Bug 132314 - Simplify tryCacheGetById
Summary: Simplify tryCacheGetById
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: JavaScriptCore (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Mark Hahnenberg
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-04-28 17:38 PDT by Mark Hahnenberg
Modified: 2014-04-28 20:11 PDT (History)
1 user (show)

See Also:

Attachments
Patch (3.68 KB, patch)
2014-04-28 17:40 PDT, Mark Hahnenberg 		oliver: review+
Details | Formatted Diff | Diff
Perf results (40.63 KB, text/plain)
2014-04-28 19:02 PDT, Mark Hahnenberg 		no flags Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Mark Hahnenberg 2014-04-28 17:38:43 PDT 
There's some duplicate logic between tryCacheGetById and tryBuildGetByIDList. We should just do self caching in tryCacheGetById, and if we fail then we should try calling tryBuildGetByIDList before giving up.
Comment 1 Mark Hahnenberg 2014-04-28 17:40:49 PDT 
Created attachment 230342 [details]
Patch
Comment 2 Filip Pizlo 2014-04-28 18:58:41 PDT 
Comment on attachment 230342 [details]
Patch

R=me too, what is the perf impact?
Comment 3 Mark Hahnenberg 2014-04-28 19:02:32 PDT 
Created attachment 230350 [details]
Perf results

Here's the results. Neutral. I re-ran sun spider with inner, outer = 8, 8, and it looked like a small progression ~0.5%.
Comment 4 Filip Pizlo 2014-04-28 19:30:58 PDT 
(In reply to comment #3)
> Created an attachment (id=230350) [details]
> Perf results
> 
> Here's the results. Neutral. I re-ran sun spider with inner, outer = 8, 8, and it looked like a small progression ~0.5%.

Yeah that looks pretty good!
Comment 5 Filip Pizlo 2014-04-28 19:31:14 PDT 
(In reply to comment #4)
> (In reply to comment #3)
> > Created an attachment (id=230350) [details] [details]
> > Perf results
> > 
> > Here's the results. Neutral. I re-ran sun spider with inner, outer = 8, 8, and it looked like a small progression ~0.5%.
> 
> Yeah that looks pretty good!

Wait ... did you enable FTL in that run?
Comment 6 Mark Hahnenberg 2014-04-28 19:47:59 PDT 
(In reply to comment #5)
> (In reply to comment #4)
> > (In reply to comment #3)
> > > Created an attachment (id=230350) [details] [details] [details]
> > > Perf results
> > > 
> > > Here's the results. Neutral. I re-ran sun spider with inner, outer = 8, 8, and it looked like a small progression ~0.5%.
> > 
> > Yeah that looks pretty good!
> 
> Wait ... did you enable FTL in that run?

Yep. I ran with the same flags and only added the "--sunspider --inner 8 --outer 8" before the VMs. Why?
Comment 7 Filip Pizlo 2014-04-28 20:06:11 PDT 
(In reply to comment #6)
> (In reply to comment #5)
> > (In reply to comment #4)
> > > (In reply to comment #3)
> > > > Created an attachment (id=230350) [details] [details] [details] [details]
> > > > Perf results
> > > > 
> > > > Here's the results. Neutral. I re-ran sun spider with inner, outer = 8, 8, and it looked like a small progression ~0.5%.
> > > 
> > > Yeah that looks pretty good!
> > 
> > Wait ... did you enable FTL in that run?
> 
> Yep. I ran with the same flags and only added the "--sunspider --inner 8 --outer 8" before the VMs. Why?

Just checking how you built since FTL is disabled by default on most OSes.
Comment 8 Mark Hahnenberg 2014-04-28 20:11:25 PDT 
Committed r167922: <http://trac.webkit.org/changeset/167922>