Author: Bertil Eklund
Date: 12:32:48 01/09/03
Go up one level in this thread
On January 09, 2003 at 11:47:54, Stephen Ham wrote: >On January 09, 2003 at 10:52:50, Christophe Drieu wrote: > >>PIII 500 Mhz >>Gambit Tiger Aggressive Hash = 24 Mb >>Ruffian 1.0.1 Hash = 32 Mb >>Own books >>Engine research tool >> >>Blitz 5 minutes + 0 >> >>GTA15 101*11110111100111 13.5 >>Ruffian 010*00001000011000 4.5 > >Dear Christophe, > >Thanks for this interesting post. But, since I'm a computer "dummy", I'm unclear >on a few minor points. As a "dummy", I'm probably going to write somethine >really stupid. > >1) I find it odd that there were no draws out of 18-games. In fact it was one draw. I have played 48 tournament games Ruffian vs Hiarcs8 and there were a lot of draws. The result was 29,5-18,5 with Ruffian on A1200 and Hiarc8 on AMD450. > >2) Why was Ruffian given more RAM? While a larger RAM amount is often >desireable, I understand that in blitz chess, it shoud be keep relatively small >in order to speed up the search. So for a fair comparison, shouldn't the engines >have the exact same RAM? Maybe 8 MB is too small a figure to be concerned about, >but in principle, I think that all variables should be held identical and >constant for both engines in order to make a valid test. The size for hash in Tiger are 6,12,24,48 and so on. > >3) Given your experience with Ruffian (hopefully you have experience at longer >time controls), do you see evidence of a playing style? For example, is it >aggressive like Junior or solid like Shredder 5.32 and CT 14? Does it exhibit >any clear weaknesses relative to other engines, such as weaker than normal >engame play or worse than normal problems in closed positions? It only have one clear weakness, in closed positions it plays weaker then most other top-engines. If out of book early,it plays the opening weaker then many other engines. Ruffian plays the endgame very well.In tactical positions it is at the same level or better then the competition. The opening book is fair but smaller then the competition. Ruffian has no learner, a thing that costs a few points in longer matches. Ruffian also gives a handicap because it often plays the game twice as fast as the opponent. > >4) Does Ruffian's evaluation function seem generally accurate (e.g. CT 14 or >Fritz 7) or does it vary a lot from ply to ply and or move to move (Nimzo 7.32 >and Shredder 7). The evaluation seems to be good. Bertil >Thanks in advance. > >All the best, > >Stephen
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.