additionally to the CEGT 40/120 dual core and single tests the CEGT 40/120 Quad testing is now started. The engines will be included in the CEGT 40/120 rating list with 2 CPU and single versions until we have enough games for maybe an additional separate list.
After those we will include more engine versions one by one. New engine versions like the announced Deep Shredder 11 will come first.
Each engine will play a 50 games match against all others like before. All engines receive 1536 MB hash. Book will be Harry Schnapp´s 8moves. It is planned to match 4 CPU engines against other 4 CPU engines only.
Results will be posted in the CEGT Forum regularly and also in Rybka and Hiarcs Forum (if those engines are involved in matches) and in German fora.
Testers owning a Quad machine are welcome to contribute, also if they are not CEGT testers up to now.
additionally to the CEGT 40/120 dual core and single tests the CEGT 40/120 Quad testing is now started. The engines will be included in the CEGT 40/120 rating list with 2 CPU and single versions until we have enough games for maybe an additional separate list.
After those we will include more engine versions one by one. New engine versions like the announced Deep Shredder 11 will come first.
Each engine will play a 50 games match against all others like before. All engines receive 1536 MB hash. Book will be Harry Schnapp´s 8moves. It is planned to match 4 CPU engines against other 4 CPU engines only.
Results will be posted in the CEGT Forum regularly and also in Rybka and Hiarcs Forum (if those engines are involved in matches) and in German fora.
Testers owning a Quad machine are welcome to contribute, also if they are not CEGT testers up to now.
Hello Heinz,
I'll be following your testing with interest. I hope you'll post the results here for us too.
Tony Thomas wrote:You have done it again, your new Quad 40/120 when it is available will be the best collection of highest quality computer games to date.
Of course Sedat has been doing 40/120 testing on Quad Hardware QX6700 for a while now
A small correction. Hash will be 1024 MB for those engines using a scheme 64, 128, etc.
If the list is only of 4 cpu I have no objecions to it but if you plan to include in the same list 2 cpu and 4 cpu then I prefer to see the same hash for both(otherwise we are not going to know how much advantage in rating is because of more processors and how much advantage is because of hash).
I assume that all engines earn from more hash and it is possible that some fast searcher with 2 cpu earn more from 1024 mbytes hash relative to slow searcher with 4 cpu.
2 CPU engines got 1024 MB before, so this will not change and you will get equal conditions here. I would like to give 2048 MB to all, but this is not possible with 4 GB RAM on each machine.
On further thought I guess that we need also some matches 4 CPU against 2 CPU to avoid different pools of matches with no connections to calculate a decent rating list. If so only the best 2 CPU versions will be chosen for this matches.
What amount of games will concern, it will last some time to have many like in the beginning of CEGT 40/120 one and a half year ago. Anyway we will prefer fewer engines with many games to many engines with few games for better statistical accuracy.
Heinz Van Kempen wrote:
What amount of games will concern, it will last some time to have many like in the beginning of CEGT 40/120 one and a half year ago. Anyway we will prefer fewer engines with many games to many engines with few games for better statistical accuracy.
Yes, I agree with this 100%.
I am curious as to the actual hardware being used and the actual time control ? Can you confirm ?
Hi, it sound great!!
It wolud be nice if you test each engine with own book to , since the Rybka-zappa match I'm still wonder which's the real diference between both engines. so if we take both test ,under generic book and own book we can bring some conclusions, Deep Junior wolud be stronger in own book than generic book, don´t you think so?. Thanks to all CEGT team
Regards
ArmyBridge wrote:Hi, it sound great!!
It wolud be nice if you test each engine with own book to , since the Rybka-zappa match I'm still wonder which's the real diference between both engines. so if we take both test ,under generic book and own book we can bring some conclusions, Deep Junior wolud be stronger in own book than generic book, don´t you think so?. Thanks to all CEGT team
Regards
testing with own book is not interesting for me
Note that zappa 2 cpu scored 42% against rybka 2 cpu in cegt and considering the fact that zappa earns more from more processors results of 5.5:4.5 for zappa with 8 cpu is not something that you can consider to be a big surprise with generic books.
,
