Author: Robert Hyatt
Date: 14:55:28 01/22/00
Go up one level in this thread
On January 22, 2000 at 12:29:04, Tom Kerrigan wrote: >On January 22, 2000 at 06:03:19, blass uri wrote: > >>He does not have to do it because he does not have to produce the best program >>at tournament time control. > >In fact, he doesn't have to produce the best program at all. How many millions >of people would buy a Deep Blue program for their PCs, regardless of strength?? >I sure would. Please send a check for $500.00 to: Robert Hyatt 105 Green Wing Circle Pelham, AL 35124 I will send you one copy of a program that will run on your PC, and will say "hello from Deep Blue for the PeeCee" when you start it up. The point being that "Deep Blue for a PC would _not_ be Deep Blue, and I don't think _anybody_ would be fooled by the idea that it is... > >Besides, there is every indication that a Deep Blue program can run at a >reasonable nodes/second rate on a PC. If its evaluation function is really >everything it's cracked up to be, then it should be the best PC program ever. > >I think the only reasonable explanation is that the DB eval function simply >isn't that great. What would happen if FHH released this DB program and it >consistently got whomped by MChess, even though they search similar NPS? His >reputation in the computer chess world would take a serious dive. According to reported numbers, "DB" has 100K lines of C. Which would need serious porting since there is no SP-type hardware to use. And then it would need an eval written that takes about 40,000 clocks for the long path thru it. That is a _bunch_ of code considering that most chess programs spend 2000-4000 clocks per node, including searching _and_ evaluation. It is a lot of code to write... a lot of code to debug... and a lot of code to optimize to make it fast... And it wouldn't be deep blue. It would just be a pretty smart PC program with a (probably) significant search depth handicap. > >I've heard the story that DB's mobility was turned up WAY too high in one game >and it still won. This is supposed to be a testament to how terrific DB is, but >if you think about it, how embarrassing is a mistake like that??? Considering >how much this match was worth, the DB team should have been running and tuning >software simulations of the chips for months. Verifying that the chips were >working correctly should have been simple, even if they were weeks late. Tuning >should not have been an issue at all. With a mistake like the one they made, it >really sounds like they were flying by the seat of their pants... =/ > >-Tom Easy to say. Hard to do. Write and tune a chess program at 200 nodes per second... then run it at 200K nodes per second and see how far off your tuning is. I've done this in the old days and got burned several times. They had the same problem...
This page took 0.01 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.