Brian Buhrow writes:
> 
> 	Bus Logic 445s users have always used the bt742a.c SCSI driver.  Although
> the aha1542.c driver would theoretically work, the bt742a.c driver has
> always provided better through put on the SCSI devices and works well with
> machines with more than 16MB of memory.
> 	Because there are different revisions of the 445s board out there that
> work in slightly different ways, I would suggest leaving all drivers in a
> state that would cause them to work with as many boards as possible.  Let
> the users decide whether or not they want to use the aha1542 driver or the
> bt1742 driver.
> 	Perhaps Dave Burgess can put a note to the effect that users of the 
> Bus Logic 445s card have the option of using the bt742a driver or the
> aha1542 driver depending on which ROM and board rev they have.  I, for
> example, use the bt742a driver with my Bus Logic 445s with ROM rev 3.37 and
> the dip switch 2/10 in the off position and find that the card is
> blindingly fast.  I would not like to lose my speed in the act of upgrading
> to the current kernel.


I agree whole heartedly.  I've been using the bt742a driver with my 445S for 
over a year and it would be a shame to suddenly break the driver on the 445.

I have 24MB on my machine and until NetBSD incorporates bounce-buffers (has
it?) I would be unable to use the aha1542 driver anyway.

For what it's worth, I'm using a June 2 current kernel right now with a 445
and it's working fine, so I don't know what all the fuss is about, but
I would probably move on from NetBSD if I were suddenly forced to sacrifice
performance that I've enjoyed since day one.


-dwh-
dhess@cs.stanford.edu

------------------------------------------------------------------------------