Subject: Re: [linux-audio-user] xf86config no_accel = yes but sucks bad
From: R Parker (rtp405_AT_yahoo.com)
Date: Wed Sep 29 2004 - 19:24:58 EEST
--- Dave Robillard <drobilla_AT_connect.carleton.ca>
wrote:
> On Wed, 2004-09-29 at 09:51, R Parker wrote:
> > Hey,
> >
> > In the ongoing attempt to eliminate xruns, I've
> set
> > Option "no_accel" "yes" in the XF86Config.
> [snip]
> >
> > Or is there another method for a user to turn
> > acceleration on and off?
> >
> > Is anyone else turning off acceleration and how do
> you
> > deal with the performance?
> >
> > ron
>
> I don't think turning off acceleration is going to
> do anything
> beneficial for you - it could actually be
> counterproductive since X is
> going to be eating a /lot/ more CPU.
>
> You might want to turn off DRI direct rendering
> though, as that will
> free up an interrupt.
Thanks for the suggestions. Acceleration is back on
and DRI is off, oh, what a relief it is.
I suspect my system has a more serious problem and
that I am essentially tweaking the leaves while the
roots are infected.
I'm running an ICP hardware raid controler with raid
level 5 and reiserfs. In addition to that, I run an
rsync mirror to an IDE drive. I'm considering
switching from raid 5 to 0 because I do see read
problems and maybe some of the xruns occur during
write. The read problems appear with Ardour while
doing REW or F/FWD seeks and the messages are
something about disk read access error.
I've tried to google up some hints for performance
tuning with hardware raid but haven't found any
obvious suggestions. With the mirror running every 24
hours, I can probably afford to forgo the hardware
redundancy.
Maybe someone has some insites into hardware scsi raid
performance. I'd be happy to investigate any
suggestions.
ron
> -DR-
>
>
__________________________________
Do you Yahoo!?
Yahoo! Mail - You care about security. So do we.
http://promotions.yahoo.com/new_mail
This archive was generated by hypermail 2b28 : Wed Sep 29 2004 - 19:28:17 EEST