Author: Dann Corbit
Date: 12:13:28 06/15/05
Go up one level in this thread
On June 12, 2005 at 18:42:15, Antonio Dieguez wrote: >ha this seems like a joke, reporting that great result with so much naturality. I won't jump and scream yet, because not enough games. Even stronger than Spike is definitely not sure from the data presented. However, any time I run more than 30 games and a program scores better than Shredder, my eyes pop out of their sockets (I literally have to search by hand on the floor for them) and also my jaw falls off and I have to search for it also. >I'm thinking some day I could post a list like this, if somebody believe me: > >"I was just in the middle of the testing of the last version of my engine. I am >not sure about the sufficiency of the number of games needed to be played to get >statistical significance, what do you think: > >1 Amyan1.7b : 2780 111 111 35 68.6 % 2689 20.0 % >2 Amyan1.7a : 2761 113 89 36 65.6 % 2692 33.3 % >3 Shredder9UCI : 2698 119 90 35 45.9 % 2697 31.4 % >4 Ruffian_105 : 2656 95 117 35 40.7 % 2706 28.6 % >5 Spike09a : 2606 96 105 35 29.1 % 2718 34.3 % I am ready to download it now. >On June 12, 2005 at 16:05:59, Dann Corbit wrote: > >>Fabien has made some improvements. >>Here is a little test tournament that I ran: >> Program Elo + - Games Score Av.Op. Draws >>1 Fruit_massy-msvc : 2750 111 111 35 58.6 % 2689 20.0 % >>2 Fruit_massy-intel-profile: 2731 113 89 36 55.6 % 2692 33.3 % >>3 Shredder9UCI : 2716 119 90 35 52.9 % 2697 31.4 % >>4 Ruffian_105 : 2676 95 117 35 45.7 % 2706 28.6 % >>5 Spike09a : 2626 96 105 35 37.1 % 2718 34.3 % >> >>You can get the games here: >>ftp://cap.connx.com/pub/chess-engines/new-approach/data.zip
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.