|
15楼#
发布于：2010-08-13 08:54
Bug 578133 – [meta] JM: Make us fast!
https://bugzilla.mozilla.org/show_bug.c ... aegerSpeed Bug 536277 – [meta] JaegerMonkey: Baseline method JIT https://bugzilla.mozilla.org/show_bug.cgi?id=536277 还有很多要改进的。 |
|
|
|
16楼#
发布于：2010-08-13 08:54
都开是不是就等于JM+TM了？ 目前老的TM也有600ms的成绩，JM的优势暂时很小，除了尝鲜似乎没啥理由升级了，除非那个v8性能关系很大？ |
|
|
17楼#
发布于：2010-08-13 08:54
刚自己测了下，有点失望，我不敢说官方的数据有多少水分，但至少提升幅度方面和我测下来的不符。。。
0813中文每夜版（只有TM引擎）：787.6ms jaegermonkey-1281688675小时版：739.8ms Opera10.6：463.2ms 之前测的0801中文每夜版：794.6ms PS. JM版的4个jit都设成true的话sunspider无法正常运行，故只将javascript.options.methodjit.chrome和javascript.options.methodjit.content设成true |
|
|
18楼#
发布于：2010-08-13 08:54
jaegermonkey测试成绩，两天前的版本
http://www2.webkit.org/perf/sunspider-0 ... sults.html?{%223d-cube%22:[20,19,19,19,19],%223d-morph%22:[17,17,17,18,17],%223d-raytrace%22:[23,23,24,23,23],%22access-binary-trees%22:[20,21,20,20,19],%22access-fannkuch%22:[19,19,20,19,19],%22access-nbody%22:[13,14,13,13,13],%22access-nsieve%22:[9,8,8,8,8],%22bitops-3bit-bits-in-byte%22:[6,8,6,6,6],%22bitops-bits-in-byte%22:[11,12,12,11,11],%22bitops-bitwise-and%22:[5,5,4,5,5],%22bitops-nsieve-bits%22:[8,8,11,8,8],%22controlflow-recursive%22:[11,11,11,11,11],%22crypto-aes%22:[20,15,15,21,16],%22crypto-md5%22:[9,10,8,9,9],%22crypto-sha1%22:[7,8,7,7,7],%22date-format-tofte%22:[43,41,42,41,42],%22date-format-xparb%22:[32,30,31,31,32],%22math-cordic%22:[11,12,12,12,12],%22math-partial-sums%22:[20,21,21,20,21],%22math-spectral-norm%22:[11,11,11,10,11],%22regexp-dna%22:[54,55,51,50,50],%22string-base64%22:[12,12,12,12,12],%22string-fasta%22:[20,20,22,20,20],%22string-tagcloud%22:[64,67,67,66,66],%22string-unpack-code%22:[63,61,60,61,60],%22string-validate-input%22:[19,21,21,20,21]} TM引擎,这个是上周的 http://www2.webkit.org/perf/sunspider-0 ... sults.html?{%223d-cube%22:[19,19,18,20,19],%223d-morph%22:[18,18,17,39,17],%223d-raytrace%22:[24,23,23,24,24],%22access-binary-trees%22:[20,20,20,22,20],%22access-fannkuch%22:[19,19,19,19,19],%22access-nbody%22:[13,13,13,13,13],%22access-nsieve%22:[11,9,9,8,9],%22bitops-3bit-bits-in-byte%22:[7,6,6,7,7],%22bitops-bits-in-byte%22:[12,12,12,11,11],%22bitops-bitwise-and%22:[5,5,4,4,5],%22bitops-nsieve-bits%22:[8,8,8,8,8],%22controlflow-recursive%22:[13,12,12,13,12],%22crypto-aes%22:[14,15,15,14,15],%22crypto-md5%22:[9,9,9,9,9],%22crypto-sha1%22:[9,9,9,9,9],%22date-format-tofte%22:[41,41,41,42,41],%22date-format-xparb%22:[34,32,32,32,32],%22math-cordic%22:[12,11,13,12,11],%22math-partial-sums%22:[20,20,20,20,20],%22math-spectral-norm%22:[11,11,11,11,11],%22regexp-dna%22:[54,52,60,60,62],%22string-base64%22:[12,13,11,11,12],%22string-fasta%22:[18,18,18,17,18],%22string-tagcloud%22:[49,48,55,55,56],%22string-unpack-code%22:[89,122,120,121,122],%22string-validate-input%22:[18,21,20,20,20]} |
|
|
19楼#
发布于：2010-08-13 08:54
just4fun：刚自己测了下，有点失望，我不敢说官方的数据有多少水分，但至少提升幅度方面和我测下来的不符。。。 arewefastyet上的数据来自于tracemonkey和jaegermonkey的trunk，测每夜版当然不对。 |
|
|
20楼#
发布于：2010-08-13 08:54
|
|
|
21楼#
发布于：2010-08-13 08:54
老兄，JM为什么在你电脑上速度不快进步不大，我不知道，或许你的电脑比较特别，或者JM有问题。 但是TM这次的速度改进主要来自于yarr，这个可没进每夜版。 如果你觉得可以类推的话，那请继续吧。 |
