[comp.sys.sun] "SMD is dying"

dan@wind.bellcore.com (Daniel Strick) (10/02/89)

I think it is more correct to say that SMD is in its "Golden Years".

SMD performance has probably reached its limit, but this is a recent event
and further refinements (new controller designs) are still possible (if
not economically exciting).  Many SMD controllers are now "mature"
products (you can buy one with reasonable assurance that it will work
correctly on your workstation and with almost any SMD drive and will give
expected performance (though it is important to know what you are doing
when you try to configure a strange drive).

SCSI performance will probably surpass SMD performance in a few years, but
SMD still holds an edge because the SCSI protocols are expensive, because
SCSI devices tend to be designed for the low end PC type systems (are
cheaply contstructed), and because current microcomputer systems are
mostly not designed to take full advantage of SCSI distributed smarts and
parallel data transfer.  New versions of SCSI and modern SCSI interface
chips will make SCSI faster (eventually).

IPI-2 will probably start out a little bit faster than SMD, partly due to
the wider bus, but mainly because the new IPI-2 controllers are designed
with faster processors (simply because they are newer).  I expect IPI-2 to
become much faster than SMD, but IPI-2 is also not quite here yet.

SMD is still king.  (but you know what happens to kings these days)

Dan Strick, Bellcore, dan@bellcore.com, bellcore!dan, (201)829-4624