this will be the merge of a lifetime : SF 80 Elo+

[syzygy]

Can someone explain this project to me in small words? I gathter it is basically Stockfish with a NN eval, or? And is the neural net part running on CPU, or GPU? Apologies, I have not been following the lengthy threads on this.

Elevator speech - Go programers ( from Japan) took the SF search code and combined it with their GO NN eval and it became quickly one of the best Go engines in the world. They then took their Go engine and converted to a chess NN engine and , got it working , came back to the SF team and saids, "you guys might want take a look at this. it is already playing near the level cur-dev-Stockfish" .. the rest is history as they say .

Caveat - I really do not know the backstory, this is my conjecture from everything I have read so far. The quote above is not real.

You probably meant to write Shogi, not Go.

[MikeB]

Can someone explain this project to me in small words? I gathter it is basically Stockfish with a NN eval, or? And is the neural net part running on CPU, or GPU? Apologies, I have not been following the lengthy threads on this.

Elevator speech - Go programers ( from Japan) took the SF search code and combined it with their GO NN eval and it became quickly one of the best Go engines in the world. They then took their Go engine and converted to a chess NN engine and , got it working , came back to the SF team and saids, "you guys might want take a look at this. it is already playing near the level cur-dev-Stockfish" .. the rest is history as they say .

Caveat - I really do not know the backstory, this is my conjecture from everything I have read so far. The quote above is not real.

You probably meant to write Shogi, not Go.

Correct.

[Raphexon]

Can someone explain this project to me in small words? I gathter it is basically Stockfish with a NN eval, or? And is the neural net part running on CPU, or GPU? Apologies, I have not been following the lengthy threads on this.

Yes, with the NN running on CPU.

Apparently the NN is 20MB but using it in the eval only halves nps, which I find quite remarkable. I haven't tried to understand the code yet, but it seems to rely heavily on vector instructions (unsurprisingly).

So it turns out that NNs are much better than humans at writing evaluation functions (even taking into account speed of execution).

It only updates parts incrementally when the board state changes. That way it doesn't need to recalculate the whole net every time.

And while 20MB, it has a lot of functionally useless features. Can't compare it 1:1 with Leela.
There are also millions of different ways to encode information inside the NN, speed is affected depending on how much it has to recalculate each ply searched.
See below for examples:
https://github.com/tttak/Stockfish/tree ... hitectures

halfkp-mobility_256x2-32-32 doesn't add a lot of size, but it causes an enormous slowdown.
halfkpe4 (which I don't really understand) quadruples size but only introduces a very minor slowdown.

[Cornfed]

This is the breakthrough I've been predicting for years! It finally happened, and I'm just glad chess looks very healthy after it (draw rates have gone down! Some lines are proving to be playable again!)

Until everyone uses these?

[mwyoung]

Is that at LTC?

It scales very well.

OK.. It scales very well...?!

Re: Stockfish NNUE (Sergio 2138) vs Stockfish 170720 (32 Thread Test) (TC = 30m+30s)
Post by mwyoung » Thu Aug 06, 2020 4:31 am

End of Match.

DESKTOP-CORSAIR, Rapid 30.0min+30.0sec 0

1 SF+NNUE PO 290720 x64 popc +17 +7/=95/-2 52.40% 54.5/104
2 Stockfish 170720 64 POPCNT -17 +2/=95/-7 47.60% 49.5/104

[Laskos]

Is that at LTC?

It scales very well.

OK.. It scales very well...?!

Re: Stockfish NNUE (Sergio 2138) vs Stockfish 170720 (32 Thread Test) (TC = 30m+30s)
Post by mwyoung » Thu Aug 06, 2020 4:31 am

End of Match.

DESKTOP-CORSAIR, Rapid 30.0min+30.0sec 0

1 SF+NNUE PO 290720 x64 popc +17 +7/=95/-2 52.40% 54.5/104
2 Stockfish 170720 64 POPCNT -17 +2/=95/-7 47.60% 49.5/104

It's your fault having 91%+ draw rate. Chose better openings.

