[PlanetCCRMA] root vs capabilities

Fernando Pablo Lopez-Lezcano nando@ccrma.Stanford.EDU
Tue Jun 1 11:23:01 2004

> > > I don't know what that means. Part of me wonders if
> > > something has changed in jackd. I guess it has
> > > undergone changes. Understanding these things is
> > > somewhat beyound me.
> > > 
> > > A couple things that have changed for me since the
> > > rock solid performance days are the use of CCRMA
> > > rather than Gentoo 
> > 
> > Wow, big change! :-)
> To not use CCRMA is to ignore the massive amount work
> that you've done. You'd have to convince that it would
> be wise to go back. :)

Well, the approach of Gentoo is very different. What I meant was that it
was not a small change (changing distros is not my idea of a small
change :-). 

> Really, there's only one
> argument and that is learning to use rpmbuild rather
> then the Gentoo method of building kernels. I'm lazy
> but I should buckle up and enter the ring. Actually,
> I've already done this but the results weren't
> satisfactory. That's another story.
> > > and I changed from an nvidia to an
> > > ATI radeon 9500 pro display card. I've installed the
> > > ATI drivers and it doesn't seem to make much
> > > difference.
> > 
> > I have used radeon cards but not the latest 9500, I
> > tend to go with slower fan-less cards (I think 9200 is the latest I
> > have used). 
> I'm trying to get a job out the door now but maybe by
> tomorrow I can experiment with the old NVIDIA card.
> I've still got it.

I suppose that's one thing to try. 

> > > I suppose there could be something to the switch from
> > > Gentoo to CCRMA but wouldn't the issues be centric to
> > > the kernel and not the distribution?
> > 
> > They _should_ be, of course. Maybe Gentoo has some
> > patches that I don't?
> All I ever patched was LL, the realtime (forget what
> that's called maybe it's something from Robert Love)
> and capabilities.

Ok, that is the same I have. I have a couple other patches (that turned
out to be not that "legal") for xruns with the radeon drm driver for
doing hardware acceleration. Without those patches you will get hits
with radeon if you use hardware acceleration. Maybe that could be the
problem if you are using the drivers and kernel modules from ATI (and
not the ones that come with the kernel), I have never experimented with

> > I don't think so, but it is always a possibility - I am assuming you are
> > using the same or similar release (2.4.26 is the current in Planet
> > CCRMA). It could also be the change in the video card. 
> It's been awhile but I think the version was 2.4.22.
> > Any clues as to when do the xruns happen? Related to some kind of
> > activity? Or just apparently random? What sort of hardware (other than
> > the video card) are you running on? What soundcard?
> The xruns seem to be random but I will experiment a
> little more to have a better sense.
> rme9652; this is the old one not the hdsp
> icp vortex raid controler for audio data

I think I have heard of latency problems in the lists when using SCSI
[UPDATE: just searched a bit and apparently the problems are specific to
the adaptec scsi controllers hogging the bus, maybe your controller does
not have that problem]. It used to be (a long time ago) the best option
but apparently it is not the best any longer - for latency at least. I
don't have pointers to the discussion or posts. Maybe that has happened
recently and the change from 2.4.22 to 2.4.26 could be the culprit. 

> WDC WD400BB-00AUA1 for the OS
> Tiger MPX (S2466) dual capable AMD MoBo with one CPU

I have a few of those around CCRMA. I do remember I needed to flash the
BIOS because of a bug with the Radeon cards (but the computer would not
even POST, much less boot, so that's not your problem). 

The change from two cpus to one could actually be the _real_ change that
has made things less reliable in your case. Maybe two cpus were masking
the SCSI problem I've heard about and with only one it becomes the

> AMD Athlon(tm) MP 2600+
> 1gig of corsair ram
> I have another box in my room that's very similar to
> the one in question--hardware redundancy. The primary
> exceptions are the display and audio cards. It has the
> old NVIDIA display card and an M-Audio 2496. In the
> next day or two, I will rebuild and compare results
> between the two.

Let us know how it works...
-- Fernando