>Douglas A. Gwyn wrote:

>> One significant advantage is that disk size is found
>> using MSCP via querying the device, which allows a
>> variety of actual (e.g. SCSI, which also has a size
>> query) disks to be used. Since PDP-11s rarely had
>> disks over a gigabyte, there is a *huge* supply of
>> used (SCSI) disks available very cheaply that are
>> as large as one could actually employ (due to OS or
>> filesystem structure limitations).
>>
>> I don't see any need to involve a whole PC in the
>> disk subsystem. There is about as much work needed
>> to interface e.g. MSCP Qbus to a PCI interface with
>> software to get at the PC's disk as there is to
>> interface MSCP to SCSI directly. As somebody else
>> observed, a C implementation of MSCP already exists
>> in SIMH, and I'm sure there are C implementations
>> of drivers for various common SCSI chips (probably
>> in Linux, certainly in Plan 9). Therefore the
>> hardware boils down to a Qbus (or Unibus) board with
>> the necessary interface and control logic, a SCSI
>> chip with a bit of support logic, paths between the
>> two, and a small microprocessor with its RAM and ROM
>> to act as central control. If my employer cared
>> about such things, I could develop one and run off
>> a modest quantity of boards; my *guess* is that it
>> would run about $400 apiece (populated). However,
>> I'm sure that any time I spent on this would work
>> against me with regard to my employer.


>I am running such a project. See http://www.geocities.com/mscpscsi
>I am making progress, though slowly.
>vax, 9000


Ok, I got the first successful test run. There are still bugs though.
and the haredware and software still need to be cleaned.

vax, 9000