Last Update: 22.12.2021
Hello Chess Friends,
I am pleased to announce,I managed to organize another great Duel too:NNUE (No Book) vs NON-NNUE (Deep Books)Meanwhile, periodically I run similar Duels: I II
And 2nd name of this Duel can be called:Battle of Brains !)
A few quick notes about current Duel: NNUE Vs BookAll Deep Books are played by CF EXT 291120 NON-NNUEAs Players, I picked the Top Books of CS LVI Super League
As NNUE (No Book), 3 (three) engines are used:Kayra 1.0 v6, Stockfish 171221, SugaR AI 2.50No Book = Played without using any opening bookNNUE = Played with their own default EvalFiles
In case of comparing NNUE (No Book) vs NON-NNUE (Deep Books)Now we see that, the Elo difference is over than 70 Elo...Impressive!For example, in one of previous Duels match: NNUE vs NON-NNUEBlitz Elo diff. about 150 Elo, but played via Perfect2022 (up to 8 moves)So with Deep books at around 70+ Elo is quite normal, no BIG surprise...It seems, NNUE's gained points are come mainly from middle games!However, once more we noticed that,Via SCCT, there is no any 100+ Elo 200+ Elo or 300+ Elo for sure )Because, I am doing my best for more fair conditions (for both sides)
Meanwhile, just I'd like to point out that,The 1st tour start was with TC 3 Minutes...but after checking...I noticed that both Top chess engines as NNUE are lost on time...E.g Stockfish 171221 and SugaR AI 2.5 lost on time at TC 3 MinAnd just in 33 games: 3 games lost on time by Stockfish...strange..Plus these time loses are not after 250 moves, about in 100 movesBtw, for anyone interested, SF buggy games available: DownloadPlus, a screenshot (the Stockfish version, which played) for DetailsNote: Kayra seems to be more stable, no any game lost on timeIn short, I canceled the 3 Min tour, and I switched to 2min+1secAnd with newer time control, all these time loses are disappearedOtherwise, Deep Books would be rated 70+ Elo...just saying...))Once more...please check more carefully...before any releases...Because, stability is most important part... at least for my side!It seems Fishtest have to check/test with more various conditionsSimply, what about as MP, Ponder ON, Blitz time control e.g 3 Min? And very sad that some of my favorite engines have such bugs...I hope the Top engine authors will fix ..in case of newer releasesThank you in advance...
I have to admit also that,No doubt that latest NNUE (Evalfiles) seems to be super strong!Sure, same words are going for the Deep opening books as well...Because, vs NNUE: forget to take a win even draw is a BIG success!
Duel's individual stats about players, which gained highest points:
Best NNUE Engine Performance: Kayra 1.0 v6 (by Mehmet Karaman)Best Opening Book Performance: NimaTiv-7 (by Ian Hurnavich)
Btw, finely DON book lost the lead...)
The Winner: NNUE - Congrats to all (including book authors as well)
# PLAYER : RATING POINTS GAMES (%) 1 NNUE (No Book) : 3786.9 221.5 360 61.5% 2 DON Free 4 : 3713.6 260.0 510 51.0% 3 Chucaro 041221 : 3708.0 271.0 540 50.2% 4 New Database : 3706.7 270.0 540 50.0% 5 NimaTiv-7 : 3706.4 317.5 630 50.4% 6 1337chess X Pro : 3706.4 239.0 480 49.8% 7 LC-X 10.0M2 : 3705.6 162.0 330 49.1% 8 Fauzi Lite 0.6 : 3705.3 316.5 630 50.2% 9 MANTIS X2 : 3703.6 227.5 450 50.6% 10 Guzel Kiz 5 (2) : 3702.0 298.5 600 49.8% 11 MANTIS 051221 : 3701.9 190.5 390 48.8% 12 1337chess 210721 : 3701.1 312.5 630 49.6% 13 Goi-light : 3700.8 174.5 360 48.5% 14 DONFupdate2 : 3700.2 196.0 390 50.3% 15 Bronco : 3700.0 225.0 450 50.0% 16 Chucaro 2 : 3696.4 208.0 420 49.5% 17 Bronco 061221 : 3696.1 187.0 390 47.9% 18 NimaTiv-8 : 3695.4 175.5 360 48.8% 19 CHESSBOT : 3693.1 145.0 300 48.3% 20 LC-X 9.9a3 : 3691.9 147.5 300 49.2% 21 RadaR 050121 : 3685.5 144.5 300 48.2% 22 Pure Sarcasm : 3677.0 140.5 300 46.8%
Games
Database contains 360 games:1 -120 games = played by Kayra121-240 games = played by Stockfish241-360 games= played by SugaR
Have fun,Sedat
Hello Chess Friends,
I am pleased to announce,I managed to organize another great Duel too:NNUE (No Book) vs NON-NNUE (Deep Books)Meanwhile, periodically I run similar Duels: I II
And 2nd name of this Duel can be called:Battle of Brains !)
