Author: leonid
Date: 12:00:23 08/24/00
Go up one level in this thread
On August 24, 2000 at 02:26:22, Stefano Gemma wrote: >On August 23, 2000 at 22:10:40, Joshua Lee wrote: > >>I was wondering what Ply equivalent for Comercial programs to equal Deep >>Thought's at 11 and 16? >>In Monty Newborn's book Kasparov versus Deep Blue it says on page 156 that >>Deep Thought 0.02 searched at 1,440,000Nodes per second to a Depth of 10 in >>typical middlegame positions(whatever those are smirk) and that many lines were >>searched to a depth of 15. >> >>While not the same as Deep Thought of 1989, I have seen it as searching 11 and >>12 Ply against Kasparov. At a rate of 2,160,000Nodes per second it says that DT >>looked between 200 and 500Million positions per move, which is impossible >>considering it was Game in 90 and for Game 1 and according to the book after >[...] > >I don't think that this is impossible. On a Pentium 450 MHz machine, my new >assembly engine can reach about 12 million positions per second. But this >version of the engine has the simplest possible evaluation function and it is >very very weak. A programmer better than me and with a faster hardware can easly >reach the numbers you've said, using more complex evaluation. Never mind how weak your engin is but 12 millions positions per second sound to me as too good to be real. But if you have them, you have bright future! Do you have some visible NPS counter in your program? After what some programmers said here, number of moves (not even positions) that their programs were able to generate was around 1.8 millions for 466Mhz computer. My program is likely to confirm this indirectly. 12 million are very far from those numbers. When it is said that it is positions/second, even more so. If you could say what kind of moves your program generate, it will be good. Brifly about your evealuation, if you can. After your like, you can respond here or send your message to my e-mail address. Please mention what language you used (NASM, MASM...) and for what system. Thanks in advance, Leonid. >Ciao!!! > >PS: the new engine is based on the Drago 16 bit engine, the sources are still >available for free on www.linformatica.com
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.