On May 25 2002, Andrew Patrikalakis wrote:
> With all the recent talk of use of assembly on the PowerPC, I came
> up with a patch to use assembly versions of memcpy. It's about 35%
> faster. Here is a sample of the memcpy speed test (which also now
> works):
You're going to die laughing. I beat t
On Sun, May 26, 2002 at 10:58:13AM -0400, Albert D. Cahalan wrote:
> Michel =?ISO-8859- writes:
> > On Sun, 2002-05-26 at 10:54, Rog?rio Brito wrote:
>
> > * Anything more than -O3 seems to be useless, I think Albert even
> > suggested trying with only -O2
>
> I suggest trying -Os as well. Some
Michel =?ISO-8859- writes:
> On Sun, 2002-05-26 at 10:54, Rogério Brito wrote:
> * Anything more than -O3 seems to be useless, I think Albert even
> suggested trying with only -O2
I suggest trying -Os as well. Sometimes it runs fastest,
because smaller code fits in the instruction cache better.
On Sun, 2002-05-26 at 10:54, Rogério Brito wrote:
> BTW, I am using gcc-3.0 to compile xine-libs
Have you tried 3.1?
> and I added some extra options to the configure script
> (-mfused-madd, -mcpu=750, -mtune=750, -O9).
According to info gcc(-3.x):
* -mfused-madd is the defau
BTW. Another place where you can get improved perfs is improving the memcpy's
used to blit the Xv source image to the DMA buffers in R128DMA(). Currently,
if using PCIGART, the target DMA buffers are cacheable. When using AGP,
though,
they aren't. In both cases, using FP registers to do the blit sh
> We can also look into getting an extra version of memcpy that
> makes the transfers with floating point registers as some
> people suggested on the Debian PowerPC mailing list.
>
> People there said that using floating point registers (which
> are 64 bits large) inst
On May 25 2002, Andrew Patrikalakis wrote:
> Hello all,
Hi, Andrew.
> With all the recent talk of use of assembly on the PowerPC, I came
> up with a patch to use assembly versions of memcpy. It's about 35%
> faster. Here is a sample of the memcpy speed test (which also now
> works):
7 matches
Mail list logo
