Bug 22821 - Combine op_bitand and op_jfalse instructions
Summary: Combine op_bitand and op_jfalse instructions
Status: RESOLVED INVALID
Alias: None
Product: WebKit
Classification: Unclassified
Component: JavaScriptCore (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC Linux
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-12-12 02:25 PST by Gabor Loki
Modified: 2012-09-06 17:52 PDT (History)
3 users (show)

See Also:


Attachments
Add op_jnbitand instruction (11.23 KB, patch)
2008-12-12 02:27 PST, Gabor Loki
no flags Details | Formatted Diff | Diff
SunSpider, V8, WindScorpion results (non-jit) (4.56 KB, text/plain)
2008-12-12 02:28 PST, Gabor Loki
no flags Details
SunSpider, V8, WindScorpion results (JIT) (4.16 KB, text/plain)
2008-12-12 02:28 PST, Gabor Loki
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Gabor Loki 2008-12-12 02:25:45 PST
I have found about 1% progression if we combine op_bitand and op_jfalse instructions.

I am going to attach the patch and the results.
Comment 1 Gabor Loki 2008-12-12 02:27:49 PST
Created attachment 25976 [details]
Add op_jnbitand instruction
Comment 2 Gabor Loki 2008-12-12 02:28:36 PST
Created attachment 25977 [details]
SunSpider, V8, WindScorpion results (non-jit)
Comment 3 Gabor Loki 2008-12-12 02:28:57 PST
Created attachment 25978 [details]
SunSpider, V8, WindScorpion results (JIT)
Comment 4 Oliver Hunt 2009-01-22 18:36:38 PST
Comment on attachment 25976 [details]
Add op_jnbitand instruction

Alas, i think the jit code for this is too out of date to be useful anymore, the code itself looks fine, so rather than r- i'll just clear the review flag.

sorry for the huge delay in review time.
Comment 5 Gabor Loki 2009-05-13 05:02:37 PDT
I did the performance test again on x86 without JIT, and there was no more progression.
Comment 6 Gavin Barraclough 2012-09-06 17:52:12 PDT
This patch won't still apply; this kind of optimization is now more appropriate in the DFG JIT.