Author: John R. Menke, Sr.
Date: 05:50:22 07/04/99
Go up one level in this thread
On July 04, 1999 at 07:42:55, leonid wrote: >On July 04, 1999 at 07:09:24, John R. Menke, Sr. wrote: > >>On July 04, 1999 at 06:23:08, walter irvin wrote: >> >>>On July 03, 1999 at 23:17:20, leonid wrote: >>> >>>>How really strong is ChessMaster 6000? I have seen its score on the recent >>>>competition and it is magnificent. Just a competition accident or what? >>>>Also, somebody can say me how many positions per second can see Hiarcs? >>>>Or his last version, that is now a Champion, or on some previous >>>>versions. The most interesting for me will be the numbers that Hiarcs >>>>demonstrate when it plays by "brute force" and when all pieces are still on >>>>the board. >>>> >>>>Thanks! >>>>Leonid. >>> >>>well chessmaster 6000 is extremly strong the only program i got that has beat >>>crafty consitantly . whats very strange is that it seems like it is only >>>searching about 15000 nps to crafty's 85000 or better . as for hiarcs i cant say >>>till wed. thats when my hiarcs 7.32 will arrive . >> >> >>I'm also curious how the "positions per second" in Chessmaster 6000 and the >>"nodes per second" in Hiarcs 7.32 compare. Initially I thought they were the >>same, just different nomenclature. But now I wonder... I plan to do some >>comparison tests when I can find the time. For example I want to use a stop >>watch (not trust the computer displays) to see how long each of them take to >>calculate the solution to a known test problem, and see how that compares with >>the "positions per second" and "nodes per second" values. I also want to do the >>same with a practical playing position to compare how long they take to "Brute >>Force" (completely) calculate to a certain ply depth. I have some initial >>impressions, but they could be wrong, so I won't speculate in advance about the >>results. If anybody has done those kinds of tests, I would be very interested >>in the results, so please post them! I guess you could fairly say that I am a >>100% hardhead, i.e. I don't believe ANYTHING until it is proved for a >>fact...usually requiring a simple test of some kind. >>--JRM > >Please, if your numbers will be too complicated to put them all on the Web, >send them at my address leonide@total.net This will be more that interesting >to me. Few questions about your expessions that I just can't understand. >What is "stop watch"? A "stop watch" is a simple hand-held device. It's an analog clock powered by a mainspring. You wind it up. Then you push the button and it starts ticking. Push the button again and it stops, giving you the elapsed time. It is used to time sports events such as the 100 meter run. --JRM >What is difference between "positions per second" >and "nodes per second"? It sound to me like it is the same. They may be the same, but I'm not sure. That's one of the things that I want to test. CM6000 and Hiarcs7.32 are different programs with different code, so they don't necessarily calculate it the same way. For example maybe the hash tables are involved. If the program finds a position in the hash table, how does it count that? As 1 node or 1 position? But the position in the hash table may represent the result of calculations involving many positions. There is also a difference between European and American electric clocks due to a difference in cycles per second (Hertz) of the alternating current (AC). I have run into that difference previously in chess software. For example, the chess clocks worked perfectly fine in England where the software was written, but were incorrect in the USA. --JRM >I did some research about the "branching factor" for the Rebel 2 and >found its number of positions per second. It was very simple for the >free Rebel 2 version that everybody can find on the Web. The exclusive >possibility with this game we have because it says frankly his number >of positions poer second and have the feature that permit demand "brute >force" search. If you would like to have the branching factor I can say you. >Leonid.
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.