Author: Ralf Elvsén
Date: 10:17:38 08/11/99
Go up one level in this thread
On August 11, 1999 at 08:35:52, blass uri wrote: >On August 11, 1999 at 05:59:25, Ralf Elvsén wrote: > >>Hi there >> >>Here is a game I won against Hiarcs 7.32 (pgn below). >>Mighty proud to have fair and square defeated the "superstrong" >>Hiarcs, I looked at the game in analysis mode, but on a different >>computer and now I am perplexed... >> >>The game was played on computer A (see below). When analyzing on >>computer B I noticed that the analysis is much faster on B, much >>more than I would expect from the hardware. What is even more >>strange is that the evaluation of the position on the two computers >>are sometimes very different. Computer A is much more pessimistic. >>For instance, after 14. Kh1 Hiarcs played 14 ... Bd7 in the game >>with an eval of 0.61 (positive good for white = me) after search >>to depth 10. On comp B the eval is around equality and Bd7 is move >>nr 4 in the analysis candidate list after depth 10, eval = 0.00. > >My hiarcs on p200 got Bd7 0.61 9/27,0.68/10/30 >I stopped it after more than 10 minutes and did not give it to finish iteration >10. >I suspect that the results of hiarcs on computer B are based on learning from >the game that it lost. > >Uri Hi Uri Eehh... Your reply still leaves me wondering. When you look at the game, Hiarcs play is almost pathetic. I still have a "feel" that Hiarcs underperformed. And look at the big difference in search depth on move 14 between the two machines (10 vs 12 in about the same time). But I don't know much about how Hiarcs search works. Maybe its not so predictable as e.g. Craftys. I'll wait with the celebrations a while. Btw, I _have_ lost to Hiarcs :) These losses would be even more depressing if something was wrong with the machine or the settings... I'll dig in to this some more if the game isn't rejected/verified by some enlightened posters (other than yourself of course). Thanks for your answer, Ralf
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.