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

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Mon Feb 27 09:16:00 PST 2012


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





--- Comment #51 from Raymond Toy <rtoy at chromium.org>  2012-02-27 09:16:00 PST ---
(In reply to comment #50)
> (In reply to comment #49)
> > (In reply to comment #48)
> > > (In reply to comment #45)
> > > > (In reply to comment #42)
> > > > > (In reply to comment #41)
> > > > > > (From update of attachment 127768 [details] [details] [details] [details] [details] [details])
> > > > > > View in context: https://bugs.webkit.org/attachment.cgi?id=127768&action=review
> > > > > > 
> > > > > > I think the commit queue bot needs to have the ChangeLog include the "Reviewed by..." line.  Please re-upload patch with this fix.
> > > > > > 
> > > > > > Xingnan, I'm assuming you've run the layout tests locally on your linux box with IPP enabled and that they passed.  Also, please verify that this page works with the IPP code running:
> > > > > > http://chromium.googlecode.com/svn/trunk/samples/audio/mag-phase.html
> > > > > > 
> > > > > Chris, I will re-run the latest layout tests and verify the demo.
> > > > Layout tests are OK.
> > > > The demo looks generally good, but I found there is deviation that the magnitude mismatched in lowpass.
> > > 
> > > This doesn't make any sense to me.  The frequency response should not depend on whether we are using IPP or not.  The only thing I can think of is that the filter coefficients are somehow slightly different between the two graphs.  The displayed values for cutoff, Q, and gain are not printed to full precision.
> > 
> > Wow, that's really interesting.  Can you try to debug a little bit more and maybe add some "printf" statements to determine what the calculated filter coefficients are in the two cases?
> > 
> > How do the filters sound when filtering the noise?
> 
> Chris and Ray,
> I think the reason is I compared the graphs in different version.
> One is the latest dev-version of chromium with IPP, another is released chrome(17.0.963.56) without IPP. 
> 
> I double checked the latest dev-version of chromium, the output with IPP and without IPP from the demo are the same, without the deviation.
> 
> So sorry for noising you.
> 
> BTW, what`s the reason of the difference of the filter output between different version?

Good question.  The only thing I can think of is that the function for computing the magnitude and phase had an issue with computing the response as the filter was being changed. I doubt this is the case here.

Since the responses are now similar, between a recent version of chromium with and without IPP, I'm satisfied.

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



More information about the webkit-unassigned mailing list