[FFmpeg-devel] [PATCH] SPARC VIS simple_idct

[Balatoni Denes]

Hi!

Saturday 25 August 2007 13:52-kor Balatoni Denes ezt ?rta:
> Saturday 25 August 2007 08:02-kor Michel Lespinasse ezt ?rta:
> > Don't know if there is any interest or how it compares with the
> > simple-idct derived code - but, it is (barely) IEEE1180 compliant and
> > does not use 32-bit multiplies.
>
> I glanced over it and speed wise it should be roughly the same as my
> simple_idct_vis (about the same amount of operations).
> But unfortunatelly I see a problem: you are using unsigned multiplies,
> which are AFAIK not available on SPARC. This also means that the code might
> not actually comply with ieee1180, because you are using the sign bit for
> data, but you can't.

I forgot to say thanks for digging up this idct Walken, btw.

Also I realized, that with some performance hit simple_idct_vis could be made 
more accurate, by using the trick of X*Y=(X-1)*Y+Y where 0.5 <= X <= 1. I 
will check that next week. I don't know if overflow would become a problem, 
though.

bye
Denes