Author: Robert Hyatt
Date: 07:21:28 01/16/00
Go up one level in this thread
On January 16, 2000 at 05:51:30, Dann Corbit wrote: >On January 15, 2000 at 00:23:34, Robert Hyatt wrote: >[snip] >>The SP isn't the best machine by far. Hsu could have done far better with a >>machine based on an SMP alpha platform, rather than the message-passing on the >>SP architecture. And then there are the Crays of course. >Now maybe, not then. At the time, the Alpha was a weak shadow of the RS/6000 >and with the full complement of CPU's, the RS/6000 is still master over the best >current Alpha box. Cray is Silicon Graphics now. And it would be a large >paradigm shift from SMP to message passing. > First, a 16 cpu alpha can, and always could, blow off a 16 cpu SP badly. The SP isn't horribly fast. The main point here is that Hsu could use no more than 32 SP processors with his existing hardware. And a 32 cpu alpha, using a real shared memory would have solved a lot of problems... It is somewhat "moot" however, for the technical reasons I have given before. Faster host processors simply mean less work on the chess processors... And don't forget Cray. They make the T3 group of machines which _are_ message- passing machines based on the alpha. With a thousand alphas or whatever you can afford to buy... BTW cray is no longer SGI. For some reason they appear to have separated once again. >>I think that most of what was accomplished could have been pulled off by any >>company with the foresight to recognize the enormous P/R potential of the >>project. Don't forget that Deep Thought was unbeatable as far as other >>computers went, losing only a couple of games over a 10 year period. Yet it >>was put together for almost nothing at a university... >Here, I agree with you. But vision at the top is even more rare than incredible >technical ability.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.