I know this is a bit simplistic, but I wonder if this is largely what it boils down to:
Stockfish sees far enough/fast enough that - more often than not - it can keep much of the edge one might think an NN engine like AllieStein would 'conjur' up to a minimum...and then as more pieces come off the board and we approach an endgame, the A/B approach of Stockfish begins to shine?
TCEC Superfinal - Stockfish leading AllieStein
Moderators: hgm, Rebel, chrisw
-
- Posts: 282
- Joined: Mon Sep 02, 2013 8:23 am
-
- Posts: 261
- Joined: Sun Nov 13, 2016 10:37 am
Re: TCEC Superfinal - Stockfish leading AllieStein
If you knew the difference between Allie and Leela, you would not ask this question.
Zevra 2 is my chess engine. Binary, source and description here: https://github.com/sovaz1997/Zevra2
Zevra v2.5 is last version of Zevra: https://github.com/sovaz1997/Zevra2/releases
Zevra v2.5 is last version of Zevra: https://github.com/sovaz1997/Zevra2/releases
-
- Posts: 708
- Joined: Mon Jan 16, 2012 6:34 am
Re: TCEC Superfinal - Stockfish leading AllieStein
-
- Posts: 3283
- Joined: Wed Mar 08, 2006 8:15 pm
Re: TCEC Superfinal - Stockfish leading AllieStein
Yesterday in endgame play SF had depth 161/161 and Allie 0/0 - some difference .
Jouni
-
- Posts: 282
- Joined: Mon Sep 02, 2013 8:23 am
Re: TCEC Superfinal - Stockfish leading AllieStein
But I did ask the question about this specific instance.Nay Lin Tun wrote: ↑Fri Oct 04, 2019 8:16 amIf you check around other websites/ tournments such as chess.com , you would not ask these questions.
-
- Posts: 708
- Joined: Mon Jan 16, 2012 6:34 am
Re: TCEC Superfinal - Stockfish leading AllieStein
Depth or Search is not that important when you have sound knowledge. That is why current 20x256 leela can play equally against 1:1000 faster Stockfish.
If T60 become really powerful, it may probably play against 1:10,000 faster Stockfish.
In theory, if you know the outcome of 10^43 positions , like 32 men TB, you just need depth 1 only. ( 7 men tablebases knows only depth 1.
In practice, if you know, a certrain trade leads certain endgame that you are sure about how to win (e.g KRP v KR), then you just need depth 1.( no need to search).
-
- Posts: 1470
- Joined: Mon Apr 23, 2018 7:54 am
-
- Posts: 282
- Joined: Mon Sep 02, 2013 8:23 am
Re: TCEC Superfinal - Stockfish leading AllieStein
My thoughts were my quick take after going thru a few of the games.Nay Lin Tun wrote: ↑Fri Oct 04, 2019 3:25 pm
In theory, if you know the outcome of 10^43 positions , like 32 men TB, you just need depth 1 only. ( 7 men tablebases knows only depth 1.
In practice, if you know, a certrain trade leads certain endgame that you are sure about how to win (e.g KRP v KR), then you just need depth 1.( no need to search).
So, are you verifying that AllieStein just doesn't play as well as Stockfish when play becomes more concrete later in the middlegame and towards and into the endgame?