Re: Faster encoding?
From: Ariel (Ariel_at_discussions.microsoft.com)
Date: 11/30/04
- Next message: Sohrab: "Fps Concepts"
- Previous message: anonymous_at_discussions.microsoft.com: "Output file cycling in media encoder"
- In reply to: Alessandro Angeli [MVP::DigitalMedia]: "Re: Faster encoding?"
- Messages sorted by: [ date ] [ thread ]
Date: Tue, 30 Nov 2004 03:31:03 -0800
Alessandro,
Thank you very much for the work you put into these replies.
They were quite helpful to me.
Thank you also, Zarax.
Ariel
"Alessandro Angeli [MVP::DigitalMedia]" wrote:
> Ariel wrote:
>
> > How much will dual processors help? Is it OK to have 2
> > processors instead or 4?
>
> There are no benchmarks on any of this (that I know of, at
> least). My guess is that having more than 1 CPU is going to
> help "a lot" since the encoders are multi-threaded as long
> as the number of CPUs is less o equal than the number of
> threads (which should be 6, counting audio and video). How
> much exactly depends on the internals of the encoders and
> nobody outside of MSFT knows for sure.
>
> > Is it better to go with multiple
> > Xeon processors or multiple P4 processors?
>
> I do not have experience with Intel CPUs. I think the cache
> size and speed and the clock speed are the most important
> factors here, unless MSFT optimized their codecs to use
> extended instruction sets (e.g. SSE3) or to depend on
> CPU-specific behavior (like branch prediction).
>
> > Can I do this
> > on a server computer (like dual Xeon 3.5 Ghz running
> > Windows Server 2003)?
>
> The OS doesn't make a difference (of course, it depends on
> the server load, if it's used for other operations at the
> same time).
>
> > I note that the dual xeon motherboards I've seen are
> > 600Mhz FSB, whereas the P4 motherboards are 1.2Ghz FSB.
> > Why? And does this make a big difference for encoding
> > Windows Media Video?
>
> I don't think so: if the on-chip cache is big enough, the
> FSB won't be used much and it will always be faster than the
> pure encoding time.
>
> > Also, if I am starting from source AVI's (not from the
> > DVD's we made from them), does the MPEG-2 decoder make a
> > difference as you mentioned?
>
> It depends on the format the AVI is compressed in: if it
> uses a fast-to-decode format or uncompressed data, it will
> be noticeably faster, but only up to a point, because than
> disk I/O and bus speed will be a limiting factor during
> sample fetching. However, most formats used in AVIs are
> usually way faster to decode than MPEG-2 video and MPEG-1 or
> AC-3 audio.
>
>
> --
>
> // Alessandro Angeli
> // MVP :: Digital Media
> // a dot angeli at psynet dot net
>
>
>
- Next message: Sohrab: "Fps Concepts"
- Previous message: anonymous_at_discussions.microsoft.com: "Output file cycling in media encoder"
- In reply to: Alessandro Angeli [MVP::DigitalMedia]: "Re: Faster encoding?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
