[webkit-reviews] review requested: [Bug 64582] JSC JIT does not inline GC allocation fast paths : [Attachment 100924] the patch
bugzilla-daemon at webkit.org
bugzilla-daemon at webkit.org
Thu Jul 14 20:47:41 PDT 2011
Filip Pizlo <fpizlo at apple.com> has asked for review:
Bug 64582: JSC JIT does not inline GC allocation fast paths
https://bugs.webkit.org/show_bug.cgi?id=64582
Attachment 100924: the patch
https://bugs.webkit.org/attachment.cgi?id=100924&action=review
------- Additional Comments from Filip Pizlo <fpizlo at apple.com>
This patch works gives a 1.3% speed-up on V8, and is neutral on SunSpider.
I've lived on it for a day or so, and it seems to work. It passes all tests.
But I'm not sure if I'm handling global objects (for initializing the structure
in op_new_object) correctly, so I'd appreciate some feedback on that!
Separately I tested the speed-up of op_new_object in isolation, and it wasn't
much (statistically significant but fraction of a percent); it definitely
appears that op_create_this is doing most of the work.
[pizlo at minime PerformanceTests] ../Tools/Scripts/sunspider-compare-results --v8
v8-v4-results/sunspider-results-2011-07-14-20.03.02.js
/Volumes/Data/pizlo/quinary/OpenSource/PerformanceTests/SunSpider/v8-v4-results
/sunspider-results-2011-07-14-20.00.09.js
TEST COMPARISON FROM TO
DETAILS
=============================================================================
** TOTAL **: 1.013x as fast 1166.2ms +/- 0.3% 1150.7ms +/- 0.3%
significant
=============================================================================
v8: 1.013x as fast 1166.2ms +/- 0.3% 1150.7ms +/- 0.3%
significant
crypto: 1.016x as fast 194.6ms +/- 0.6% 191.6ms +/- 0.4%
significant
deltablue: 1.014x as fast 240.7ms +/- 0.8% 237.4ms +/- 0.5%
significant
earley-boyer: 1.028x as fast 129.7ms +/- 0.3% 126.2ms +/- 0.4%
significant
raytrace: 1.050x as fast 73.8ms +/- 0.8% 70.3ms +/- 1.4%
significant
regexp: ?? 105.9ms +/- 0.6% 106.2ms +/- 0.4%
not conclusive: might be *1.003x as slow*
richards: - 223.8ms +/- 0.6% 223.8ms +/- 0.9%
splay: 1.013x as fast 197.7ms +/- 0.5% 195.2ms +/- 0.9%
significant
[pizlo at minime PerformanceTests] ../Tools/Scripts/sunspider-compare-results
sunspider-1.0-results/sunspider-results-2011-07-14-20.04.12.js
/Volumes/Data/pizlo/quinary/OpenSource/PerformanceTests/SunSpider/sunspider-1.0
-results/sunspider-results-2011-07-14-20.04.56.js
TEST COMPARISON FROM TO
DETAILS
=============================================================================
** TOTAL **: - 174.4ms +/- 0.3% 173.9ms +/- 0.4%
=============================================================================
3d: - 25.2ms +/- 1.0% 25.1ms +/- 0.9%
cube: ?? 8.9ms +/- 1.0% 9.0ms +/- 0.8%
not conclusive: might be *1.009x as slow*
morph: - 7.3ms +/- 3.3% 7.1ms +/- 1.4%
raytrace: ?? 9.0ms +/- 0.0% 9.0ms +/- 0.9%
not conclusive: might be *1.004x as slow*
access: - 22.1ms +/- 0.4% 22.1ms +/- 0.5%
binary-trees: ?? 2.0ms +/- 2.0% 2.0ms +/- 2.8%
not conclusive: might be *1.010x as slow*
fannkuch: - 11.1ms +/- 0.8% 11.1ms +/- 0.6%
nbody: - 6.0ms +/- 0.0% 6.0ms +/- 0.0%
nsieve: ?? 3.0ms +/- 0.0% 3.0ms +/- 1.3%
not conclusive: might be *1.007x as slow*
bitops: ?? 15.4ms +/- 1.2% 15.5ms +/- 1.0%
not conclusive: might be *1.006x as slow*
3bit-bits-in-byte: - 2.0ms +/- 0.0% 2.0ms +/- 0.0%
bits-in-byte: ?? 5.4ms +/- 2.6% 5.5ms +/- 2.6%
not conclusive: might be *1.019x as slow*
bitwise-and: - 3.1ms +/- 2.5% 3.0ms +/- 1.9%
nsieve-bits: ?? 5.0ms +/- 1.1% 5.0ms +/- 0.0%
not conclusive: might be *1.008x as slow*
controlflow: ?? 1.1ms +/- 6.4% 1.1ms +/- 7.8%
not conclusive: might be *1.038x as slow*
recursive: ?? 1.1ms +/- 6.4% 1.1ms +/- 7.8%
not conclusive: might be *1.038x as slow*
crypto: - 11.0ms +/- 0.5% 10.9ms +/- 1.1%
aes: 1.017x as fast 7.0ms +/- 0.6% 6.9ms +/- 1.5%
significant
md5: ?? 2.0ms +/- 2.0% 2.0ms +/- 2.8%
not conclusive: might be *1.010x as slow*
sha1: - 2.0ms +/- 0.0% 2.0ms +/- 0.0%
date: 1.020x as fast 21.9ms +/- 0.6% 21.5ms +/- 0.9%
significant
format-tofte: 1.036x as fast 13.8ms +/- 0.9% 13.4ms +/- 1.0%
significant
format-xparb: ?? 8.0ms +/- 0.7% 8.1ms +/- 1.1%
not conclusive: might be *1.007x as slow*
math: - 16.0ms +/- 0.0% 16.0ms +/- 0.0%
cordic: - 6.0ms +/- 0.0% 6.0ms +/- 0.0%
partial-sums: - 7.0ms +/- 0.0% 7.0ms +/- 0.0%
spectral-norm: - 3.0ms +/- 0.0% 3.0ms +/- 0.0%
regexp: ?? 10.0ms +/- 0.0% 10.0ms +/- 0.4%
not conclusive: might be *1.002x as slow*
dna: ?? 10.0ms +/- 0.0% 10.0ms +/- 0.4%
not conclusive: might be *1.002x as slow*
string: - 51.7ms +/- 0.4% 51.6ms +/- 0.9%
base64: 1.062x as fast 5.5ms +/- 2.6% 5.2ms +/- 2.0%
significant
fasta: 1.036x as fast 7.0ms +/- 0.6% 6.7ms +/- 1.9%
significant
tagcloud: ?? 13.0ms +/- 0.3% 13.0ms +/- 0.4%
not conclusive: might be *1.002x as slow*
unpack-code: *1.021x as slow* 20.2ms +/- 0.5% 20.6ms +/- 1.2%
significant
validate-input: ?? 6.1ms +/- 1.4% 6.1ms +/- 1.5%
not conclusive: might be *1.003x as slow*
[pizlo at minime PerformanceTests]
More information about the webkit-reviews
mailing list