Bug 112202 - [iOS] Add optimized version of StringImpl's equal(LChar*,LChar*) for Apple CPUs
Summary: [iOS] Add optimized version of StringImpl's equal(LChar*,LChar*) for Apple CPUs
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: New Bugs (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Benjamin Poulain
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2013-03-12 15:38 PDT by Benjamin Poulain
Modified: 2013-03-13 13:46 PDT (History)
10 users (show)

See Also:

Attachments
Patch (2.83 KB, patch)
2013-03-12 15:42 PDT, Benjamin Poulain 		no flags Details | Formatted Diff | Diff
Patch (4.25 KB, patch)
2013-03-12 17:06 PDT, Benjamin Poulain 		no flags Details | Formatted Diff | Diff
Patch (4.31 KB, patch)
2013-03-13 00:18 PDT, Benjamin Poulain 		no flags Details | Formatted Diff | Diff
Show Obsolete (2) 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 Benjamin Poulain 2013-03-12 15:38:40 PDT 
[iOS] Add optimized version of StringImpl's equal(LChar*,LChar*) for Apple CPUs
Comment 1 Benjamin Poulain 2013-03-12 15:42:56 PDT 
Created attachment 192827 [details]
Patch
Comment 2 Benjamin Poulain 2013-03-12 15:47:42 PDT 
<rdar://problem/12437844>
Comment 3 Early Warning System Bot 2013-03-12 16:01:05 PDT 
Comment on attachment 192827 [details]
Patch

Attachment 192827 [details] did not pass qt-ews (qt):
Output: http://webkit-commit-queue.appspot.com/results/17174180
Comment 4 Early Warning System Bot 2013-03-12 16:13:42 PDT 
Comment on attachment 192827 [details]
Patch

Attachment 192827 [details] did not pass qt-wk2-ews (qt):
Output: http://webkit-commit-queue.appspot.com/results/17074383
Comment 5 WebKit Review Bot 2013-03-12 16:44:08 PDT 
Comment on attachment 192827 [details]
Patch

Attachment 192827 [details] did not pass chromium-ews (chromium-xvfb):
Output: http://webkit-commit-queue.appspot.com/results/17201136
Comment 6 Benjamin Poulain 2013-03-12 17:06:08 PDT 
Created attachment 192836 [details]
Patch
Comment 7 Gavin Barraclough 2013-03-12 18:09:35 PDT 
Comment on attachment 192836 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=192836&action=review

Looks good!

> Source/WTF/wtf/text/StringImpl.h:933
> +        "b      66f\n"

I think you could invert this to:
-> "bneq 66f" "b 0b"

And if you then move the label "0:" down a line, you could branch past the cbz:
-> "bneq 66f" "cbnz 0b"

This would save an unconditional branch from the loop.
Don't know if this might help?
Comment 8 Peter Beverloo (cr-android ews) 2013-03-12 18:55:35 PDT 
Comment on attachment 192836 [details]
Patch

Attachment 192836 [details] did not pass cr-android-ews (chromium-android):
Output: http://webkit-commit-queue.appspot.com/results/17188165
Comment 9 Benjamin Poulain 2013-03-12 19:27:40 PDT 
> > Source/WTF/wtf/text/StringImpl.h:933
> > +        "b      66f\n"
> 
> I think you could invert this to:
> -> "bneq 66f" "b 0b"
> 
> And if you then move the label "0:" down a line, you could branch past the cbz:
> -> "bneq 66f" "cbnz 0b"
> 
> This would save an unconditional branch from the loop.
> Don't know if this might help?

Good idea! This is faster indeed.
Comment 10 Benjamin Poulain 2013-03-13 00:18:32 PDT 
Created attachment 192876 [details]
Patch
Comment 11 Benjamin Poulain 2013-03-13 13:46:23 PDT 
Committed r145748: <http://trac.webkit.org/changeset/145748>