John Foust wrote:
> I asked a friend at Astronautics about the ZS-1, and he said:
>
> > I talked to a guy who has used the one we used to have here. He says for
> > scalar operations, it's as good as a 386. For purely vector operations, it
> > might be equivalent to a 1 GHz Pentium.
> > If you are looking for scrap value, it has 60 amp 5 volt power supplies and
> > old SCSI hard drives.
He neglected why this is a historically significant machine. Among other things,
it's an early example of dynamic instruction scheduling, decoupled access-
execute and distinct float and integer pipes. It's also a pseudo vector machine.
All in all, it's an impressive hunk of technology for its vintage (~1988, I
believe). The DAE and dynamic scheduling aspects are particularly interesting;
dynamic scheduling generally produces better performance that the more typical
static scheduling, and the hardware queues used to support both the DAE and
DS aspects of the architecture illustrate an alternative to the more
typical register renaming seen in superscalar architectures.
> > He also said you might have trouble with your electric bill.
He's probably not wrong. I believe the thing is implemented in ECL.
If I could figure out how to get one of these systems from there to
here I'd jump at the chance. These things belong in a museum, not
a scrap heap.
--
Chris Kennedy
chris_at_mainecoon.com
http://www.mainecoon.com
PGP fingerprint: 4E99 10B6 7253 B048 6685 6CBC 55E1 20A3 108D AB97
Received on Tue Jan 08 2002 - 09:41:10 GMT