Re: The Petaflops Boondoggle Computer (was PET_ard)

[jim bell <jimbell@pacifier.com>]

At 10:00 AM 9/29/96 -0800, Timothy C. May wrote:
>(Hoist by their own petards indeed! Don't tell our Russian what petard means.)

Uh, wasn't that the name of the bald captain on Star Trek Next Generation?  
You know, "Jean-Luc Petard"?

>>      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.

Check out an article in about the September issue of Scientific American, 
1966, on the subject of the Illiac IV, which was one of the first attempts 
at a multiprocessor machine.  Originally it was conceived as a 256-processor 
unit, at 4 million (floating point?) operations per second per processor 
which would have been 1 giga ops per second, but it was eventually built as 
a 64-processor unit and turned on in about 1972 or so.  The succeeding 
factor-of-1000 improvement appears (if the item above is accurate) to have 
taken  24 years to accomplish, so it's hard to imagine that the next factor 
of 1000 will arrive appreciably sooner than year 2020.


>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.

Oddly enough, however, we're getting somewhat of an echo of the "big single 
processor" phenomenon with the micros.  We all know that in supercomputers, 
multiprocessors won out over single processors, and mainframes were just 
about defeated by microcomputers. 

Yet a look at Intel's pricing for Pentiums shows that they sell a 120-MHz 
chip for about $135, while they sell a 200-megahertz version for around $550 
or so.   Arithmetic suggests that a person would be far better off with a 
4-120-MHz-processor Pentium (cumulative clock rate 480 MHz) than a single, 
200-megahertz version.  (admittedly, peripheral logic costs will adjust this 
a little.)   Of course, this would also leave Intel flat on its ass 
attempting to compete with AMD, Cyrix, etc, because a somewhat higher speed 
per cpu is just about the only advantage they have.

Jim Bell
jimbell@pacifier.com