4 Feb
2013
4 Feb
'13
3:24 p.m.
On 2/4/2013 1:46 PM, Rich Alderson wrote:
From: Jeff Jonas
Sent: Friday, February 01, 2013 6:43 PM
CMD now Silicon Imaging, who have made a lot off of a PCI to IDE chip
had a CI disk subsystem which hooked to Vax systems. The chip they used
escapes me, but it was a SCSI chip of some sort that was persuaded to
parse the info coming from the VAX. I believe they dispatched action in
the controller within a couple of their clocks from having the complete
command received from the VAX. The DEC hardware didn't stand a chance
performance wise, as it had to read the command an chew on it for a
while in some processor before anything happened. If the Cache in the
CMD subsystem had the data it was turned around from the buffer starting
practically a soon as the command byte went out.
Jim
Long ago when SCSI was young
and just an 8 bit parallel bus,
Ampro littleboards advocated using the SCSI bus
for peer to peer communications, not just master/slave.
Did anyone else do that?
At XKL, I once mentioned the use of CI-based disks on
TOPS-20 (major
feature of v6.0) to one of the SCSI card engineers for the Toad-1.
He was intrigued by the idea, read the docs and the TOPS-20 source
code for the feature, and implemented MSCP on the FASTWIDE SCSI bus.
It never made it past the experimental stage (unfortunately, in my
opinion).
Computer Interconnect (CI) was a 50Mbit ethernet[1] designed for the
Jupiter (follow-on to the DEC-20, to be called the DECSYSTEM-4050),
and later retrofitted to the VAX, so I would think that our little
experiment qualifies as an edge case. :-)
[1] Lower case is intentional.
Rich Alderson
Vintage Computing Sr. Systems Engineer
Vulcan, Inc.
505 5th Avenue S, Suite 900
Seattle, WA 98104
mailto:RichA at vulcan.com
mailto:RichA at LivingComputerMuseum.org
http://www.LivingComputerMuseum.org/