🚀 go-pugleaf

I tried to reply by E-mail, but it got bounced somewhere in oz...

In comp.arch you write:

>I would like to find out what happens when you intermix synchronous
>and asynchronous SCSI devices on the one bus? Does the bus drop to
>the lowest common denominator, or does it work at sync speeds to sync
>devices etc.? What happens when both types of devices are doing
>intermixed transfers? Does the speed drop significantly for the sync
>devices? I am thinking here of high speed SCSI disks and slow speed
>cartridge tapes or 8mm tapes. Thanks for any information.

Typically negotions for doing sync mode transfers happen only at boot
time or on opens, since they are relatively expensive (on the order of
5-10 ms).

Since all SCSI transfer are only between pairs of devices, the sync
transfers happen at full sync speed (limited by the negotiated offset
and transfer period, but typically somewhat slower).

The async transfers also happen at their full rated/negotiated speed.

Almost all SCSI chips that can do sync mode transfers switch from sync
to async just by re-programming 1 or 2 registers, so it's not hard (or
expensive) to switch back and forth between modes when both types of
devices are on the same SCSI bus.  So far I haven't run into any SCSI
tapes devices that do sync mode transfers (QIC, 8mm, or 9 track).  I
have heard claims that some of the DAT drives will do sync transfers,
but haven't yet actually seen it demonstrated.

In a mixed environment (sync and async), a major advantage of sync
(aside from higher sustained transfer rates for fast devices) is better
SCSI bus utilization when several devices want to use the bus at the
'same' time.

With the DAT's being advertised as only 185 Kbytes/sec (sustained), and
with large (512Kbyte to 1Mbyte) buffers, the increased bus utilization
would be the only advantage of sync mode; even slow async SCSI devices
have no problems running 185Kbytes/sec over the SCSI bus (as opposed to
sustained transfer rate).
	Dave Olson

It's important to keep an open mind, but not so open
that your brains fall out. -- Stephen A. Kallis, Jr.