https://github.com/booot76/Booot-chess-engine
Hi community!
New ideas, new release.
All the best
Alex Morozov.
booot 7.4 release
Moderator: Ras
-
Steve Maughan
- Posts: 1278
- Joined: Wed Mar 08, 2006 8:28 pm
- Location: Florida, USA
Re: booot 7.4 release
Quick question: how does the speed of the engine vary if compiled using Delphi vs FPC?
— Steve
P.S. I'm in the middle of creating a Delphi based chess engine. It should be a few months before it sees the light of day
— Steve
P.S. I'm in the middle of creating a Delphi based chess engine. It should be a few months before it sees the light of day
http://www.chessprogramming.net - Juggernaut & Maverick Chess Engine
-
booot
- Posts: 86
- Joined: Sun Jul 03, 2016 10:29 pm
Re: booot 7.4 release
Quick answer : i did not noticed any difference in speed. Engine mostly does AVX2 instructions for matrix multiplications, so no matter what compiler maid the rest of interface.
Good luck with Pascal! Not bad language at all
Good luck with Pascal! Not bad language at all
-
Wolfgang
- Posts: 983
- Joined: Sat May 13, 2006 1:08 am
-
Wolfgang
- Posts: 983
- Joined: Sat May 13, 2006 1:08 am
-
AndrewGrant
- Posts: 1959
- Joined: Tue Apr 19, 2016 6:08 am
- Location: U.S.A
- Full name: Andrew Grant
Re: booot 7.4 release
Hey. I wanted to ask if you were still using self-play data to train Booot's NNUE, or if you had swapped to Leela data for example.
-
booot
- Posts: 86
- Joined: Sun Jul 03, 2016 10:29 pm
Re: booot 7.4 release
I still use own generated data but this time i 'recalibrate' them after generating on Python side to 'fill' all the range [0..1] for more effective net learning. It's additional part of the art.