On Mon, 29 May 2006 02:46:38 +0200 Marschall Kurt <[EMAIL PROTECTED]> wrote:
> Hi, > > I was a bit too much enthusiastic about this speed thing. > The resizing and clipping options of course don't have the same purpose. > But generally -B is faster than -X (shrinking is faster than expanding). Duh, looks like I should read the whole thread before to answer :P > I would love too know if there is a way to speed up the resizing thing or > the deinterlacing. I get totaly insane with the slowness of the encoding > (And I have a amd64 Athlon 3200+). I encode with xvid4 (with speed > optimisations for amd64), deinterlace option 3 (the slowest) and lots of > resizing as you could see in the previous message. I don't get more than 13 > fps. Without all the resizing and deinterlacing stuff I get 31 fps. > With the resizing and without deinterlacing i get 19 fps. They are less or more my numbers (I have an athlon64 3200+ too). As general advice, the idea is to do the lower amount of processing that's needed, because slowdown isn't linear when adding processing operations. > How can I speed up transcode. Good question, but a proper answer will involve a fair amount of architectural work and a bunch of asm optimizations :) A more feasible path it's to move, if wasn't already done, all clipping operations as early as is possible, in order to reduce as much as is possible the frame region to be further elaborated. (I.E.: prefer -j over -Y where feasible). Best regards, -- Francesco Romani - Ikitt ['people always complain, no matther what you do'] IM contact: (email-me, I have antispam default deny!) icq://27-83-87-867 some known bugs: http://www.transcoding.org/cgi-bin/transcode?Bug_Showcase
