Future of computer chess: 2 top engines?
Posted: Mon Oct 12, 2020 1:40 pm
I fear, that soon we have only 2 top engines. Stockfish + derivates and Lc0 + derivates. I hope to be wrong .
It has really been the case since mid to late 2018 when Komodo and Houdini (which was later discovered to be a Stockfish clone) stalled on progress and no other engine except for Ethereal was able to catch up to Komodo and Houdini before NNUE (and now Ethereal has stopped development which is a shame). Meanwhile Stockfish and Leela strength skyrocketed far beyond anything seen before, leaving all other engines in the dust. Since then, the only engines around Stockfish/Leela strength are derivatives like Allie, Crystal, and Shashchess, and the private engine Stoofvlees which really doesn't count.
I guess that winning contests will be divided between those two for a while.
On the other hand, NNUE really makes it easy for chess engines to gain a lot of elo, Nemorino already reached Komodo/Houdini level just by applying NNUE, so if an engine with a stronger search like Xiphos or rofchade were to implement NNUE they could reach Leela's level I feel. Unless you feel NNUE causes an engine to be a Stockfish derivative.Dann Corbit wrote: ↑Mon Oct 12, 2020 5:40 pm I guess that winning contests will be divided between those two for a while.
But computer chess is doing fine. Look at Graham's contests on the tournament forum.
There are a boatload of actively developed engines. They just are not Elo monsters.
Komodo 11.2 is only 26 Elo points behind version 14, at CCRL 40/15. It was released on July 17, 2017. SF had been the clear number one for over a year, by the time Lc0 matured, and now it is again; this last T60 run has been rather disappointing, not to mention incredibly long.Madeleine Birchfield wrote: ↑Mon Oct 12, 2020 5:11 pmIt has really been the case since mid to late 2018 when Komodo and Houdini (which was later discovered to be a Stockfish clone) stalled on progress and no other engine except for Ethereal was able to catch up to Komodo and Houdini before NNUE (and now Ethereal has stopped development which is a shame). Meanwhile Stockfish and Leela strength skyrocketed far beyond anything seen before, leaving all other engines in the dust. Since then, the only engines around Stockfish/Leela strength are derivatives like Allie, Crystal, and Shashchess, and the private engine Stoofvlees which really doesn't count.
There is new technology just around the corner that is going to breathe new life into the LC0 category of programs.Ozymandias wrote: ↑Mon Oct 12, 2020 6:14 pmKomodo 11.2 is only 26 Elo points behind version 14, at CCRL 40/15. It was released on July 17, 2017. SF had been the clear number one for over a year, by the time Lc0 matured, and now it is again; this last T60 run has been rather disappointing, not to mention incredibly long.Madeleine Birchfield wrote: ↑Mon Oct 12, 2020 5:11 pmIt has really been the case since mid to late 2018 when Komodo and Houdini (which was later discovered to be a Stockfish clone) stalled on progress and no other engine except for Ethereal was able to catch up to Komodo and Houdini before NNUE (and now Ethereal has stopped development which is a shame). Meanwhile Stockfish and Leela strength skyrocketed far beyond anything seen before, leaving all other engines in the dust. Since then, the only engines around Stockfish/Leela strength are derivatives like Allie, Crystal, and Shashchess, and the private engine Stoofvlees which really doesn't count.
AMD cards need a lot of power, that means less potatoes I would prefer if the software side would allow me to better profit my rather new 1660s.Dann Corbit wrote: ↑Mon Oct 12, 2020 6:19 pm There is new technology just around the corner that is going to breathe new life into the LC0 category of programs.
AMD is creating a topology where the GPUs and CPUs can have syncrhonized access to the same memory sources.
That removes the one huge problem that exists today for LC0.