Hi Tom,
Can Roc be compiled so as to support Syzygy, like Jose's Gull 3.1 does?
New engine "Roc"
Moderators: hgm, Rebel, chrisw
-
- Posts: 40
- Joined: Fri Jul 22, 2016 7:51 pm
Re: New engine "Roc"
I am testing that now; don't know how much Elo impact it will have.
At first I will be releasing only binaries for Roc with TB support. I used Basil Falcinelli's Fathom code, but I had to mess with it to get a Windows build.
At first I will be releasing only binaries for Roc with TB support. I used Basil Falcinelli's Fathom code, but I had to mess with it to get a Windows build.
"Wise and cruel was the Bird, and wise and cruel were the Sons of the Bird."
-
- Posts: 582
- Joined: Wed May 10, 2006 7:28 pm
- Location: Birmingham, England
Re: New engine "Roc"
Thanks, Tom, for your work.
In infinite analysis the engine doesn't report the time used for each line.
Please always compile a non-popcount release.
In infinite analysis the engine doesn't report the time used for each line.
Please always compile a non-popcount release.
Marek Soszynski
-
- Posts: 50
- Joined: Sun Jan 30, 2011 1:08 pm
"Roc" - it is an improvement !!
" ... congrats to Tom Hyer it is an improvement !!! "
http://computer-chess.org/forum/index.p ... ad&id=3163
" Hawkeye 1.03 and Gull 16021521 were NOT"
http://computer-chess.org/doku.php?id=c ... ments:gull
http://computer-chess.org/forum/index.p ... ad&id=3163
" Hawkeye 1.03 and Gull 16021521 were NOT"
http://computer-chess.org/doku.php?id=c ... ments:gull
-
- Posts: 2435
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: "Roc" - it is an improvement !!
I started a testrun of Roc 0.2 for my Stockfish-gamebase. 7000 games. 70''+700ms, singlecore: 1000 games versus Stockfish 8, Komodo 10.3, Houdini 5, Shredder 13, Fizbo 1.9, Critter 1.6a, Fire 4...
If Roc 0.2 is running stable and if it is measureable stronger than Gull 3, then Roc 0.2 will replace Gull 3 as an opponent for my Stockfish- and asmFish-testruns. Result not before Wednesday.
Stay tuned.
Stefan (SPCC)
http://spcc.beepworld.de
If Roc 0.2 is running stable and if it is measureable stronger than Gull 3, then Roc 0.2 will replace Gull 3 as an opponent for my Stockfish- and asmFish-testruns. Result not before Wednesday.
Stay tuned.
Stefan (SPCC)
http://spcc.beepworld.de
-
- Posts: 4606
- Joined: Wed Oct 01, 2008 6:33 am
- Location: Regensburg, Germany
- Full name: Guenther Simon
Re: "Roc" - it is an improvement !!
I think it should also be compared to LazyGull and Gull 3.1 forked by Basil Falcinelli.pohl4711 wrote:I started a testrun of Roc 0.2 for my Stockfish-gamebase. 7000 games. 70''+700ms, singlecore: 1000 games versus Stockfish 8, Komodo 10.3, Houdini 5, Shredder 13, Fizbo 1.9, Critter 1.6a, Fire 4...
If Roc 0.2 is running stable and if it is measureable stronger than Gull 3, then Roc 0.2 will replace Gull 3 as an opponent for my Stockfish- and asmFish-testruns. Result not before Wednesday.
Stay tuned.
Stefan (SPCC)
http://spcc.beepworld.de
https://github.com/basil00/Gull
-
- Posts: 1600
- Joined: Mon Feb 21, 2011 9:48 am
Re: "Roc" - it is an improvement !!
Guenther wrote:Just for the record, Gull 3.1 is not a basil forked.pohl4711 wrote:I started a testrun of Roc 0.2 for my Stockfish-gamebase
I think it should also be compared to LazyGull and Gull 3.1 forked by Basil Falcinelli.
https://github.com/basil00/Gull
It is all possible equal to the project of ThinkingALot,
Nor do I think there are differences of game between the twos versions.
https://github.com/Velmarin/Chess-Gull-Syzygy/releases
-
- Posts: 40
- Joined: Fri Jul 22, 2016 7:51 pm
Re: "Roc" - it is an improvement !!
Thanks. This is very good news; it exceeds my expectations.
"Wise and cruel was the Bird, and wise and cruel were the Sons of the Bird."
-
- Posts: 40
- Joined: Fri Jul 22, 2016 7:51 pm
Re: New engine "Roc"
"Wobble" is an idea I am experimenting with, to change the move ordering a little bit based on the depth. The idea is that each iterative search will have a chance to find a move that the previous iteration missed. I hope this might be useful at long time control.
If you are able to test Wobble=1 against Wobble=0 for any significant number of games, I'd be interested to know how it does.
Btw, Roc 0.1 has Wobble=1 hardcoded, Roc 0.2 has Wobble=0 as the default.
If you are able to test Wobble=1 against Wobble=0 for any significant number of games, I'd be interested to know how it does.
Btw, Roc 0.1 has Wobble=1 hardcoded, Roc 0.2 has Wobble=0 as the default.
"Wise and cruel was the Bird, and wise and cruel were the Sons of the Bird."
-
- Posts: 3186
- Joined: Sat Feb 16, 2008 7:38 am
- Full name: Peter Martan
Re: New engine "Roc"
Thanks again for explaining. Before I'd do Celo- comparison by eng-eng, I'd like to see some Time to Depth, Time to Solution and Time to Eval- changes in certain test positions.THyer wrote:"Wobble" is an idea I am experimenting with, to change the move ordering a little bit based on the depth. The idea is that each iterative search will have a chance to find a move that the previous iteration missed. I hope this might be useful at long time control.
So to get you fully right: the higher Wobble is the lower "selectivity" of search
Peter.