On Fri, Jul 09, 1999 at 01:37:35PM -0600, Victor Yodaiken wrote:
> On Fri, Jul 09, 1999 at 08:41:58PM +0200, Thomas Wuensche wrote:
> > That may depend on the hardware and PCI bridge configuration,
> > but from experience with our ISA CAN boards I would expect the
> > time to be in the range of 1-3us. We achieve complete cycles
> > with synchronisation to the processor on the board and more
> > than one ISA access in about 20us.
>
>
> Worst case? Dig out an old ISA ethernet card and try the test with a
> "ping -f" running in the background.
>
Or, read/write to a floppy. DMA cycles on an ISA bus, especially
on an 8 bit bus, take a *long* time. If you have anything that uses
DMA, you can kiss your ISA latency goodbye. If you disable ISA DMA,
you *should* be able to do better than 20 us.
Does anyone have experience with a decent, cheap, and fast PCI
timer card? For decent, I'm thinking of one with a timer reload
register, or (better), one that can trigger when the counter
reaches a specified value, and keep counting. Atomic reads/updates
of registers is a must. If a card like this can be obtained for
USD ~100-150, I think that would be a decent alternative to the
8254. I'll write a driver...
(Actually, now that I think about it, the 3com 3c905b has such a
counter, and is ~USD 60. What a bargain. What a hack. But hey,
if it works...)
dave...
--- [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/
