Subject: Re: Oooooh... IDE to Q-bus/Unibus adapters.
To: None <port-vax@NetBSD.ORG>
From: John Wilson <firstname.lastname@example.org>
Date: 07/08/1999 02:34:03
>From: Thor Lancelot Simon <email@example.com>
>Okay, but you're not going to find a DMA IDE controller that's not PCI. Do
>you see why I was going on about how you ought to be drawing the comparison
>to cheap _PCI_ SCSI adapters?
I haven't seen the specs for the more recent IDE modes but if it's the same as
it used to be only with shorter cycles, you're giving the protocol entirely
too much credit. In fact "controller" is an awfully strong word for what
you need to talk to an IDE drive, building one for Q-bus out of loose TTL
would be straightforward, actually the hard part is just making sure that's
what you want -- for a complicated protocol like MSCP you're probably going
to be better off bouncing everything through an on-board micro so it can take
care of what goes where (and the Q-bus DMA logic can be used to do other
things while disk I/O is going on).
ANYWAY, if we're talking about a MV2/MV3/PDP-11 Q-bus controller, then this
is a totally pointless flame war. The throughput of the Q-bus is a tiny
fraction of what *any* modern hard drive is capable of, so that's where the
limit is going to be. Dunno how much rosier the picture is on the desktop
PCI is a total nightmare. Bridging it to Q-bus just to borrow some multi-I/O
chip that probably will be discontinued next month, doesn't sound like it
would be worth it. The only interesting thing about such chips is that they
do bus-mastering DMA, which is still a pretty novel idea on PCs but it's
the *only* way that Q-bus has ever done DMA and of course the details are
all different so it's no help. Q-bus disk controllers have made it this
far without scatter/gather capability so I think we can continue to live
>From: firstname.lastname@example.org (Allison J Parent)
>Neither have I. I think ISA is not fast enough to support the faster USB
>rates. I'd think Qbus would have problems too.
I guess it depends on traffic, USB is only slightly faster than Ethernet and a
lot of what goes over the wire is bureaucratic crap so it really might not be
such a load on the Q-bus assuming you've got an on-board micro taking care of
the busywork. ScanLogic makes a chip called the "SL11H" (hope I'm remembering
that right) which is supposed to be an easy-to-interface USB controller, I've
never worked with it though so I don't know whether it truly supports the
whole protocol or whether it's just designed for specific applications only.
I'm way behind on learning about USB, I got their regular SL11 peripheral chip
(boy they sure are jerks about samples) and stuck one on an ISA board just to
play with it eons ago and *still* haven't written code to make it do anything.
Using something like that to make a PDP-11 act as a USB *peripheral* might
be useful too though...