| Rank | Engine | Score | Ro | Ry | |
|---|---|---|---|---|---|
| 1 | RobboLito_009_w32 | 61.0/100 | 1==11=0=0===0==100=1=1=0= ==111=1====011===0=11011= 1=1=0=0=11==0===11011=011 =1111111===0=011=====111= | ||
| 2 | Rybka 3 1-cpu w32 | 39.0/100 | 0==00=1=1===1==011=0=0=1= ==000=0====100===1=00100= 0=0=1=1=00==1===00100=10 0=0000000===1=100=====000= |
100 games played / Tournament is finished
Tournament start: 2010.02.12, 18:23:44
Latest update: 2010.02.13, 23:41:30
Level: Blitz 5/0
Hardware: AMD Turion(tm) 64 X2 Mobile Technology TL-60 with 2,048 MB Memory
Operating system: Microsoft Windows Vista Home Edition Service Pack 2 (Build 6002)
PGN-File: Download the games here
Table created with: Arena 2.0.1
Dear Colleague, I am not as entusiastic as you in terms of Robbolito 0.09. I've just finished blitz (4'+2'') turnament on 32-bit platform (Intel Core 2 Duo 1,5GHz; 2GB RAM, Vista 32-bit) and Robbolito gained only the 5th place among 7 engines.
ReplyDeleteResults:
1.Stockfish 5.0/6
2.Rybka 2.2n2 mp 32-bit 4.5/6
3.Fritz 11 3.0/6 (9.00)
4.Glaurung 2.2 3.0/6 (7.25)
5.RobboLito 0.09 w32 2.5/6
6.Toga II 1.4 beta5c 2.0/6
7.bright-0.5c 1.0/6
Yes, maybe. Different settings,total number of rounds,format of the tournament/match can lead to different results. Would be interested to know other testers' results. I would run an individual match with Robbo and the rest of the engines when I got time. BTW, what were your settings (Ponder, Tablebases, opening,etc),tournament format,number of rounds,GUI and would you mind to share the pgn game?
ReplyDeleteSeems that your format is single round robbin. well, if you beleive that 6-games is enough to make a conclusion then you are right, but I doubt it will be a true reflection of any of the engine strength. try run engine match say 50 games each pair.
ReplyDeleteYou are right both. You can manipulate different settings to justify your thesis.
ReplyDeleteYes, it was single round robin. I am not sure if games are valuable but if you wish let me know.
I used defaulted settings for each engine, hash table 512MB, TB 3-4-5, Fritz GUI.
I noticed that some engines don’t like each other. Next observation is that some engines create almost winning position and then lose it rapidly.
My last experiment was to limit hash table to 256MB with the same other settings and results are surprisingly totally different. You can find them at my blog http://lionswordintheair.blogspot.com/