|
|
22楼#
发布于：2010-08-13 08:54
给我一个不能类推的理由。JM的速度按照官方数据会比你说的yarr后的TM更高，而按你的意思yarr后TM的性能又要高于每夜版。也就是说JM相比每夜版更应该高出不少才对，而我测下来并不是那么回事。 至于原因，要么官方还有什么更牛b的JM版本没放出来，要么就如我前面的“PS”所说，是两个jit没有打开的缘故。 |
|
|
23楼#
发布于：2010-08-13 08:54
淡定，最多几周就能见成效了，再说，你既没做出贡献，也没付出金钱，不要这么激动吧
|
|
|
24楼#
发布于：2010-08-13 08:54
告诉你sunspider测试跟你的电脑配置是挂钩的 所以会出现以上问题 V8测试我就不知道了 |
|
|
25楼#
发布于：2010-08-13 08:54
|
|
|
26楼#
发布于：2010-08-13 08:54
是的，我电脑上sunspider0.9 Firefox 4.0b3pre 20100801 606.0ms Chrome 6.0.472.14 Dev 400.0ms |
|
|
27楼#
发布于：2010-08-13 08:54
配置高点也会好很多
|
|
|
28楼#
发布于：2010-08-13 08:54
当然有一点需要说明的是，Firefox的 Javascript支持乱序执行，而大多数测试却都是顺序执行。由于很多浏览器都对顺序执行专门做过优化，因此在跑分时 Firefox难免会有些吃亏，这一点还是值得特别注意的！
转贴，实际应用才是检验的真理，跑分只是一个侧面 |
|
|
29楼#
发布于：2010-08-13 08:54
jaegermonkey 引擎速度真是快 感觉和chrome没啥区别了 仅靠感觉
============================================ RESULTS (means and 95% confidence intervals) -------------------------------------------- Total: 610.4ms +/- 0.8% -------------------------------------------- 3d: 115.0ms +/- 2.0% cube: 41.4ms +/- 3.4% morph: 27.4ms +/- 4.1% raytrace: 46.2ms +/- 1.2% access: 80.8ms +/- 2.3% binary-trees: 21.6ms +/- 5.1% fannkuch: 31.4ms +/- 2.2% nbody: 9.0ms +/- 0.0% nsieve: 18.8ms +/- 5.5% bitops: 22.2ms +/- 7.3% 3bit-bits-in-byte: 0.8ms +/- 69.5% bits-in-byte: 7.6ms +/- 9.0% bitwise-and: 2.0ms +/- 0.0% nsieve-bits: 11.8ms +/- 8.8% controlflow: 10.0ms +/- 8.8% recursive: 10.0ms +/- 8.8% crypto: 51.8ms +/- 5.2% aes: 27.6ms +/- 8.2% md5: 17.2ms +/- 3.2% sha1: 7.0ms +/- 0.0% date: 74.0ms +/- 2.4% format-tofte: 37.4ms +/- 3.0% format-xparb: 36.6ms +/- 1.9% math: 57.2ms +/- 5.2% cordic: 35.6ms +/- 8.8% partial-sums: 13.2ms +/- 4.2% spectral-norm: 8.4ms +/- 8.1% regexp: 48.8ms +/- 3.3% dna: 48.8ms +/- 3.3% string: 150.6ms +/- 0.9% base64: 8.4ms +/- 8.1% fasta: 27.2ms +/- 5.0% tagcloud: 51.4ms +/- 1.3% unpack-code: 49.8ms +/- 2.1% validate-input: 13.8ms +/- 4.0% ----------------------------------------------------------- chromeplus 1.4.1.0 ============================================ RESULTS (means and 95% confidence intervals) -------------------------------------------- Total: 426.6ms +/- 2.8% -------------------------------------------- 3d: 70.4ms +/- 5.4% cube: 27.6ms +/- 15.1% morph: 23.2ms +/- 16.3% raytrace: 19.6ms +/- 5.7% access: 37.8ms +/- 5.4% binary-trees: 2.4ms +/- 46.3% fannkuch: 13.8ms +/- 7.5% nbody: 17.0ms +/- 7.3% nsieve: 4.6ms +/- 14.8% bitops: 32.6ms +/- 5.8% 3bit-bits-in-byte: 2.8ms +/- 19.9% bits-in-byte: 6.6ms +/- 10.3% bitwise-and: 9.4ms +/- 11.8% nsieve-bits: 13.8ms +/- 17.3% controlflow: 3.0ms +/- 0.0% recursive: 3.0ms +/- 0.0% crypto: 26.0ms +/- 7.6% aes: 10.2ms +/- 5.5% md5: 8.6ms +/- 21.9% sha1: 7.2ms +/- 7.7% date: 54.4ms +/- 7.7% format-tofte: 26.4ms +/- 10.3% format-xparb: 28.0ms +/- 8.3% math: 46.2ms +/- 11.8% cordic: 15.4ms +/- 16.7% partial-sums: 22.0ms +/- 17.0% spectral-norm: 8.8ms +/- 6.3% regexp: 14.8ms +/- 7.0% dna: 14.8ms +/- 7.0% string: 141.4ms +/- 11.0% base64: 15.0ms +/- 8.3% fasta: 21.4ms +/- 11.3% tagcloud: 37.4ms +/- 8.3% unpack-code: 38.8ms +/- 2.7% validate-input: 28.8ms +/- 46.1% |
|
