Hi Mark and thanks for an extensive answer.

I've read parts of Jan's paper, a lot of good info, very interesting your parts regarding filesystem. I wouldn't have guessed that ReisterFS would have been the least intrusive of the tested filesystems!

My current system is an Athlon XP 1700+, MSI 266(something) m/b with 256MB ram and two ide disks 60gb,80gb. Using an sb-live and an Delta44 (envy24) soundcard.
I've only recently starting to trim it for lowlatency for _real_. And this system needs trimming. By taking your advice of shuffling the actual cards and also trying different settings in the bios has made noticable difference wrt latency.

Right now I'm stuck on the freezing kernel problem that Fernando has been reporting about. Since he has posted a patch today(yesterday) I'm in the process of building it, hopefully the problems will disappear and latency is in the right ballpark :)

WRT a previous comment about not using Jack at all, how would I use 5
soft synths, Ardour and Rosegarden at the same time without it, and only
using Alsa? Even if I didn't require low latency which I do) how would I do
that? I'm not clear?

For a pro/semipro this is probably not an option, unless you are content working with just one program at the time, with MusE this works pretty good since MusE has internal softsynths, but they aren't that many.
Also, there are cards, e.g the SB-Live/Audigy that supports multiple open of the output port. I think it supports up to 64 simultaneous streams that then will be mixed on the card. But I don't think SB-Live/Audigy would suffice for a pro/semipro user. (it's ok for me, but I don't do much serious work). Mixdown is a real pain in this situation, what you do is that you set the soundcards mixer to record from the output stream.
This would not work for jack based applications like ardour, then you'd need jack anyway. I guess you could run multiple instances of jack, one for each app, if you really wanted to. But then you would be missing the point by a few thousand square kilometers.

Jack is a much cleaner solution that has the possibility to really bring Linux audio together and sustain it for a very long time!
That is, if I ever succeed with running it without xruns ;-)

Regards
/Robert


Reply via email to