A few quick notes about current Duel: NNUE Vs BookAll Deep Books are played by CF EXT 291120 NON-NNUEAs Players, I picked the Top Books of CS LVI Super League
As NNUE (No Book), 3 (three) engines are used:Kayra 1.0 v6, Stockfish 171221, SugaR AI 2.50No Book = Played without using any opening bookNNUE = Played with their own default EvalFiles
In case of comparing NNUE (No Book) vs NON-NNUE (Deep Books)Now we see that, the Elo difference is over than 70 Elo...Impressive!For example, in one of previous Duels match: NNUE vs NON-NNUEBlitz Elo diff. about 150 Elo, but played via Perfect2022 (up to 8 moves)So with Deep books at around 70+ Elo is quite normal, no BIG surprise...It seems, NNUE's gained points are come mainly from middle games!However, once more we noticed that,Via SCCT, there is no any 100+ Elo 200+ Elo or 300+ Elo for sure )Because, I am doing my best for more fair conditions (for both sides)
Meanwhile, just I'd like to point out that,The 1st tour start was with TC 3 Minutes...but after checking...I noticed that both Top chess engines as NNUE are lost on time...E.g Stockfish 171221 and SugaR AI 2.5 lost on time at TC 3 MinAnd just in 33 games: 3 games lost on time by Stockfish...strange..Plus these time loses are not after 250 moves, about in 100 movesBtw, for anyone interested, SF buggy games available: DownloadPlus, a screenshot (the Stockfish version, which played) for DetailsNote: Kayra seems to be more stable, no any game lost on timeIn short, I canceled the 3 Min tour, and I switched to 2min+1secAnd with newer time control, all these time loses are disappearedOtherwise, Deep Books would be rated 70+ Elo...just saying...))Once more...please check more carefully...before any releases...Because, stability is most important part... at least for my side!It seems Fishtest have to check/test with more various conditionsSimply, what about as MP, Ponder ON, Blitz time control e.g 3 Min? And very sad that some of my favorite engines have such bugs...I hope the Top engine authors will fix ..in case of newer releasesThank you in advance...
I have to admit also that,No doubt that latest NNUE (Evalfiles) seems to be super strong!Sure, same words are going for the Deep opening books as well...Because, vs NNUE: forget to take a win even draw is a BIG success!
Duel's individual stats about players, which gained highest points:
Best NNUE Engine Performance: Kayra 1.0 v6 (by Mehmet Karaman)Best Opening Book Performance: NimaTiv-7 (by Ian Hurnavich)
Btw, finely DON book lost the lead...)
The Winner: NNUE - Congrats to all (including book authors as well)
# PLAYER : RATING POINTS GAMES (%) 1 NNUE (No Book) : 3786.9 221.5 360 61.5% 2 DON Free 4 : 3713.6 260.0 510 51.0% 3 Chucaro 041221 : 3708.0 271.0 540 50.2% 4 New Database : 3706.7 270.0 540 50.0% 5 NimaTiv-7 : 3706.4 317.5 630 50.4% 6 1337chess X Pro : 3706.4 239.0 480 49.8% 7 LC-X 10.0M2 : 3705.6 162.0 330 49.1% 8 Fauzi Lite 0.6 : 3705.3 316.5 630 50.2% 9 MANTIS X2 : 3703.6 227.5 450 50.6% 10 Guzel Kiz 5 (2) : 3702.0 298.5 600 49.8% 11 MANTIS 051221 : 3701.9 190.5 390 48.8% 12 1337chess 210721 : 3701.1 312.5 630 49.6% 13 Goi-light : 3700.8 174.5 360 48.5% 14 DONFupdate2 : 3700.2 196.0 390 50.3% 15 Bronco : 3700.0 225.0 450 50.0% 16 Chucaro 2 : 3696.4 208.0 420 49.5% 17 Bronco 061221 : 3696.1 187.0 390 47.9% 18 NimaTiv-8 : 3695.4 175.5 360 48.8% 19 CHESSBOT : 3693.1 145.0 300 48.3% 20 LC-X 9.9a3 : 3691.9 147.5 300 49.2% 21 RadaR 050121 : 3685.5 144.5 300 48.2% 22 Pure Sarcasm : 3677.0 140.5 300 46.8%
Games
Database contains 360 games:1 -120 games = played by Kayra121-240 games = played by Stockfish241-360 games= played by SugaR
Have fun,Sedat