Author: Per Jørgensen
Date: 02:48:36 01/11/06
Go up one level in this thread
On January 11, 2006 at 05:09:12, enrico carrisco wrote: >On January 11, 2006 at 04:11:10, Per Jørgensen wrote: > >>On January 11, 2006 at 01:43:55, Harvey Williamson wrote: >> >>>On January 10, 2006 at 09:35:10, Per Jørgensen wrote: >>> >>>>Hello Chesscomputerfriends! >>>> >>>>As I described in this Forum on the 30th of December and earlier I'm making a >>>>new Ratinglist based on 15 openingpositions. 10 of these are known from the >>>>CSS-Ranglist (for further information check this website www.computerschach.de), >>>>here are the 5 new openingpositions: >>>> >>>>C30 = Kings Gambit (1.e4,e5 2.f4) >>>> >>>>E17 = Queenindian (1.d4,Sf6 2.c4,e6 3.Sf3,b6 4.g3,Lb7 5.Lg2,Le7) >>>> >>>>B02 = Scandinavian (1.e4,d5 2.exd5) >>>> >>>>D35 = Orthodox Queens Gambit (1.d4,d5 2.c4,e6 3.Sc3,Sf6) >>>> >>>>A57 = Modern Ben Oni (1.d4,Sf6 2.c4,c5 3.d5) >>>> >>>>It's my intention to create an élite Ratinglist of 17 engines so that each >>>>engine has played 480 games (30 x 16) when the list is full. I'm testing on the >>>>following Hardware: >>>> >>>>Asus A8V Motherboard, Socket 939, FSB 1.000 MHz >>>>AMD Athlon 64 4000+, 2.4 GHz, 1 MB Level 2 Cache, 90 nm San Diego Core >>>>2 x 512 MB Corsair PC 3200XLPT RAM (CAS Latency 2-2-2-5) >>>> >>>>And here are the specific testconditions: >>>> >>>>Shredder 9 GUI >>>>Thinkingtime for each engine: 40 moves in 14 minutes repeatedly >>>>Hash Tables for each engine: 128 MB PC 3200 Ram (CAS Latency 2-2-2-5) >>>>8 MB Cache for TB's >>>>TB's: Nalimov - 3 und 4-pieces >>>>Pondern OFF >>>> >>>>Since more than 5 weeks the tests are running and the first result were as >>>>follows: >>>> >>>>Fritz 9 - Fruit 2.2.1 19-11 >>>>Shredder 9 - Fritz 9 18-12 >>>>Shredder 9 - Fruit 2.2.1 11½-18½ >>>>Toga II 1.1 - Fritz 9 14-16 >>>>Toga II 1.1 - Fruit 2.2.1 10-20 >>>>Toga II 1.1 - Shredder 9 16-14 >>>>Loop List - Fruit 2.2.1 11-19 >>>>Loop List - Fritz 9 8½-21½ >>>>Loop List - Shredder 9 12½-17½ >>>>Loop List - Toga II 1.1 9½-20½ >>>>Ktulu 7.5 - Fritz 9 11½-18½ >>>>Ktulu 7.5 - Fruit 2.2.1 12-18 >>>>Ktulu 7.5 - Shredder 9 11-19 >>>>Ktulu 7.5 - Toga II 1.1 12-18 >>>>Ktulu 7.5 - Loop List 13½-16½ >>>>Spike 1.0a Mainz - Fritz 9 8-22 >>>>Spike 1.0a Mainz - Fruit 2.2.1 11½-18½ >>>>Spike 1.0a Mainz - Shredder 9 14½-15½ >>>>Spike 1.0a Mainz - Toga II 1.1 13-17 >>>>Spike 1.0a Mainz - Ktulu 7.5 13½-16½ >>>>Spike 1.0a Mainz - Loop List 12½-17½ >>>>Hiarcs 10 - Fritz 9 11½-18½ >>>>Hiarcs 10 - Fruit 2.2.1 13½-16½ >>>>Hiarcs 10 - Shredder 9 11-19 >>>>Hiarcs 10 - Toga II 1.1 11½-18½ >>>>Hiarcs 10 - Ktulu 7.5 18-12 >>>>Hiarcs 10 - Loop List 16-14 >>>>Hiarcs 10 - Spike 1.0a Mainz 19½-10½ >>>> >>>>And here comes the latest results: >>>> >>>>Chess Tiger 15.0 - Fritz 9 11-19 >>>>Chess Tiger 15.0 - Fruit 2.2.1 8-22 >>>>Chess Tiger 15.0 - Shredder 9 7-23(!) >>>>Chess Tiger 15.0 - Toga II 1.1 9½-20½ >>>>Chess Tiger 15.0 - Hiarcs 10 10-20 >>>>Chess Tiger 15.0 - Loop List 15-15 >>>>Chess Tiger 15.0 - Ktulu 7.5 11½-18½ >>>>Chess Tiger 15.0 - Spike 1.0a Mainz 13½-16½ >>>> >>>>Although Chess Tiger 15.0 doesn't belong to the gruop of Topengins of today I >>>>was expecting a littte bit stronger performance. Especially Shredder 9 managed >>>>to almost humiliate Chess Tiger with a 23-7 victory (the