The so called "illness" is more related to the PCI-bus scheduler.
It is the job of the scheduler to find a clever algorithm for that traffic.
Unfortunately the default every PC comes with, is to do best
benchmarks not to do something clever or usefull :(
If you have the chipset register docs it is nearly easy to trim this :)
!!avoid this if possible, it has to be adapted to every new chipset!!
Jens Michaelsen
Nachricht geschrieben von Guilherme Nelson F De Souza
> Anyway, an ill-behaved PCI card (with a wrong choice of latency
and under a heavily loaded demand) can be really bad for any application
(real time or not). I've read about some "pathological" cases where a
card hold the bus (holding the CPU, other PCI cards, etc) forever by
taking too long to release the bus (larger latency) and getting back
in the end of the queue for another transfer before any other card
can gain access.
<
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/
