Random PC musings...
Barry Shein
First, any reactions to the Apple "AI Workstation" (that means it runs
lisp) announcement this week? Maybe someone can at least summarize
what the announcement was about.
I was at Usenix a couple of weeks ago and stopped by Uniforum for an
afternoon. I was amazed at the number of workstations or
workstation-like (eg. small towers with a few boards, but no screens)
systems there are now. There were dozens of respectable systems on the
floor. The range seems to be from about 1MIP to 15..20MIPs (everyone
heard about Apollo's announcement, sounds impressive!)
I was looking at one company running X-windows on a 19" 1200x1600
monochrome screen on a 386 (using a co-processor to do the graphics.)
Also very impressive and quite inexpensive (a useable system like
that is probably $6K-ish.)
We keep hearing rumors about Sun's 386 system. I am beginning to
wonder if we're about to see a massive explosion of 386 Unix systems
doing serious work, they can certainly deliver the price-performance.
Unix hides the architectural details well enough that I suspect those
who dislike the '86 architecture will give in on this, besides the 386
helps hide most of the earlier problems at the hardware level. Of
course, new introductions at the low-end based on Sun's SPARC (from
AT&T?) and Motorola's 68030 (and/or Motorola's to-be-announced RISC)
could offer interesting alternatives. I still say the fate of Unix
remains in the hands of the disk manufacturers and it looks like
they're making good on their promises, fast 100MB winchesters are
finally coming down into the less than $1K range.
Conversely, I will make the bold prediction that the Mac has peaked,
the novelty is wearing off and its non-standard environment and lack
of things like networking (regardless of all the attempts to make it
seem otherwise) will start to push people towards these standard
systems. The Mac/OS will become Apple's VMS and they will start to
struggle internally (just like DEC has) about what to do with Unix.
I expect to see MicroSoft put most of their software products into the
Unix system in the next year, with others to inevitably follow.
Products like NeXT will probably make all this even more clear (if
NeXT is ever released, I don't know enough about it to really predict
whether it will be a success, it's not clear Steve Jobs does
either...)
Where does all this leave OS/2 (half an operating system)? It will
enjoy some popularity no doubt, but I hardly expect anything exciting
or interesting to come out of it. There will always be a market for
grey-flannel computing.
-Barry Shein, Boston University
David Barto
>Unix hides the architectural details well enough that I suspect those
>who dislike the '86 architecture will give in on this, besides the 386
>helps hide most of the earlier problems at the hardware level.
I agree. I just started looking for a system and the 386 based systems
are acceptable. However I have yet to find a system which is reasonable
in price ($3K) which does not require me to purchase a bound in keyboard
and expect a special monitor. For this reason I am still looking.
(If anyone finds one which will take a standard CRT as the
console, and runs *nix please mail me.)
On the mainframe side you mention that you can purchase a small box (or
collection of boxes) which will fill the need of large groups without
constraining those groups to a single vendor (read IBM or DEC). My
father-in-law, president of a medium sized corporation in San Diego,
changed his entire computing environment from a few large mainframes to
LOTS of Suns and (Celerity) Accels and other reasonable priced
processors, with 1 to a few people per machine.
The entire computer center quit (or were fired) over the change, and
most of the scientists griped that 'they' were losing control of
'their' computing environment. Now, after several months of the new
environment, I doubt that they would go back to the centralized
approach.
I think that those people who care about the future of SCIENTIFIC
computing, are aware of the changes and are doing as the above
company has done. Those doing the more mundain day-to-day computing
of large databases will continue doing it 'the way we have always
done it -- The IBM way'.
-barto
--
David Barto sdcsvax!sdcc6--\
ba...@net1.ucsd.edu seismo-!s3sun--!megatek!barto