Date: Mon Jun 20 08:30:57 2022 +0200
Timestamp: 1655706657);( good calibrated evaluation.)
Use NNUE complexity in search, retune related parameters
This builds on ideas of xoto10 and mstembera to use more output from NNUE in the search algorithm.
Great Job! Should be SF 15.1.
(I did bot like SF 15 Master. :D)
Stockfish rules again.
|Hi..maybe you could change to gcc 11.3.1 ?
Author: Giacomo Lorenzetti
Date: Sun May 29 18:54:19 2022 +0200
Small speedup in futility_move_count
The speedup is around 0.25% using gcc 11.3.1 (bmi2, nnue bench, depth 16
and 23) while it is neutral using clang (same conditions).
|Why not convert it all to increment so that engines will have a certain amount of time per move?
Unless we are testing time management, that sounds like a better option to me because then
the engine won't make mistakes because of not having enough time...
Probably the devs have a good reason for that though !
|i think the time control should be reduce at stc from 10+100ms to 6+160ms this way more games could be played because i think the engine quality of moves is more determine by the increment than the total game duration|
|But I still haven't found an engine that suits me Fischer's style.
A kind of evolutionary Terminator who in 1970 took apart the Soviet school by himself.
Then if withdrawn like a prophet as to say : if some Nasa genius in 100 years will permit to some Christmas tree playing chess remember, in his birth time, to mention my name and call iy like , Psfish, Stickfish, Pfufish or I don't know how ...
the service offered by this site is excellent,
however I ask if you can make available a simple configurator module that allows you to read and configure the parameters of the Stockfish chess engine exe file (Hash, Ponder, SyzygyPath, EvalFile NNUE Path); in this way each user can use the exe file of the customized chess engine also in other applications without necessarily going through the standard GUI such as Arena, Tarrash, Cute Chess .....
thanks in advance for any reply
Greetings from Marco's Italy
|> it accepted my draw offer
engines can not receive offers. UCI protocol lack such commands. You offered draw to GUI and your GUI accepted it.
|33362||Mark 1880 chess dot com|
|Played against the latest development version of stockfish as of May 03, 2022, it accepted my draw offer despite it being in the winning position (which is unusual as it has never accepted a draw whenever I offer it out of fun) . I used it on the Fat Fritz 2 GUI updated to the latest GUI (version 30) and for its opening I used the Hiarcs 15d opening book optimized. Time control 5mins + 2 secs per move, no handicap whatsoever, played it fair and square.
1. e4 e5 2. d4 exd4 3. Qxd4 Nc6 4. Qd1 Nf6 5. Bd3 d5 6. Bg5 dxe4 7. Bb5 Qxd1+ 8. Kxd1 Ng4 9. Bxc6 bxc6 10. Nh3 h6 11. Bf4 Bc5 12. f3 exf3 13. gxf3 O-O 14. Kd2 Nf6 15. Ng1 Nd5 16. Ne2 Rd8 17. Ke1 Re8 18. Kd1 Ba6 1/2-1/2
|My chess engine rates 1427 ELO on CCRL *~*|
|So it shouldn't be hard to jump straight to Stockfish 17 or 18. or keep the two branches separate but interconnected.
It would be exciting ,,,
|Why not make a version of stockfish that uses a different gpu, NVIDIA?
Then let's compare the traditional Stockfish and the gpu one ...
|yes, not is missed my message|
|I would prefer to be cited correctly.
My message was:
Koivisto is definetely not stronger than Stockfish 15. We have made many many tests and ran hundreds of thousands games. We also refer to CCRL, TCEC and CCC which all show that SF15 is the strongest engine that exists.
And yes, Koivisto 8 does use a neural network. It's not the same as stockfish but related.
|Any one check Koivisto latest 8.0 ? its 3570 elo ,I checked 25 game against SF dev build candirufish version , result 24 draw and 1 win koivisto ,see author germen engineer messaged me below
Koivisto is definitely stronger than Stockfish 15. We have made many
many tests and ran hundreds of thousands games. but We refer to CCRL,
TCEC and CCC, which all show that SF15 is the strongest engine that exists.
And yes, Koivisto 8 does use a neural network. It’s not the same as
stockfish but related.
|yeah +26-8 is pretty brutal for a final
You're right an easy win is less thrilling than a battle down to the wire
I thought it had to do with not having any suspense. Stockfish just demolished komodo...
If it was closer, with komodo winning, we may have had stockfish fans griefing and then
more celebration when stockfish won a game. Perhaps many fans went away, I don't know...
Perhaps new ones came around.
This looks like the biggest difference in tcec history(perhaps it's not, I don't know)
|@sad13: It's quite a common thing. The journey to the top is always more thrilling than staying at the top. I remember in 2014/2015 when SF finally managed to win TCEC (+13-7 iirc), it was amazing. But now that it's winning everything back to back, we get used to see it winning.
also don't forget there was a lot of griefing here, which made reasonable people go away at some point hence less fans.
|@ jayakumar987 - Thank you! That site is a very useful resource and perfectly answers my question. One day later, Stockfish 15 comes out so maybe I was on to something. Thanks again. :)|
|Just around the corner?
|@ Dev f
|Anyone have an ELO estimate on the current build? I'm curious about the gain since the release of 14.1.
If we're talking 25 points or more, is that enough to release the current build as "Stockfish 15"?
|when stockfish wasn't as much better than komodo as it is today I remember more cheering for sf|
|33347||Lalit Mohan Bankoti|
|@sad13,we know stockfish 14.1 is strongest chess engine.komodo is a kid in front of sf.actually we are waiting for sf 15.then we will cheer together.|
|I am surprised there isn't cheering about stockfish from stockfish fans here.
It's doing fantastic against komodo right now in tcec.
|stockfish is doing fantastic right now in tcec S22 superfinal !|