WebKit Bugzilla
New
Browse
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
Bug 64630
DFG JIT - Where arguments passed are integers, speculate this.
https://bugs.webkit.org/show_bug.cgi?id=64630
Summary
DFG JIT - Where arguments passed are integers, speculate this.
Gavin Barraclough
Reported
2011-07-15 14:37:46 PDT
Presently the DFG JIT is overly aggressively predicting double. Use a bit of dynamic information, and curtail this a little.
Attachments
The patch
(3.91 KB, patch)
2011-07-15 14:40 PDT
,
Gavin Barraclough
sam
: review+
Details
Formatted Diff
Diff
View All
Add attachment
proposed patch, testcase, etc.
Gavin Barraclough
Comment 1
2011-07-15 14:40:08 PDT
Created
attachment 101044
[details]
The patch TEST COMPARISON FROM TO DETAILS ============================================================================= ** TOTAL **: 1.006x as fast 235.3ms +/- 0.5% 233.8ms +/- 0.3% significant ============================================================================= 3d: - 34.0ms +/- 0.0% 34.0ms +/- 0.0% cube: - 13.0ms +/- 0.0% 13.0ms +/- 0.0% morph: - 10.0ms +/- 0.0% 10.0ms +/- 0.0% raytrace: - 11.0ms +/- 0.0% 11.0ms +/- 0.0% access: - 26.1ms +/- 3.8% 25.5ms +/- 1.5% binary-trees: - 4.0ms +/- 0.0% 4.0ms +/- 0.0% fannkuch: - 14.1ms +/- 6.9% 13.5ms +/- 2.8% nbody: - 5.0ms +/- 0.0% 5.0ms +/- 0.0% nsieve: - 3.0ms +/- 0.0% 3.0ms +/- 0.0% bitops: 1.045x as fast 18.6ms +/- 2.0% 17.8ms +/- 1.7% significant 3bit-bits-in-byte: 1.60x as fast 1.6ms +/- 23.1% 1.0ms +/- 0.0% significant bits-in-byte: - 5.0ms +/- 0.0% 5.0ms +/- 0.0% bitwise-and: - 6.0ms +/- 0.0% 5.8ms +/- 5.2% nsieve-bits: - 6.0ms +/- 0.0% 6.0ms +/- 0.0% controlflow: - 2.0ms +/- 0.0% 2.0ms +/- 0.0% recursive: - 2.0ms +/- 0.0% 2.0ms +/- 0.0% crypto: 1.069x as fast 14.0ms +/- 0.0% 13.1ms +/- 1.7% significant aes: 1.111x as fast 9.0ms +/- 0.0% 8.1ms +/- 2.8% significant md5: - 3.0ms +/- 0.0% 3.0ms +/- 0.0% sha1: - 2.0ms +/- 0.0% 2.0ms +/- 0.0% date: - 30.0ms +/- 0.0% 30.0ms +/- 0.0% format-tofte: - 18.0ms +/- 0.0% 18.0ms +/- 0.0% format-xparb: - 12.0ms +/- 0.0% 12.0ms +/- 0.0% math: ?? 22.5ms +/- 1.7% 22.8ms +/- 1.3% not conclusive: might be *1.013x as slow* cordic: - 8.0ms +/- 0.0% 8.0ms +/- 0.0% partial-sums: ?? 11.5ms +/- 3.3% 11.8ms +/- 2.6% not conclusive: might be *1.026x as slow* spectral-norm: - 3.0ms +/- 0.0% 3.0ms +/- 0.0% regexp: - 14.1ms +/- 1.6% 14.0ms +/- 0.0% dna: - 14.1ms +/- 1.6% 14.0ms +/- 0.0% string: ?? 74.0ms +/- 0.8% 74.6ms +/- 0.5% not conclusive: might be *1.008x as slow* base64: - 8.0ms +/- 0.0% 8.0ms +/- 0.0% fasta: - 10.0ms +/- 0.0% 10.0ms +/- 0.0% tagcloud: ?? 18.6ms +/- 2.0% 18.9ms +/- 1.2% not conclusive: might be *1.016x as slow* unpack-code: - 28.0ms +/- 0.0% 28.0ms +/- 0.0% validate-input: ?? 9.4ms +/- 3.9% 9.7ms +/- 3.6% not conclusive: might be *1.032x as slow*
Gavin Barraclough
Comment 2
2011-07-15 14:57:45 PDT
Fixed in
r91115
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug