Author: enrico carrisco
Date: 00:02:26 04/04/02
A week or so ago, I mentioned that I would make an attempt to get to the bottom of the recent "HIARCS 8 Beta" engine that has seemed to make its way around the chess community. The fact that it was so widely distributed and even mentioned to be used by some that post in this forum is very sad. Everyone wants to see consistent increases in the strength and development of commercial chess engines, but some directly hinder such growth because they feel they deserve the hard work of others for free. That being said, the imposter HIARCS 8 engine has the following attributes: HIARCS8.DLL [Size: 192 KB (196,608 bytes)] HIARCS8.ENG [Size: 143 KB (146,944 bytes)] It is a developmental version from August 1999. It has been doctored to have a different learn file name (learn.log) and the "HIARCS 8 Beta" name has been doctored into the .eng. The original name was removed. This fake/imposter beta version is way short of where the true HIARCS 8 Beta stands today. In order to demonstrate the huge differences between the real HIARCS 8 and this fake, the following tests have been provided as working examples: EXAMPLE #1 ---------- 1NQ5/k1p1p3/7p/pP2P1P1/2P5/2pq4/1n6/6K1 w - - b6+!! is the key move All timings run on a PIII 750Mhz laptop: * HIARCS 7.32 takes 9 minutes 9 seconds to find b6+!! at 10/30 plies after 58211K nodes * FAKE HIARCS 8 beta takes 33 minutes 15 seconds to find b6+!! at 11/31 plies after 207071K nodes * The real HIARCS 8 takes 24 seconds to find b6+!! at 8/31 plies after 2452K nodes EXAMPLE #2 ---------- 1. e4 e5 2. Nf3 Nc6 3. Bc4 Bc5 4. c3 Nf6 5. d4 exd4 (All timings run on a PIII 750Mhz laptop) HIARCS 7.32 : 6. e5 eval=+0.06 depth=11/31 time=20:09 nodes=94444K FAKE HIARCS 8 beta: 6. e5 eval=+0.06 depth=11/30 time=14:10 nodes=66717K Real HIARCS 8 beta: 6. e5 eval=+0.05 depth=11/31 time= 3:15 nodes=15425K (in fact, the real HIARCS 8 Beta reaches the 12th iteration and pv at 12/34 +0.05 9:05 43028K nodes) Hence, the real HIARCS 8 Beta is significantly stronger and faster than HIARCS 7.32 and the fake/imposter engine. It will be available before the Summer 2002 (when an actual release date becomes available, I will post it here...) We're running the last string of tests against all other major chess engines at this time; I will post those results here as well. The games consist of 60/0 time controls on an AMD XP 2100+, both engines using the trusty old "general.ctg" with reset weights and alternate colors turned on. Enrico L. Carrisco HIARCS 8 Development Team hiarcs8@hotmail.com
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.