[Webkit-unassigned] [Bug 94810] Vector::shrinkToFit should use realloc when canCopyWithMemcpy is true

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=94810


Yong Li <yoli at rim.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #161042|0                           |1
        is obsolete|                            |
 Attachment #161235|                            |review?
               Flag|                            |




--- Comment #25 from Yong Li <yoli at rim.com>  2012-08-29 08:37:19 PST ---
Created an attachment (id=161235)
 --> (https://bugs.webkit.org/attachment.cgi?id=161235&action=review)
Updated

I changed the number "10x" because it is actually an estimate after deducting the extra malloc/free in every cycle. Also I made a mistake that it was on 8K->4K shrinking. To be clear, for 2K-1K shrinking, the time it spends given by the test reduces from 100% to 60-70%, For 8k-4k shrinking, the elapsed time reduces from 100% to ~55%.. I'm not sure how to calculate this performance boost percentage. Now I'm using this formula: (<old time cost> - <new time cost>) / <old time cost>. Then this cannot explain why we see "2x times faster" in some logs...

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.