Fat Fritz 2

[Sylwy]

Hi, thanks: Whats the difference between FatFritz2.exe and FatFritz2mycompile.exe.
Whats the version of the included FatFritz2_v1.bin? Is it official or the one from github that some say is older version?
thanks

This weights file is surprisingly strong. I have my doubts if the Chessbase version is really stronger, not impossible that CB has mixed up the NNUE files and now sells the weaker one embedded in the engine, while the stronger is open src in Github

+10000000.............0
The apotheosis of marketing !

[carldaman]

This FF2 is a royal huckstering (con)job. Very disappointing.

Maybe they can truly improve on it later on, where the claims can have substance, but if so, they should have better waited until such a time.

[gerhard]

... with this case Chessbase has driven itsself into massively bad reputation !!!

[twobeer]

A.Silver said, that the github net is weaker. And it is obvious, that this must be true. If not, nobody needed to buy FF2, so it would be crazy, if Chessbase would provide the strong commercial net on github for free.

hmm.. That would be just as crazy as selling a chess-enginge including network that is weaker than free SF versions ripping the same source, and claim it is stronger, but certainly not impossible... Can you help out with the MD5s of these network so we can actually verify that differnet rating lists are using the same net...

WIthout proper info on actual network versions used it all becomes a trust, mumbo-jumo issue.. Even for Pohls rating-list.

[Gabor Szots]

WIthout proper info on actual network versions used it all becomes a trust, mumbo-jumo issue.. Even for Pohls rating-list.

In a very short match (20 games) with 30s+0.2s I obtained this result:

Score of Fat_Fritz_2-x64 (commercial) vs Fat_Fritz_2-x64 (github net): 6 - 0 - 14 [0.650]
Elo difference: 107.5 +/- 78.2, LOS: 99.3 %, DrawRatio: 70.0 %
20 of 20 games finished.

Maybe not convincing in itself but a strong indication that the downloadable NNUE is much weaker.

[connor_mcmonigle]

Yes. The free, GitHub, network provided with the source is likely to be significantly weaker than the commercial network in my testing. By separating the network weights from the provided binary, there is no longer a GPL violation. However, the already distributed integrated binaries with GPL attached remain problematic legally, in my understanding...

Below are the results of my testing with the 4moves_noob (known to inflate differences in elo) opening book against SF dev at 6s+0.06s TC with 5 man adjudication. The results are consistent with a far weaker network:

Code: Select all

stockfish-ff2net bench:

Total time (ms) : 1995
Nodes searched  : 5156767
Nodes/second    : 2584845



stockfish-masternet bench:

Total time (ms) : 1745
Nodes searched  : 3766422
Nodes/second    : 2158408

Code: Select all

   # PLAYER                 : RATING  ERROR   POINTS  PLAYED    (%)
   1 stockfish-masternet    :    0.0   ----    611.0     974   62.7%
   2 stockfish-ff2net       :  -91.2   15.8    363.0     974   37.3%

[gaard]

Yes. The free, GitHub, network provided with the source is likely to be significantly weaker than the commercial network in my testing. By separating the network weights from the provided binary, there is no longer a GPL violation. However, the already distributed integrated binaries with GPL attached remain problematic legally, in my understanding...

Below are the results of my testing with the 4moves_noob (known to inflate differences in elo) opening book against SF dev at 6s+0.06s TC with 5 man adjudication. The results are consistent with a far weaker network:

Code: Select all

   # PLAYER                 : RATING  ERROR   POINTS  PLAYED    (%)
   1 stockfish-masternet    :    0.0   ----    611.0     974   62.7%
   2 stockfish-ff2net       :  -91.2   15.8    363.0     974   37.3%

By separating the network weights from the provided binary, there is no longer a GPL violation.

Not quite. If the NN doesn't allow you to build an executable that is functionally identical to what is being distributed by ChessBase, then the GPL violation remains. Sleight-of-hand makes their case look worse.

[connor_mcmonigle]

By separating the network weights from the provided binary, there is no longer a GPL violation.

Not quite. If the NN doesn't allow you to build an executable that is functionally identical to what is being distributed by ChessBase, then the GPL violation remains. Sleight-of-hand makes their case look worse.

Yes. This sleight of hand doesn't look good ethically, but, of course, none of this does in the first place. By separating the weights from the executable, ChessBase is now selling two separate items when one purchases FF2: the weights and the executable. The conveyed executable can be, now, exactly reproduced by compiling the provided GitHub source and, therefore, future sales of FF2 won't be a GPL violation in my understanding. Previous sales of the integrated binary were likely a GPL violation though.

[gaard]

By separating the network weights from the provided binary, there is no longer a GPL violation.

Not quite. If the NN doesn't allow you to build an executable that is functionally identical to what is being distributed by ChessBase, then the GPL violation remains. Sleight-of-hand makes their case look worse.

Yes. This sleight of hand doesn't look good ethically, but, of course, none of this does in the first place. By separating the weights from the executable, ChessBase is now selling two separate items when one purchases FF2: the weights and the executable. The conveyed executable can be, now, exactly reproduced by compiling the provided GitHub source and, therefore, future sales of FF2 won't be a GPL violation in my understanding. Previous sales of the integrated binary were likely a GPL violation though.

I don't follow. If the publicly available NN is not equivalent to the one distributed by CB then how can you exactly reproduce what is distributed by CB? Or, are you under the assumption that the publicly available NN is equivalent to the embedded one that CB is selling?

[Graham Banks]

Latest from the blitz list.