First engine to play in it is stockfish developement version from 14.2.2024 at 100K nodes per second no tablebases
2048 mbytes hash.
People who are interested to participate need to send games that I can verify to be reproducable.
A new entry is accepted only if the following conditions happens:
1)It does better than previous entries against at least one engine in the tournament
2)I can verify it by playing with my computer and it does not take too much time(no more than 10 seconds per move in my computer).
3)The engine is freeware that I can use(you can try some stockfish version at some fixed nodes setting or contempt and you can try also different free engines)
I use 2048 mbytes hash for all engines and to make things simple no tablebases.
First entry need at least to get 1.5-0.5 result against stockfish last developement version from 14.2.2024 with 100k nodes per move.
You can download it here.
https://abrok.eu/stockfish/
Author: Disservin
Date: Wed Feb 14 21:39:31 2024 +0100
Timestamp: 1707943171
Next entry to enter need to do better than at least one engine.
For example if first new entry could get a win with white and draw with black against stockfish100k from 14.2 then it is enough for the second entry to get a win with black against stockfish100k or to get 1.5-0.5 score against first entry.
Every participant in the tournament is going to play 2 games with white and black against the other engines and particpants need to be free engines in some setting that I can use in my computer(it can be also stockfish)
Note that winning with black against stockfish developement version at 100k is possible and dragon3.1 contempt50 proved it by the following game but dragon3.1 is not participating because it is not a free engine and also 100M with a single core takes too much time in my computer(Dragon may do it also with less nodes but even if it does it with 1M nodes per move then it is not enough because dragon is not free.
invitation to a new deterministic no opening tournament
Moderator: Ras
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
The first engine to join based on my choice is Berserk12 5,000k nodes per move that could beat stockfish 100k nodes per move with white and draw against it with black.
I chose berserk because it is second best free single cpu engine based on CCRL blitz rating
Third engine to join the tournament need to beat stockfish 100k with black(a possible task) or to beat Berserk12 5,000k nodes per move 1.5-0.5
Here are the games.
I checked that I can reproduce the games in my computer.
I chose berserk because it is second best free single cpu engine based on CCRL blitz rating
Third engine to join the tournament need to beat stockfish 100k with black(a possible task) or to beat Berserk12 5,000k nodes per move 1.5-0.5
Here are the games.
I checked that I can reproduce the games in my computer.
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
next opponent that I decided to add because it has a tcec rating.
Stockfish15 3000K nodes per move
It could beat the first stockfish with black so passed the test.
Surprisingly failed with the white pieces.
It is going to play next match against berserk 5M nodes per move when I expect 2 draws
Stockfish15 3000K nodes per move
It could beat the first stockfish with black so passed the test.
Surprisingly failed with the white pieces.
It is going to play next match against berserk 5M nodes per move when I expect 2 draws
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
2 draws as I expected.
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
Rubichess contempt 50 50,000,000 nodes per move could win against stockfish15 3M nodes per move but it is too slow to participate in my tournament.
I am disappointed from the choice 1.e4 e5 by rubichess and I do not think that it is a good idea to play for a win with black but maybe I am wrong.
I am surprised that with white Rubi could win against the berlin defense of stockfish.
I am disappointed from the choice 1.e4 e5 by rubichess and I do not think that it is a good idea to play for a win with black but maybe I am wrong.
I am surprised that with white Rubi could win against the berlin defense of stockfish.
-
- Posts: 200
- Joined: Fri Oct 12, 2012 12:48 pm
- Location: Seville (SPAIN)
- Full name: Javier Ros
Re: invitation to a new deterministic no opening tournament
The idea is very interesting, but if you want the experiment to be completely deterministic, the programs must run on a single processor thread. Otherwise, depending on the number of threads, the search may produce different results for the same number of analysed positions.
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
The games are with a single processor thread and I checked that I can reproduce the gamesJavier Ros wrote: ↑Tue Feb 20, 2024 11:59 am
The idea is very interesting, but if you want the experiment to be completely deterministic, the programs must run on a single processor thread. Otherwise, depending on the number of threads, the search may produce different results for the same number of analysed positions.
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
Rubichess contempt 50 100M nodes per move could beat berserk12 5M nodes per move 2-0(berserk practically stopped the search faster than 5M nodes but I do not care as long as the result is derministic.
I doubt if stockfish can do it regardless of number of nodes per move.
Next opponent of Rubichess is stockfish15 3M nodes per second(with 50M nodes per move it could get 1.5-0.5 result).
Here are the games
I doubt if stockfish can do it regardless of number of nodes per move.
Next opponent of Rubichess is stockfish15 3M nodes per second(with 50M nodes per move it could get 1.5-0.5 result).
Here are the games
-
- Posts: 10787
- Joined: Thu Mar 09, 2006 12:37 am
- Location: Tel-Aviv Israel
Re: invitation to a new deterministic no opening tournament
Rubichess contempt50 100M nodes per move won with white and drew with black against stockfish15 3K
-
- Posts: 640
- Joined: Fri Mar 30, 2018 7:20 am
- Full name: Andreas Matthies
Re: invitation to a new deterministic no opening tournament
What's that? You promised that "it [RubiChess] is too slow to participate in my tournament".