I see based on the results that 8 second per game +0.08 seconds per move is popular and the faster time control that I could find is 5+0.05.
I wonder why not testing at faster time control and what is the fastest time control that you practically can test(what about 0.1+0.001 time control).
http://chess.grantnet.us/
The reason that I think testing at very fast time control may be interesting is that I have the idea of some chess engine A when the evaluation is based on the result of B play against itself at very fast time control when is better to have B as strong as possible so evaluation of one position(win draw loss is usually correct and does not take too much time).
What is the fastest time control that developers use
Moderator: Ras
-
- Posts: 10696
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
-
- Posts: 1936
- Joined: Tue Apr 19, 2016 6:08 am
- Location: U.S.A
- Full name: Andrew Grant
Re: What is the fastest time control that developers use
These time controls are kinda meaningless by thesmselves. The 8+.08s you see on OpenBench is relative to {some computer}. In the case of Weiss that you probably looked at, its based on a thread of a Ryzen 3700x. If instead it was based on a core of a 7950x, it would look more like a 3+.03s test.Uri Blass wrote: ↑Tue Dec 24, 2024 9:16 am I see based on the results that 8 second per game +0.08 seconds per move is popular and the faster time control that I could find is 5+0.05.
I wonder why not testing at faster time control and what is the fastest time control that you practically can test(what about 0.1+0.001 time control).
http://chess.grantnet.us/
The reason that I think testing at very fast time control may be interesting is that I have the idea of some chess engine A when the evaluation is based on the result of B play against itself at very fast time control when is better to have B as strong as possible so evaluation of one position(win draw loss is usually correct and does not take too much time).