On Sun, 08 Feb 2004 11:46:04 +0100
Gabriel Bouvigne <[EMAIL PROTECTED]> wrote:

> > Side note: I don't _expect_ a performance gain in using a AMD64 compiled
> > version. Switching from 32bit to 64bit results in a larger executable
> > (as pointers grow from 32 to 64 bits). This may result in a _slower_
> > executable, as the resulting code may not fit into the CPU cache
> > anymore. The cache would play ping-pong in this case (switching between
> > parts of the code). I've read about a performance loss of 20-30% for
> > some specific applications (but with sparc64, not amd64).
> 
> But on x86, we might benefit from the increased number of visible registers.

Yes, this could help here.

> > If someone compares the performance between an ia32 and amd64 compiled
> > version of LAME, I would be interested to know the actual numbers.
> 
> http://anandtech.com/cpu/showdoc.html?i=1884&p=17

Woohoo! Very nice! Ok, obviously the increased number of visible
registers helps alot.

> Moreover, we do not have any amd64 computer. I think that this is a 
> major blocker.

I've access to the FreeBSD amd64 machine (btw. it builds lame 3.95.1
just fine: no error logs on http://bento.freebsd.org/ for lame), but
this doesn't help Fred.

Bye,
Alexander.

-- 
           I will be available to get hired in April 2004.

http://www.Leidinger.net                       Alexander @ Leidinger.net
  GPG fingerprint = C518 BC70 E67F 143F BE91  3365 79E2 9C60 B006 3FE7
_______________________________________________
mp3encoder mailing list
[EMAIL PROTECTED]
http://minnie.tuhs.org/mailman/listinfo/mp3encoder

Reply via email to