[mwyoung]

Is that at LTC?

It scales very well.

OK.. It scales very well...?!

Re: Stockfish NNUE (Sergio 2138) vs Stockfish 170720 (32 Thread Test) (TC = 30m+30s)
Post by mwyoung » Thu Aug 06, 2020 4:31 am

End of Match.

DESKTOP-CORSAIR, Rapid 30.0min+30.0sec 0

1 SF+NNUE PO 290720 x64 popc +17 +7/=95/-2 52.40% 54.5/104
2 Stockfish 170720 64 POPCNT -17 +2/=95/-7 47.60% 49.5/104

It's your fault having 91%+ draw rate. Chose better openings.

Book is set to 6 moves. The same settings as when I got a huge win factor over Stockfish also at short time controls. Sorry you need a new theory! And you need to do the leg work before advertising NNUE scales very well!

[Laskos]

Is that at LTC?

It scales very well.

OK.. It scales very well...?!

Re: Stockfish NNUE (Sergio 2138) vs Stockfish 170720 (32 Thread Test) (TC = 30m+30s)
Post by mwyoung » Thu Aug 06, 2020 4:31 am

End of Match.

DESKTOP-CORSAIR, Rapid 30.0min+30.0sec 0

1 SF+NNUE PO 290720 x64 popc +17 +7/=95/-2 52.40% 54.5/104
2 Stockfish 170720 64 POPCNT -17 +2/=95/-7 47.60% 49.5/104

It's your fault having 91%+ draw rate. Chose better openings.

Book is set to 6 moves. The same settings as when I got a huge win factor over Stockfish also at short time controls. Sorry you need a new theory! And you need to do the leg work before advertising NNUE scales very well!

Even your crappy test shows +7 -2, an excellent W/L ratio, probably denoting that at 60% draw rate, the difference could easily be 80 Elo points.

[mwyoung]

Is that at LTC?

It scales very well.

OK.. It scales very well...?!

Re: Stockfish NNUE (Sergio 2138) vs Stockfish 170720 (32 Thread Test) (TC = 30m+30s)
Post by mwyoung » Thu Aug 06, 2020 4:31 am

End of Match.

DESKTOP-CORSAIR, Rapid 30.0min+30.0sec 0

1 SF+NNUE PO 290720 x64 popc +17 +7/=95/-2 52.40% 54.5/104
2 Stockfish 170720 64 POPCNT -17 +2/=95/-7 47.60% 49.5/104

It's your fault having 91%+ draw rate. Chose better openings.

Book is set to 6 moves. The same settings as when I got a huge win factor over Stockfish also at short time controls. Sorry you need a new theory! And you need to do the leg work before advertising NNUE scales very well!

Even your crappy test shows +7 -2, an excellent W/L ratio, probably denoting that at 60% draw rate, the difference could easily be 80 Elo points.

It could be, and if my grandma had wheels she would be a wagon. Stockfish NNUE scales good enough. The point is you never tested at LTC. So you don't know anything as usual.

[Raphexon]

Is that at LTC?

It scales very well.

OK.. It scales very well...?!

Re: Stockfish NNUE (Sergio 2138) vs Stockfish 170720 (32 Thread Test) (TC = 30m+30s)
Post by mwyoung » Thu Aug 06, 2020 4:31 am

End of Match.

DESKTOP-CORSAIR, Rapid 30.0min+30.0sec 0

1 SF+NNUE PO 290720 x64 popc +17 +7/=95/-2 52.40% 54.5/104
2 Stockfish 170720 64 POPCNT -17 +2/=95/-7 47.60% 49.5/104

It's your fault having 91%+ draw rate. Chose better openings.

Book is set to 6 moves. The same settings as when I got a huge win factor over Stockfish also at short time controls. Sorry you need a new theory! And you need to do the leg work before advertising NNUE scales very well!

Even your crappy test shows +7 -2, an excellent W/L ratio, probably denoting that at 60% draw rate, the difference could easily be 80 Elo points.

NNUE's "lack of scaling" posts can be summarized in just a few words: Coin flipping and elo compression.