v7: <arithmetic> 48.81535+-0.53607 48.78308+-0.55469 might be 1.0007x faster <geometric> * 7.39751+-0.05609 ? 7.41337+-0.08361 ? might be 1.0021x slower <harmonic> 1.21854+-0.02729 ? 1.22522+-0.03327 ? might be 1.0055x slower BASE PATCH Kraken: ai-astar 500.636+-0.830 ? 501.509+-0.752 ? audio-beat-detection 247.558+-2.759 ^ 237.977+-0.764 ^ definitely 1.0403x faster audio-dft 297.394+-0.584 296.687+-0.409 audio-fft 141.293+-0.206 ^ 138.251+-0.400 ^ definitely 1.0220x faster audio-oscillator 239.815+-0.872 ^ 236.472+-0.557 ^ definitely 1.0141x faster imaging-darkroom 277.984+-0.421 ! 279.554+-0.721 ! definitely 1.0056x slower imaging-desaturate 150.844+-0.197 150.632+-0.078 imaging-gaussian-blur 418.371+-0.290 ? 418.949+-1.844 ? json-parse-financial 76.993+-0.182 ^ 76.312+-0.182 ^ definitely 1.0089x faster json-stringify-tinderbox 96.368+-0.279 ? 96.919+-0.304 ? stanford-crypto-aes 103.979+-0.711 103.969+-0.409 stanford-crypto-ccm 99.441+-7.253 ? 102.757+-7.329 ? might be 1.0333x slower stanford-crypto-pbkdf2 265.350+-0.994 ^ 262.072+-1.283 ^ definitely 1.0125x faster stanford-crypto-sha256-iterative 117.640+-0.625 ? 118.064+-0.515 ? <arithmetic> * 216.691+-0.527 215.723+-0.506 might be 1.0045x faster <geometric> 184.305+-0.910 183.654+-0.881 might be 1.0035x faster <harmonic> 157.884+-1.198 157.676+-1.150 might be 1.0013x faster BASE PATCH JSBench: amazon 9.6667+-0.3128 ? 9.8333+-0.2473 ? might be 1.0172x slower facebook 45.0000+-1.0837 45.0000+-0.9768 google 76.7500+-1.2452 76.5000+-0.8779 twitter 20.4167+-1.4704 ? 21.3333+-1.4921 ? might be 1.0449x slower yahoo 7.0000+-0.2709 6.9167+-0.3272 might be 1.0120x faster <arithmetic> * 31.7667+-0.6633 ? 31.9167+-0.2680 ? might be 1.0047x slower <geometric> 21.6376+-0.5386 ? 21.8232+-0.2623 ? might be 1.0086x slower <harmonic> 15.0907+-0.3953 ? 15.1695+-0.3289 ? might be 1.0052x slower BASE PATCH JSRegress: adapt-to-double-divide 21.3873+-0.0410 ? 21.4557+-0.0455 ? aliased-arguments-getbyval 0.9938+-0.0374 0.9462+-0.0299 might be 1.0503x faster allocate-big-object 3.8024+-1.2864 ? 3.8289+-1.3248 ? arity-mismatch-inlining 0.8124+-0.0196 ? 0.8332+-0.0280 ? might be 1.0256x slower array-access-polymorphic-structure 8.0899+-1.7841 ? 8.0994+-1.7983 ? array-nonarray-polymorhpic-access 188.0348+-3.7579 186.3077+-3.6770 array-with-double-add 5.6256+-0.0562 5.6063+-0.0351 array-with-double-increment 3.9105+-0.0262 ? 4.0048+-0.0825 ? might be 1.0241x slower array-with-double-mul-add 7.2449+-0.0515 ? 7.2563+-0.0282 ? array-with-double-sum 7.6467+-0.1391 ^ 7.4265+-0.0142 ^ definitely 1.0296x faster array-with-int32-add-sub 9.9902+-0.0597 ? 10.0450+-0.0231 ? array-with-int32-or-double-sum 7.5450+-0.0318 7.5403+-0.0323 big-int-mul 4.6112+-0.0428 ? 4.6114+-0.0509 ? boolean-test 4.1355+-0.0308 ? 