WTB: Emulex MT-02 controller
cclist at sydex.com
Thu Apr 24 16:54:54 CDT 2008
Date: Thu, 24 Apr 2008 15:47:39 -0500
From: Jules Richardson
> Hmm, maybe the designers were sensible, and any critical timings are
> solely on the device response side - i.e. the host can take as long as it
> wants to process things, but the device must respond within a certain time
> period (and then just happily sits there until the host acknowledges).
QIC-02 is buffered via drive-local RAM, so it's not really fussy on
the host-side timing. These tapes are streamers, however, so you
want to keep the tape moving; otherwise, it's back-up-a-piece-and-
take-a-running-start-at-reading-the-next-block. Really slow--and
hard on the tape.
> Of course none of this probably helps Chuck, who quite possibly doesn't
> want to be messing with designing hardware and writing driver software :-)
> It's just nice to speculate that it could be done once QIC interface
> boards are unobtainable...
Oh, Chuck has both QIC-36 and QIC-02 ISA inteface boards. He just
wants the convenience of not dealing with buggy NetBSD drivers (the
Linux QIC-02 drivers don't work at all anymore) and not having to
look for spare DMA and IRQs for the boards. A QIC-02 host interface
board is scarcely more complicated than an IDE interface. OTOH, a
QIC-36 interface is usually a full-length board, packed with a data
separator, CPU, RAM and other logic. From the CPU side, both appear
as pretty much the same thing.
More information about the cctalk