[Webkit-unassigned] [Bug 77509] Enable IPP for Biquad filter

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=77509





--- Comment #11 from Raymond Toy <rtoy at chromium.org>  2012-02-02 09:24:49 PST ---
(In reply to comment #7)
> (In reply to comment #5)
> > (In reply to comment #1)
> > > Created an attachment (id=124881)
 --> (https://bugs.webkit.org/attachment.cgi?id=124881&action=review) [details] [details] [details]
> > > Patch
> > > 
> > > From my test in linux, the performance increased ~27% for the processing of biquad.
> > 
> > To clarify, this gives 27% increase over the original code.  And the other version in bug 75528 (https://bugs.webkit.org/show_bug.cgi?id=75528) that uses SSE2 gives about 20%?
> 
> These two patches are different. I just optimized the multiply-add in biquad process in bug 75528, but in bug 77509 I used the whole IIR filter in IPP to implement the biquad, in which I guess highly optimized method may be used so it achieves a better performance, not being sure because I could not touch the IPP source code either.
> 
> These two patches also are used in different scenarios. If IPP is enable, we could use IIR filter in IPP. Or we could use the optimization in bug 75528 for common case as long as the platform supports SSE2.

Thanks for clarifying this.  I understand speed difference now.

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