4.1437+-0.0474 ? branch-fold 15.6613+-0.0680 15.4183+-0.2578 might be 1.0158x faster cast-int-to-double 13.4604+-0.0396 ? 13.5197+-0.0494 ? cell-argument 13.6243+-0.0231 ? 13.6604+-0.0218 ? cfg-simplify 3.5991+-0.0391 3.5623+-0.0085 might be 1.0103x faster cmpeq-obj-to-obj-other 13.8341+-0.3669 13.4664+-0.2580 might be 1.0273x faster constant-test 8.1694+-0.0374 8.1547+-0.0518 direct-arguments-getbyval 0.7406+-0.0119 ! 0.7769+-0.0235 ! definitely 1.0490x slower double-pollution-getbyval 10.2322+-0.0448 10.2249+-0.0379 double-pollution-putbyoffset 6.2791+-0.5816 6.2599+-0.6063 empty-string-plus-int 13.2319+-0.5126 13.1191+-0.6010 external-arguments-getbyval 2.3097+-0.2045 2.1503+-0.1822 might be 1.0741x faster external-arguments-putbyval 3.6681+-0.3275 3.5223+-0.3084 might be 1.0414x faster Float32Array-matrix-mult 13.7786+-0.5272 ? 14.2005+-0.8213 ? might be 1.0306x slower fold-double-to-int 20.2194+-0.0262 ! 24.7994+-0.7005 ! definitely 1.2265x slower function-dot-apply 2.8550+-0.0130 ! 2.9300+-0.0230 ! definitely 1.0263x slower function-test 4.5477+-0.0731 ? 4.5989+-0.0829 ? might be 1.0113x slower get-by-id-chain-from-try-block 6.4857+-0.1247 ? 6.6065+-0.0934 ? might be 1.0186x slower HashMap-put-get-iterate-keys 80.4716+-1.2814 ^ 73.8619+-1.1423 ^ definitely 1.0895x faster HashMap-put-get-iterate 83.9164+-0.5552 ^ 78.5512+-0.8288 ^ definitely 1.0683x faster HashMap-string-put-get-iterate 73.8752+-0.6889 71.7851+-1.4092 might be 1.0291x faster imul-double-only 0.2782+-0.0095 ? 0.2815+-0.0092 ? might be 1.0121x slower imul-int-only 0.2810+-0.0083 ? 0.2870+-0.0082 ? might be 1.0216x slower imul-mixed 0.2878+-0.0058 ? 0.2889+-0.0071 ? indexed-properties-in-objects 3.9407+-0.0295 3.9268+-0.0067 inline-arguments-access 1.3727+-0.0323 ? 1.4049+-0.0340 ? might be 1.0235x slower inline-arguments-local-escape 24.2666+-0.2398 23.9844+-0.3015 might be 1.0118x faster inline-get-scoped-var 6.4433+-0.0546 ? 6.4577+-0.0732 ? inlined-put-by-id-transition 14.2969+-0.2204 14.1698+-0.2214 int-or-other-abs-then-get-by-val 8.4292+-0.0529 8.3988+-0.0313 int-or-other-abs-zero-then-get-by-val 34.8242+-0.0701 ? 34.9314+-0.1521 ? int-or-other-add-then-get-by-val 9.6831+-0.0226 ? 9.8327+-0.1506 ? might be 1.0155x slower int-or-other-add 10.0285+-0.0501 10.0021+-0.0597 int-or-other-div-then-get-by-val 5.9379+-0.0461 5.9217+-0.0388 int-or-other-max-then-get-by-val 6.8973+-0.0545 6.8946+-0.0420 int-or-other-min-then-get-by-val 6.8865+-0.0143 ? 6.8977+-0.0323 ? int-or-other-mod-then-get-by-val 5.7429+-0.0283 5.7337+-0.0391 int-or-other-mul-then-get-by-val 6.1238+-0.0347 6.1128+-0.0500 int-or-other-neg-then-get-by-val 7.4912+-0.0322 ? 7.5318+-0.0675 ? int-or-other-neg-zero-then-get-by-val 34.8872+-0.1834 ? 