[mythtv] Re: [committed] denoise3d "black spots" fix

[Andrew Mahone]

On Thu, 29 Jan 2004 21:54:59 -0500, "D Banerjee" <davatar at comcast.net>
said:
> Well, the only thing I can see is that you're not using an asm
> volatile. I'm not sure what gcc will do with that code, since if you
> don't indicate volatile it's free to move it around if it feels it can
> make an optimization. That may be your problem. Also why I stuck to
> doing single blocks of asm in quickdnr, mmx is already hand optimized,
> I want the compiler to do as little as possible to it.

This is *definitely* not the issue.  If you read mmx.h, where the macros
I used here are defined, each MMX instruction ends up in its own __asm__
__volatile__.

I don't think the MMX has anything to do with it.  I wrote debug code
that used C math to duplicate the operations which produce the values in
wbuf, and those are identical.  And if I use a C loop to duplicate the
lookup table operation, and store the values into a test array, those
values are not the same ones in cbuf.

> This revision seems to be much better, no spots, lines or other
> artifacts visible.

Good, I couldn't see the lines from the start, but those B&W cards
between shows during Adult Swim highlight the black spots bug quite well,
and got a bit irritating to look at.
-- 
  Andrew Mahone
  andrewmahone AT eml DOT cc

-- 
http://www.fastmail.fm - Email service worth paying for. Try it for free