I haven't tried 2.6 ..yet, but I was under the impression that you had to
enable a bunch of stuff in the kernel-configuration before you got good
performance. I've got no real basis for this assumption, apart from that it
used to be that way with the old kernel...
I've heard "someone", can't remember who, report very good performance with
2.6, though sadly that's pretty much all I've heard of it so it's hardly
Saturday 15 November 2003 10.24 skrev Thorsten Mika:
> Hi all,
> having used a 2.6.0-test9-mm3 kernel and enjoying the benefits of the
> pre-emption thing (everything seems _much_ more reactive than before), I
> ended up in returning to my old 2.4.21 because of 2.6's extremely poor
> realtime performance. It is unfortunately completely unusable for
> low-latency audio work.
> Does anyone have an idea whether this is going to change or whether
> there are already RT-patches around? I am not a coder, so all I did was
> perform the suggested changes (on the JACKD site) in capability.h, but
> this is of course not enough.
> I searched the net, but apart from 2 threads on LAD there does not seem
> to be any valuable information around (especially regarding plans of
> implementation etc).
> Thank you,