biggest victory of all >>>>so far in my testing). Here follows the updated ranking: >>>> >>>>1. Fritz 9 146½ Points >>>>2. Fruit 2.2.1 143½ P. >>>>3. Shredder 9 137½ P. >>>>4. Toga II 1.1 134½ P. >>>>5. Hiarcs 10 121 P. >>>>6. Ktulu 7.5 107 P. >>>>7. Loop List 104½ P. >>>>8. Spike 1.0a Mainz 100 P. >>>>9. Chess Tiger 15.0 85½ P. >>>> >>>>Compared to the latest ranking there are only a few important changes: Fruit >>>>2.2.1 has reduced the lead of Fritz 9 and Ktulu 7.5 has passed Loop List. Even >>>>with a fine 20-10 victory for Hiarcs 10 against Chess Tiger it's fair to say >>>>that this engine so far has been a disappointment. When the next enginetest is >>>>finished I will have 10 engines tested and from that point (and in the future) I >>>>will publish real ELO-lists. >>>> >>>>The next engine to be tested is the brandnew SmarThink 1.00. The results so far >>>>for this russian engine have been quite mixed so I am excited to see how it will >>>>perform under my testconditions. Needless to say; as soon as Rybka 1.2 is >>>>available (probably 1st of February) I'll test it and the same goes >>>>for Shredder 10. >>>> >>>>Best regards >>>>Per >>> >>> >>>Did you switch off poaition learning for Shredder 9? >> >>Hello Harvey! >> >>Thanks for your interest in my testwork. Yes, position learning ist allowed for >>all engines who is capable of using this option. However, I don't see the big >>victory against Chess Tiger 15.0 as a result of this feature. I have been >>testing for a long time and I haven't recoqnized such a pattern or development >>in the results and the same goes for the important CSS-Ratinglist. Needless to >>say, position learning ON has an influence on the results but in my opinion it >>is a marginal one. Like I wrote in my posting I have just started the test of >>SmarThink 1.00 and right now the russian engine has an impressive 3-0 lead >>against Shredder 9. It goes up and down for Shredder 9 as it does for all >>engines. >> >>Best regards >>Per > >Hello Per. > >I believe Harvey was asking because he, like myself, is part of HIARCS >development and we didn't have any match results less than 55% against Shredder >9. However, it's only been 30 games with your tournament, so perhaps things >will still turn around. > >Regards, > >-elc. Hello Enrico! So far Hiarcs 10 hasn't made a topperformance in my testingsystem. The reason or reasons for this is really an enigma for me. My Hardware is quite fast and Up-to-date, my testconditions are in my opinion very fair (i.e. using the fixed openingpositions in stead of books) and so on. What makes it even more mysterious is the fact that Klaus Wlotzka that runs the CSS-list achieved a much better testresult for Hiarcs 10 resulting in an amazing 2nd place just a few points behind Fruit 2.2.1. And our testconditions are indeed comparable (i.e. using the fixed openingpositions in stead of books). Like Klaus I have been testing Hiarcs 10 on default-settings. As you see it's really strange how come Hiarcs 10 hasn't been performing better in my testings. I have to admit that I several times have been wondering if I got a "Mondayversion" of Hiarcs 10. If you are interested I would be happy to mail you the games of Hiarcs 10. Best regards Per
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.