Commercial Release of Ethereal 13.50

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

AndrewGrant
Posts: 1787
Joined: Tue Apr 19, 2016 6:08 am
Location: U.S.A
Full name: Andrew Grant

Commercial Release of Ethereal 13.50

Post by AndrewGrant »

Please see: http://chess.grantnet.us/Ethereal/ for the non-technical information

Owners of Ethereal 13.00 & 13.25 should have received emails by now containing a link to download the updated binaries. If you fall into that category, and have not received an email from andrewgrantethereal@gmail.com, please first check your spam folder and then reach out to me via andrewgrantethereal@gmail.com or andrew@grantnet.us. I would prefer to provide any support needed via email, and not here on talkchess.

Code: Select all

13.50: Fix an extremely rare crash case first seen by the CCRL team
13.49: During MultiPV mode, report the lines in sorted order after each depth
13.48: Increase the threshold for using HCE instead of NNUE to 2000cp
13.47: Allow fail-highs to be voted on by thread selection during SMP play
13.46: Allow back-to-back Null Moves for the same side
13.45: Expand the NNUE cap from [-1000, 1000] to [-2000, 2000]
13.44: Don't needlessly apply/rever the TT move during Multi-Cut
13.43: Change some pruning conditions to better detect mates in TB positions
13.42: Save information into the TT after a Probcut under some conditions
13.41: Increase R during NMP when we are following a tacical move
13.40: Move the Thread's Stack State to its own encapsulated Structure (SPEED)
13.39: Index continuation History by the tactical state of the move (Stockfish)
13.38: Remove evaluation conditions from ProbCut
13.37: Short-circuit the QSearch when SEE suggests we shatter the beta window
13.36: Modify Ethereal's internal PGN processor to only handle stripped PGNs
13.35: Scale up the Midgame by 40% during NNUE evaluation, scale down the Endgame
13.34: Update the default NNUE file to one trained with a lower LR
13.33: Update the default NNUE file to one trained with a lower LR
13.32: Update the default NNUE file to one using 2x192 instead of 2x256
13.31: Update the default NNUE file to one trainer with a lower LR
13.30: Small Makefile fix to handle Linux cases where USER= is defined accidentally
13.29: Update the default NNUE file to one trained using Mirroed HKP
13.28: Update the makefile to produce release binaries more easily
13.27: Remove the ready lock to comply with the UCI protocol
13.26: Butfix for UCI options when playing under Cutechess
13.25: Force abort when compiled for NNUE, but no NNUE file is given

Code: Select all

// Standard chess using a very drawish book
ELO   | 42.79 +- 1.69 (95%)
CONF  | 60.0+0.60s Threads=1 Hash=64MB
GAMES | N: 40000 W: 7550 L: 2648 D: 29802
http://chess.grantnet.us/test/23535/

// Fischer Random Chess using a very shallow book
ELO   | 125.65 +- 2.30 (95%)
CONF  | 60.0+0.60s Threads=1 Hash=64MB
GAMES | N: 40000 W: 16368 L: 2501 D: 21131
http://chess.grantnet.us/test/23536/
Ethereal 13.50 can already been seen on CCRL's Fischer Random Chess rating list, having reestablished itself as the gate keeper between Stockfish, Komodo, Leela, and the remainder of the field.

Image

---

While Ethereal without NNUE still exists, and can be compiled directly from the source with ease, I have not provided binary files. This is because the strength of NNUE makes the standard version obsolete for high tier game-play, like rating lists, tournaments, and correspondence analysis.
#WeAreAllDraude #JusticeForDraude #RememberDraude #LeptirBigUltra
"Those who can't do, clone instead" - Eduard ( A real life friend, not this forum's Eduard )
peter
Posts: 3187
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Commercial Release of Ethereal 13.50

Post by peter »

AndrewGrant wrote: Fri Jan 21, 2022 12:31 am Owners of Ethereal 13.00 & 13.25 should have received emails by now containing a link to download the updated binaries.
Thanks a lot for update and email, dear Andrew!
Peter.
Krzysztof Grzelak
Posts: 1534
Joined: Tue Jul 15, 2014 12:47 pm

Re: Commercial Release of Ethereal 13.50

Post by Krzysztof Grzelak »

Thank you very much Andrew.
User avatar
Rubinus
Posts: 1161
Joined: Thu Jan 18, 2007 4:05 pm
Location: Prague
Full name: Pavel Háse

Re: Commercial Release of Ethereal 13.50

Post by Rubinus »

Thank you and good luck.
User avatar
pohl4711
Posts: 2457
Joined: Sat Sep 03, 2011 7:25 am
Location: Berlin, Germany
Full name: Stefan Pohl

Re: Commercial Release of Ethereal 13.50

Post by pohl4711 »

Ratinglist-testrun of Ethereal 13.50 finished. Thanx to A.Grant for sending me a free version for testing...


https://www.sp-cc.de

(Perhaps you have to clear your browsercache or reload the website)
Damir
Posts: 2804
Joined: Mon Feb 11, 2008 3:53 pm
Location: Denmark
Full name: Damir Desevac

Re: Commercial Release of Ethereal 13.50

Post by Damir »

Thanks for the update and e-mail Andrew. :) :) :D
User avatar
mclane
Posts: 18762
Joined: Thu Mar 09, 2006 6:40 pm
Location: US of Europe, germany
Full name: Thorsten Czub

Re: Commercial Release of Ethereal 13.50

Post by mclane »

Thanks for the update
What seems like a fairy tale today may be reality tomorrow.
Here we have a fairy tale of the day after tomorrow....
kasinp
Posts: 251
Joined: Sat Dec 02, 2006 10:47 pm
Location: Toronto
Full name: Peter Kasinski

Re: Commercial Release of Ethereal 13.50

Post by kasinp »

Great news, thank you Andrew.

A couple of questions; answers may benefit others:

a) assuming the CPU can handle all, which of the included compiles is best? I suspect avx2, but then what about pext-avx2?
b) would bmi2 be another option or is it equivalent to one of the included versions?

Thanks again,
PK
bastiball
Posts: 5341
Joined: Tue Oct 20, 2020 4:18 am
Full name: Basti Dangca

Re: Commercial Release of Ethereal 13.50

Post by bastiball »

kasinp wrote: Fri Jan 21, 2022 4:47 pm Great news, thank you Andrew.

A couple of questions; answers may benefit others:

a) assuming the CPU can handle all, which of the included compiles is best? I suspect avx2, but then what about pext-avx2?
b) would bmi2 be another option or is it equivalent to one of the included versions?

Thanks again,
PK
I think bmi and pext are the same.
Basti Dangca
CCRL testing group
Vernon Crawford
Posts: 73
Joined: Wed Sep 01, 2021 2:05 am
Location: London, England
Full name: Vernon Crawford

Re: Commercial Release of Ethereal 13.50

Post by Vernon Crawford »

Congrats on the new release & the Elo gain...

I appreciate commercial engines (especially Komodo, Shredder, & Hiarcs),
but with all the bugs, the unprofessionalism, and the fact that you come across somewhat arrogant
https://talkchess.com/forum3/viewtopic. ... 3&start=44
is a real turnoff.

I think I'll pass.