On May 21 2002, Ralf Ertzinger wrote:
> Rogério Brito <[EMAIL PROTECTED]> wrote:
> > Anyway, one of the first things I see is that xine uses a
> > function called xine_fast_memcpy, which is an alternative
> > memcpy function possibly written in assembly (if available) or
> > the standard glibc, if no other version is available, as is
> > the case with PPC.
>
> Call me stupid, but I seem to recall that xine is benchmarking the
> kernel memcpy function on x86,
Yes, that is true, for x86.
> so there has to be a mechanism inside xine that connects to the
> kernel function.
No, there is a mechanism for testing generic memcpy's in the
file xine-utils/memcpy.c.
But reading the source of that file, it seems to need a
reimplementation of the possible memcpy's, which includes the
linux kernel memcpy.
Given that the functions are implemented there, the test
occurs, provided that there is a way to benchmark them (which
uses the rdstc instruction on x86; on PPC, something along the
lines of http://cr.yp.to/hardware/ppc.html should be used, I
think).
> Maybe it is possible to use this hook in ppc, too?
I sure hope so. :-)
I don't know if that would improve the situation, but every
tiny bit of extra CPU time seems to count for G3s, I'd guess
(unless some funny, unwanted interaction with the cache
happens).
Please, someone correct me if I am incorrect.
[]s, Roger...
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Rogério Brito - [EMAIL PROTECTED] - http://www.ime.usp.br/~rbrito/
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
