On Sat, 2007-02-24 at 16:55 +0000, Paul wrote:
> On Saturday 24 February 2007 15:56, Philippe Gerum wrote:
> > > Initial results look promising with the latency test reporting figures
> > > in the 0.5 to 4 uSec range under light load. A kernel compile bumps the
> > > wost up to 9-14uSec.
> > Not bad. Is this a UP or SMP config, and which kind of box did run the
> > tests precisely?
> AMD X2 4600+ with SMP config - MoBo is an Asus A8V with a Via K8T800Pro
> chipset and a measly 512M of RAM.
> > > Firing up X,KDE, and a gamut of desktop apps really kills the
> > > numbers though - ~267uSec being the worst recorded so far,
> > Ok, the latter looks clearly pathological. The trace should tell us
> > more.
> The latency log ?
Rather the CONFIG_IPIPE_TRACE feature, with the IRQs off tracking option
set, but the latency log already has some value, thanks.
> Attached is some three hours worth of log - Should point out there are no
> proprietory video drivers being used, just the regular Xorg stuff and what
> ever the kernel provides.
Looking at this log, the good news is that latency peaks are not spread
all over the place, but only a few of them are bugging us. This tend to
confirm a bad interference exists with something specific to the X
startup procedure (maybe/likely? related to the graphic card). Does the
peak still occur if you disable hw acceleration for the graphic card in
your XF86Config file?
> Regards, Paul.
Xenomai-core mailing list