Bug 45270 – [v8] Inline hot methods for V8 to WebCore and back conversions

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[anton muhin]

[v8] Inline hot methods for V8 to WebCore and back conversions

[anton muhin]

This slightly increases the size of binary (< 0.1% for chrome built on Ubuntu for 32-bit platform and ~0.1% of chrome.dll for release, but not official branding), but gives 3--5% of speed up on Windows and up to 8% on 32bit Ubuntu.

[anton muhin]

Created attachment 66651 [details]
Patch

[Adam Barth]

Comment on attachment 66651 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=66651&action=prettypatch

> WebCore/ChangeLog:12
> +        That slightly increases the size of binary (within 0.1% for both Ubuntu
> +        and Windows, but those builds are slightly different from official ones),
> +        but gives performance boost (3--5% on Windows, up to 8% on Ubuntu).
On which benchmark?  Just curious.

[anton muhin]

(In reply to comment #3)
> (From update of attachment 66651 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=66651&action=prettypatch
> 
> > WebCore/ChangeLog:12
> > +        That slightly increases the size of binary (within 0.1% for both Ubuntu
> > +        and Windows, but those builds are slightly different from official ones),
> > +        but gives performance boost (3--5% on Windows, up to 8% on Ubuntu).
> On which benchmark?  Just curious.

Thanks a lot for a quick review on Labour Day, Adam.

I should have mentioned it explicitly, but the benchmark is Dromaeo DOM core.

[WebKit Commit Bot]

Comment on attachment 66651 [details]
Patch

Clearing flags on attachment: 66651

Committed r66840: <http://trac.webkit.org/changeset/66840>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.