35.3324+-0.6152 ? might be 1.0128x slower int-or-other-sub-then-get-by-val 9.6979+-0.0451 9.6894+-0.0409 int-or-other-sub 7.7451+-0.0310 ? 7.7612+-0.0405 ? int-overflow-local 11.2064+-0.0528 11.1561+-0.0164 Int16Array-bubble-sort 94.2755+-1.6931 93.7767+-1.2190 Int16Array-load-int-mul 1.8600+-0.0186 ? 1.9090+-0.0350 ? might be 1.0264x slower Int8Array-load 6.5061+-0.0610 6.4567+-0.0269 integer-divide 14.5612+-0.0289 ? 14.5791+-0.0300 ? integer-modulo 1.8699+-0.0229 ? 1.9019+-0.0252 ? might be 1.0171x slower make-indexed-storage 4.2468+-0.6140 4.2362+-0.6114 method-on-number 24.2399+-0.3364 23.8131+-0.1479 might be 1.0179x faster negative-zero-divide 0.3235+-0.0101 0.3222+-0.0127 negative-zero-modulo 0.3160+-0.0120 ? 0.3307+-0.0255 ? might be 1.0463x slower negative-zero-negate 0.2992+-0.0117 0.2956+-0.0063 might be 1.0124x faster nested-function-parsing-random 368.9416+-11.6692 ? 373.4030+-11.0124 ? might be 1.0121x slower nested-function-parsing 54.9313+-3.1723 54.8957+-3.2386 new-array-buffer-dead 3.6772+-0.0848 ? 3.7164+-0.1052 ? might be 1.0107x slower new-array-buffer-push 13.8055+-2.3158 12.4037+-2.3424 might be 1.1130x faster new-array-dead 26.9615+-0.0825 ? 26.9633+-0.0566 ? new-array-push 11.5011+-1.8352 ? 11.5137+-1.7983 ? number-test 4.0621+-0.0324 ? 4.0925+-0.0299 ? object-closure-call 8.2380+-0.1754 ! 10.9695+-1.9251 ! definitely 1.3316x slower object-test 4.4688+-0.0296 4.4193+-0.0423 might be 1.0112x faster poly-stricteq 87.9036+-0.5866 ^ 86.4678+-0.4603 ^ definitely 1.0166x faster polymorphic-structure 20.0117+-0.1527 19.9555+-0.1834 polyvariant-monomorphic-get-by-id 11.8686+-0.0291 ? 11.8974+-0.0697 ? rare-osr-exit-on-local 19.0707+-0.0828 19.0498+-0.0805 register-pressure-from-osr 29.8273+-0.0792 ? 29.8721+-0.0779 ? simple-activation-demo 32.8762+-0.2876 ? 33.0950+-0.2822 ? slow-array-profile-convergence 4.2171+-0.2445 4.1717+-0.2136 might be 1.0109x faster slow-convergence 3.4445+-0.0616 ? 3.4617+-0.0567 ? sparse-conditional 1.2267+-0.0281 1.2253+-0.0265 splice-to-remove 48.9070+-0.7501 ^ 47.4499+-0.1390 ^ definitely 1.0307x faster string-concat-object 4.7962+-1.3424 ? 4.8372+-1.3586 ? string-concat-pair-object 4.7803+-1.3515 4.7740+-1.3813 string-concat-pair-simple 18.9302+-0.6459 ? 19.1429+-0.7021 ? might be 1.0112x slower string-concat-simple 19.0452+-0.5305 19.0322+-0.4916 string-cons-repeat 14.2651+-0.8958 ? 14.2810+-0.8786 ? string-cons-tower 14.6724+-0.8409 14.6491+-0.8555 string-equality 111.5154+-2.1951 ^ 99.6235+-0.6686 ^ definitely 1.1194x faster string-hash 2.4866+-0.0333 ^ 2.4425+-0.0078 ^ definitely 1.0181x faster string-repeat-arith 46.0893+-0.2429 ! 47.0634+-0.5247 ! definitely 1.0211x slower string-sub 89.7635+-1.2982 ? 91.4326+-2.1428 ? might be 1.0186x slower string-test 4.0297+-0.0381 ? 4.0514+-0.0327 ? structure-hoist-over-transitions 3.7215+-0.5904 ? 