[FFmpeg-devel] transcoding on nvidia tesla

Mike Melanson mike
Fri Feb 1 01:24:18 CET 2008


Dan Khasis wrote:
> The performance enhancements are 100% real. For example,
> http://www.o0o.it/pro/, shows that a Mac Pro with two dual core xeon
> woodcrests at 2.66ghz crank out no more than 60gigaflops. So a tesla at a
> minimum of 8 teraflops per pc is 130 times more powerful. 

Eh, my new Mac Mini Core 2 Duo 2GHz takes, like, 5 seconds to switch
between 2 running applications. :)

> I'm not claiming this is my unique idea, but I'm not the only one who sees
> the benefits in doing something like this.
> 
> http://forums.nvidia.com/index.php?showtopic=35698&hl=ffmpeg

Interesting, the original poster proposes using it for decoding. Seems
like a pricey and overblown bit of hardware for the task.

> If it [tesla] wasn't that much higher in performance, I'm sure nvidia
> wouldn't have invested tens of millions or more into developing such a
> system, if it wasn't powerful. Also, ATI made an announcement to release
> something similar to tesla in 09.

So you're indicating that perhaps we need to at least lay some
groundwork for these architectural matters that are on the horizon?
Seems reasonable.

> Additionally, everyone knows that GPU's have higher flops than regular
> processors, so it's a natural fit for video transcoding, hence offloading to
> gpu's that some people tried doing.

FLOP = floating point operation. Video codecs generally operate in
integer space. No matter; the hardware can probably do wonders with
integers, and with many more at a time than floats.

I don't dispute that the card has exciting possibilities. But it will
require someone knowledgeable, talented and interested to get ahold of
some hardware before this initiative goes anywhere.

-- 
	-Mike Melanson




More information about the ffmpeg-devel mailing list