Athlon XP1800+
Fritz 10 GUI
128mb hash each
3-4-5 piece tablebases
Ponder off
HS-10moves.ctg book
40 moves in 75 minutes repeating (adapted for the CCRL 40/40)
2 cycles 30 rounds
The top three engines are likely to be promoted into Division 3 for the following season and the bottom three engines are likely to be relegated into Division 5.
Updated versions will not be introduced during the tournament.
Ovyron wrote:Hamsters 0.5, Hermann 2.0, Horizon 4.3
NOT Horizon!
Graham is using common books.
Perhaps Graham's question should be: "Which are the 3 best engines if they are supplied with a random book?" I guess that will give you the 3 best analysis engines, but it won't tell you much about playing strength for engines that have tuned books. I put more than half my development time into improving the book and steering away from troublesome positions. It's very painful for me when TDs throw away my hard work.
Disclaimer: Graham has every right to play his tournaments any way he likes. I'm not complaining about his tournament conditions or intending this to be a personal attack. I'm merely advising Ovyron (Ulysses P.) that he made a wrong choice with his pick.
In the main group there is a discussion indeed on engines with its own book or book generics.
I think that there are many factors that can influence in which your engine is above or no. And indeed having own books I do not believe that he is most influential.
It can influence that the computer is AMD or INTEL, can influence the time control, it can influence the GUI, it can influence ponder, if the programs use more of one processor, the number of rounds, if tablebases is used, the memory hash…
I think that the books are single a small part of everything what can influence that I have commented. With this I do not say that I prefer generic books.
I don't know of any other author that works harder improving the opening book than I do. I need to do something while I wait for test results, so I work on the book. I fix up about 15 losing lines per day (more than 5,000 per year). Sometimes I need to spend several days creating a good defense against an opponent's trap move. There is no measurable difference in strength for any single change. But the cumulative effect for 1,000's of changes is noticeable.
Horizon's book makes a difference. I've checked this over 1,000's of games. The difference started at about 30 Elo and seems to be about 50 now after my first year of book improvement. I'm hoping that in 5 years the Elo difference will be close to 100 points but it may be a case of diminishing returns.
50 points are many coverall when engines are very similar in force and when the duration of the games is great, perhaps you are right and is the difference between being above and down.