3.7391+-0.6209 ? tear-off-arguments-simple 2.2065+-0.0567 2.1469+-0.0475 might be 1.0278x faster tear-off-arguments 3.3256+-0.0453 3.2783+-0.0468 might be 1.0144x faster temporal-structure 20.0640+-0.0975 ? 20.1285+-0.0757 ? to-int32-boolean 20.7739+-0.0509 ? 20.7879+-0.0407 ? undefined-test 4.1944+-0.0185 ? 4.2030+-0.0106 ? <arithmetic> 21.2807+-0.2452 21.1032+-0.2487 might be 1.0084x faster <geometric> * 7.9629+-0.1182 ? 7.9673+-0.1157 ? might be 1.0006x slower <harmonic> 2.6583+-0.0323 ? 2.6747+-0.0251 ? might be 1.0062x slower BASE PATCH DSP: filtrr-posterize-tint 51.0953+-1.6954 ^ 41.7866+-1.2746 ^ definitely 1.2228x faster filtrr-tint-contrast-sat-bright 79.2356+-1.3352 ^ 68.4036+-0.3833 ^ definitely 1.1584x faster filtrr-tint-sat-adj-contr-mult 96.3122+-4.9107 ^ 81.1607+-4.4092 ^ definitely 1.1867x faster filtrr-blur-overlay-sat-contr 221.3623+-4.0859 ^ 208.3153+-4.5467 ^ definitely 1.0626x faster filtrr-sat-blur-mult-sharpen-contr 273.1110+-5.3634 ^ 246.6976+-4.4370 ^ definitely 1.1071x faster filtrr-sepia-bias 39.7671+-1.2545 ^ 32.7256+-3.1302 ^ definitely 1.2152x faster route9-vp8 x5 1107.6637+-5.3893 ? 1108.0560+-8.6864 ? starfield x5 1157.5269+-4.7271 1154.0246+-4.1049 bellard-jslinux x5 3035.9167+-16.3195 ? 3084.5833+-32.8582 ? might be 1.0160x slower zynaps-quake3 x5 1302.2194+-15.6498 ? 1306.2599+-20.9950 ? zynaps-mandelbrot x5 1117.2992+-4.4787 1115.0208+-4.1578 ammojs-asm-js x5 243.3205+-13.3205 238.4339+-13.4732 might be 1.0205x faster ammojs-regular-js x5 234.7532+-9.9576 231.1267+-12.0002 might be 1.0157x faster <arithmetic> 1018.3995+-2.8506 ? 1021.1370+-5.5250 ? might be 1.0027x slower <geometric> * 620.7194+-7.4518 605.7849+-8.0869 might be 1.0247x faster <harmonic> 299.8736+-8.2546 ^ 269.4401+-10.5573 ^ definitely 1.1130x faster BASE PATCH All benchmarks: <arithmetic> 222.3552+-0.6179 ? 222.7046+-1.0179 ? might be 1.0016x slower <geometric> 22.9315+-0.2308 22.8301+-0.1910 might be 1.0044x faster <harmonic> 3.3584+-0.0437 ? 3.3758+-0.0441 ? might be 1.0052x slower BASE PATCH Geomean of preferred means: <scaled-result> 39.2496+-0.3696 39.0978+-0.2523 might be 1.0039x faster

Darin Adler

Comment 3 2013-06-12 19:57:23 PDT

Comment on attachment 204552 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=204552&action=review > Source/JavaScriptCore/runtime/JSObject.cpp:2203 > + JSFunction* function = > + JSFunction::create(exec, globalObject, functionLength, name, nativeFunction, intrinsic); Seems unhelpful to split this into two lines.

Geoffrey Garen

Comment 4 2013-06-13 08:29:56 PDT

Committed r151548: <http://trac.webkit.org/changeset/151548>

Attachments
Patch (16.05 KB, patch) 2013-06-12 19:00 PDT, Geoffrey Garen	oliver: review+	Details Formatted Diff Diff
View All Add attachment proposed patch, testcase, etc.