The Petaflops Boondoggle Computer (was PET_ard)

["Timothy C. May" <tcmay@got.net>]

(Hoist by their own petards indeed! Don't tell our Russian what petard means.)

At 1:16 PM +0000 9/29/96, John Young wrote:
>   Sci, 20 Sept 1996:
>
>   "Redefining the Supercomputer"
>
>      The word is petaflops, computer jargon for 1000
>      trillion computations per second. Think of it as a
>      year's labor for a powerful workstation compressed
>      into 30 seconds. Think of it, also, as 1000 times the
>      speed of the current computing benchmark, a trillion
>      operations a second --  teraflops -- which is on the
...

I doubt this will be ever be built, at least not as a government-funded
"G-job" "one-off" machine. It would, as the full article state, necessitate
a kind of "Apollo program" for supercomputers.

This, as funding for mega-projects fades. This, as Cray Computer went
bankrupt, as Thinking Machines went into Chapter 11 and only recently
emerged as a pale shadow of its former self (concentrating on software
only), and as Floating Point Systems, NCube, MasPar, etc. are foundering.
(Actually, some have already been absorbed into other companies, and in
many cases, dissolved. I think FPS was absorbed...)

(I could go on...Elxsi, Denelcor, Steve Chen's supercomputer company,
Control Data Corporation (pulled the plug on its supercomputers years ago),
etc. Probably two dozen companies have tried to enter the "next generation
supercomputer" business....)

Cray Research (not to be confused with Cray Computer, of course) is now a
unit of Silicon Graphics. And my old employer, Intel, is now struggling
with its "Supercomputer" business unit (which was once doing moderately
well, and was even the performance leader for a while, but which is now
being scaled back....)

The reasons for the collapse of the market are well-known: the end of
communism has lessened certain needs, the cut-backs in defense spending,
"the attack of the killer micros" (arrays of cheap micros give better
bang-for-the-buck), and, related to the themes of this list, NSA's
code-breaking just ain't what it used to be.

To wit, if even a petaflops machine, costing billions of dollars and
needing a nuclear power plant to power it, cannot make headway on cracking
a garden-variety PGP-encrypted message.....

(I grant that computers, supercomputers, workstations, arrays of
special-purpose hardware, etc. are useful for all sorts of related things,
such as signals analysis, filtering of voices, recognition of voices,
traffic analysis, etc. But I rather doubt that a single petaflops machine
is a good way to go for this.)

The "speculative" applications--the "miraworld" simulation environment, for
example--are nonsensical. There is no reason for a multibillion dollar
petaflops machine to be built so that researchers can schedule a few
minutes on it! (They'd rather have 0.1% the peak performance, but constant
or assured access, I'm sure.)

And so on. I don't see it happening.

--Tim May





We got computers, we're tapping phone lines, I know that that ain't allowed.
---------:---------:---------:---------:---------:---------:---------:----
Timothy C. May              | Crypto Anarchy: encryption, digital money,
tcmay@got.net  408-728-0152 | anonymous networks, digital pseudonyms, zero
W.A.S.T.E.: Corralitos, CA  | knowledge, reputations, information markets,
Higher Power: 2^1,257,787-1 | black markets, collapse of governments.
"National borders aren't even speed bumps on the information superhighway."