Author: Uri Blass
Date: 06:37:52 01/22/06
Go up one level in this thread
On January 22, 2006 at 08:47:58, Tord Romstad wrote: >On January 21, 2006 at 09:57:04, Jon Dart wrote: > >>I have built up a test suite from positions from >>Arasan's games (plus a few from other sources). >>Most of these have been verified by running them at >>long time controls. They are probably comparable in >>overall difficulty to ECMGCP. > >No. ECMGCP is relatively easy on modern hardware, but >your test suite seems to be very hard. At 10 seconds per >move on a Pentium IV 2.4 GHz, the latest version of my >engine scores 159/183 at ECMGCP, but only about 50% >in the Arasan suite (I don't remember the exact score). > >Tord I think that for some reason your program is relatively better in ECMGCP and not in arasan test suite but I also agree that probably scoring half of ECMGCP is easier than scoring half of arasan for most programs. Note that the difference is not very big and I think that if you take the easy half of ECMGCP out of the test then the tests are going to be of similiar difficulty(glaurung may still be better in ECMGCP but part of the other programs will score better in arasan). I think that scoring 50% of arasan and scoring 75% of ecmgcp are probably problems of similiar difficulty. Uri
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.