Page 7 of 9
Re: SF PB (Stockfish)
Posted: Mon Mar 11, 2024 9:58 pm
by Skynet
SF-PB-110324 "VVLTC search tune" (next patch is about nothing), sorry forgot to add SlowMover Option.
https://pixeldrain.com/u/UWSsfHXT
Re: SF PB (Stockfish)
Posted: Wed Mar 13, 2024 3:51 am
by Skynet
SF-PB-120324 "
Fix Raspberry Pi Compilation"
https://pixeldrain.com/u/aqetRZeE
feel free to share the engine in other forums, i don't have time.
Re: SF PB (Stockfish)
Posted: Wed Mar 13, 2024 9:04 am
by ZamChess
Thank you Skynet !
Re: SF PB (Stockfish)
Posted: Wed Mar 13, 2024 6:41 pm
by Solista
This engine is a disaster for analyses.
Fix Raspberry Pi Compilation in Stockfish 16.1 would be better for Raspberry Pi. SF 16.1 is better engine.
Re: SF PB (Stockfish)
Posted: Thu Mar 14, 2024 2:39 am
by Skynet
Solista wrote: ↑Wed Mar 13, 2024 6:41 pm
This engine is a disaster for analyses.
Fix Raspberry Pi Compilation in Stockfish 16.1 would be better for Raspberry Pi. SF 16.1 is better engine.
Hey Eduard. You remind an amateur with your statements. SF is not designed for solving tactics only, the main goal of the team is to gain elo and make it stronger. There is so many tactical positions that SF, Komodo, Berserk etc can't solve them, but people like them and they use them for all the purposes. As one of the authors said some years ago "engine can't solve the concrete position, because it will never bring it self in this position in first place". As for evaluation, it may change from patch to patch and
it is normal, it's impossible to have everything at once and there is no need to criticize every
development version because they are constantly evolving. If you don't like the engine, which has won multiple world championships - don't use it. Take CorChess, Cristal, Black Diamond, Zappa etc and do your job in peace.
Re: SF PB (Stockfish)
Posted: Thu Mar 14, 2024 2:50 am
by Solista
It's not the actual search code. The problem is only threshold for Small Network. Threshold is set quite low and the small network has too much influence. But well, I have a different opinion than you. We don't have to agree. No problem. Have fun compiling.
Re: SF PB (Stockfish)
Posted: Thu Mar 14, 2024 7:11 am
by ConRog
Solista wrote: ↑Thu Mar 14, 2024 2:50 am
It's not the actual search code. The problem is only threshold for Small Network. Threshold is set quite low and the small network has too much influence. But well, I have a different opinion than you. We don't have to agree. No problem. Have fun compiling.
Hi Solista,
Skynet is right, stockfish is the best engine for playing, and it is build for this purpose. It's not an engine for searching best move at all positions. It goes very fast in depth search, with high pruning methods.
Personally I use ShashChess, with 4 or 8 cores, and with HighTal UCI option enabled (only this option). HighTal option forces ShashChess to explore a little more variants both sides (wider search) but at the expense of a little less depth : a lot more hard positions can be found than SF.
A good "analyser" engine must go a little wider in search than SF, but not too wide because it will not go far in depth and could be inefficient in a lot of situations. An engine like Chrystal has a very wide search but could not go very far in depth, unless you let it work 15 minutes.
Best regards.
Re: SF PB (Stockfish)
Posted: Thu Mar 14, 2024 2:06 pm
by Prometheus
@Skynet
Stockfish has never entered the World Championships, you mean the TCEC championships.
Re: SF PB (Stockfish)
Posted: Thu Mar 14, 2024 4:49 pm
by Solista
ConRog wrote: ↑Thu Mar 14, 2024 7:11 am
Solista wrote: ↑Thu Mar 14, 2024 2:50 am
It's not the actual search code. The problem is only threshold for Small Network. Threshold is set quite low and the small network has too much influence. But well, I have a different opinion than you. We don't have to agree. No problem. Have fun compiling.
Hi Solista,
Skynet is right, stockfish is the best engine for playing, and it is build for this purpose. It's not an engine for searching best move at all positions. It goes very fast in depth search, with high pruning methods.
Personally I use ShashChess, with 4 or 8 cores, and with HighTal UCI option enabled (only this option). HighTal option forces ShashChess to explore a little more variants both sides (wider search) but at the expense of a little less depth : a lot more hard positions can be found than SF.
A good "analyser" engine must go a little wider in search than SF, but not too wide because it will not go far in depth and could be inefficient in a lot of situations. An engine like Chrystal has a very wide search but could not go very far in depth, unless you let it work 15 minutes.
Best regards.
I created a version with changed settings for the small network (Big-SF 110324) and this version clearly beats SF dev with my test book with variants that we play on servers in practice. I prefer to use this version because it is clearly better for analysis. Download on my homepage. Take Stockfish dev and everything will be fine. I prefer the modified version. By the way, it work network with Dimension 3072. A new 3072 network is currently being tested on Fishtest today.
I compiled two test versions with the new 3072 network today. One is the one that is currently being tested on Fishtest against SF dev 2560 and one is with changed parameters for the small network. The rest identical. Download also on my homepage for anyone who would like to test it.
Re: SF PB (Stockfish)
Posted: Sun Mar 24, 2024 9:02 pm
by Skynet
SF-PB-200324_x64 "Add cmath header to movepick.h"
https://pixeldrain.com/u/ncNP6m9P