> > This differs from my prev. version: I no longer do usleep(1) in tight
> > loop. Instead, I know that ISA read will cost at least 0.125usec (is
> > there better estimate available?). Please, roll this in.
>
> Applied.
Thanx.
> Btw, as far as I know, an ISA read will cost closer to 1 usec rather than
> 1/8 usec.
>
> Yes, the ISA bus is 8MHz, but when you actually do a read that's not just
> one bus cycle, even in theory it's at least two cycles (one to assert the
> address, one to get the result), and in practice it seems to be noticeably
> more.
Hmm, seems that my machine gets 3.5MB/sec from this 8bit card, so
access can not take much more than 2 ISA cycles. 2 cycles / read mean
that I can make timeout 2 times less. I'll test it a bit, expect tiny
patch somewhere later. Fortunately, timeouts are not too critical in
this code.
> Nitpickers will add to that the subtractive decoding of PCI->ISA,
Unfortunately there are machines without PCI...
Pavel
-- I'm really pavel@atrey.karlin.mff.cuni.cz. Pavel Look at http://atrey.karlin.mff.cuni.cz/~pavel/ ;-).- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html