Author: Christophe Theron
Date: 01:34:00 10/13/97
Go up one level in this thread
On October 13, 1997 at 03:24:16, Keith Ian Price wrote: >Well, Baby Blue with one chess processor would still be capable of >analysing 1-2 million positions per second, which is 5-10 times as fast >as the current fastest chess program for the PC, and with a fuller >(though not yet proven to be better) evaluation. While I would in no >wise expect Deep Blue in a desktop, I would still be fascinated by Baby >Blue's analysis. If I could get a Son of DB it would do 10 million nps, >and this would be something. It may not be the Theory of Relativity, but >relative to PC programs of today, it will be relatively fast. I will >still sign the check at nearly 300000 kps. Are you sure a single processor being 5 times faster than a good PC program would play better? The current selection sheme of PC programs are quite complex (I use much selection in Chess Tiger), and I'm not sure the logic of the chip is able to do something like that. Take out the selection algorithms (or just make it a little less efficient) in a deep computation (level 12-13 for example), and you will have to wait 100 times more to get the same result. So maybe the power of DB comes from the huge number of processors more than from the fast NPS. BTW, I've heard DB is brute force (no selection). Has anybody information about that?
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.