[Webkit-unassigned] [Bug 73789] Need SSE optimization for SincResampler::Process()

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=73789





--- Comment #6 from xingnan.wang at intel.com  2011-12-04 22:49:17 PST ---
Thanks for your fast response and comments.
(In reply to comment #3)
> (From update of attachment 117838 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=117838&action=review
> 
> > Source/WebCore/platform/audio/SincResampler.cpp:-249
> > -            // FIXME: add SIMD optimizations for the following. The scalar code-path can probably also be optimized better.
> 
> This could be changed to
> // FIXME: add ARM NEON optimizations for the following. The scalar code-path can probably also be optimized better.
> :)
> 
OK, I`ll add it.
> > Source/WebCore/platform/audio/SincResampler.cpp:262
> > +                input = *inputP++;
> > +                sum1 += input * *k1;
> > +                sum2 += input * *k2;
> > +                ++k1;
> > +                ++k2;
> 
> You should put your code under the #define CONVOLVE_ONE_SAMPLE and use the macro here.
> 
All right.
> > Source/WebCore/platform/audio/SincResampler.cpp:267
> > +            int group = n / 4;
> 
> Typically you can just compute the end pointer so you do not have to keep track of that.
> 
I don`t quite understand here, do you mean we just use end pointer to count the loop? 
> > Source/WebCore/platform/audio/SincResampler.cpp:355
> > +                input = *inputP++;
> > +                sum1 += input * *k1;
> > +                sum2 += input * *k2;
> > +                ++k1;
> > +                ++k2;
> 
> Same regarding CONVOLVE_ONE_SAMPLE.
All right.

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