Search found 845 matches

by dkappe
Sun Apr 18, 2021 10:45 pm
Forum: Computer Chess Club: General Topics
Topic: Introducing Toga III chess engine
Replies: 25
Views: 6495

Re: Introducing Toga III chess engine

Toga III 0.3 is now available. The only update is a new net, trained using reinforcement learning. It’s approximately 20 elo stronger in self-play. The Linux and Windows binaries can be found here: https://www.patreon.com/posts/50193195 (github seems to have lowered its file size limits).
by dkappe
Fri Apr 16, 2021 4:17 pm
Forum: Computer Chess Club: General Topics
Topic: Komodo is dominating TCEC now
Replies: 2
Views: 463

Re: Komodo is dominating TCEC now

Jouni wrote:
Fri Apr 16, 2021 4:12 pm
Dragon has 2 wins in row: FRC and Swiss :) . Now they have updated to 512GB RAM like CCC. Serious hardware.
Very small sample size and small margins of victory.
by dkappe
Thu Apr 08, 2021 5:14 am
Forum: Computer Chess Club: General Topics
Topic: LcFish??
Replies: 6
Views: 801

Re: LcFish??

The April Fools joke that keeps on giving.
by dkappe
Thu Mar 25, 2021 6:17 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish One Thread Challenge
Replies: 9
Views: 1406

Re: Stockfish One Thread Challenge

In CCC one thread SF (NNUE) beats 250 threads Ethereal! And plays equal to 250 threads Komodo or SF classic. Definitely doubling of cores don't give 70 ELO anymore :) . Looking at the first event, Dragon had 37 point out of 60 vs 1 thread SF. Now I don’t know if that’s a better or worse than expect...
by dkappe
Fri Mar 19, 2021 8:25 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: CCRL 40/15, 40/2 and FRC lists updated (6th March 2021)
Replies: 17
Views: 1708

Re: CCRL 40/15, 40/2 and FRC lists updated (6th March 2021)

... because testing GPU engines is a total pain in the arse. No concurrency, even if you had a machine with dozens of cores you can run only one match. Hi, Why is that? Do the GPU engines use all the available cores? I am thinking about getting a new PC with a GPU, that's what I am interested. I've...
by dkappe
Mon Mar 15, 2021 4:01 pm
Forum: Computer Chess Club: General Topics
Topic: Zahak, a GoLang based chess engine
Replies: 38
Views: 3185

Re: Zahak, a GoLang based chess engine

I am using a garbage collected language, I cannot really control when memory is freed when searching. So, the increase during search is probably normal. But why does it stay high even after search, I might have some memory leak, need to check, which is not exactly an easy task :( Ideally you would ...
by dkappe
Sun Mar 14, 2021 11:27 pm
Forum: Computer Chess Club: General Topics
Topic: Zahak, a GoLang based chess engine
Replies: 38
Views: 3185

Re: Zahak, a GoLang based chess engine

amanjpro wrote:
Sun Mar 14, 2021 10:27 pm
dkappe wrote:
Sun Mar 14, 2021 8:00 pm
Nice work. I played 4 games against it with leela2200 (lc0 running a 16x2 T10 distilled net on cpu) on lichess. All tied up at 2 wins a piece.
Oh, thank you. I think leela2200 won the mini match 3-2 though ;)
It appears that white won every game.
by dkappe
Sun Mar 14, 2021 8:00 pm
Forum: Computer Chess Club: General Topics
Topic: Zahak, a GoLang based chess engine
Replies: 38
Views: 3185

Re: Zahak, a GoLang based chess engine

Nice work. I played 4 games against it with leela2200 (lc0 running a 16x2 T10 distilled net on cpu) on lichess. All tied up at 2 wins a piece.
by dkappe
Sat Mar 13, 2021 8:32 pm
Forum: Computer Chess Club: General Topics
Topic: NNUE Research Project
Replies: 61
Views: 5678

Re: NNUE Research Project

Note that nets trained on non-stockfish data often have to have their scores multiplied by a factor in order to fall into a good range for alpha-beta. Values between 1.5 to 2.5 are common. I use a python script to adjust the final layer for this. The toga net used a 1.55 multiplier, if I recall. If...
by dkappe
Sat Mar 13, 2021 7:36 pm
Forum: Computer Chess Club: General Topics
Topic: NNUE Research Project
Replies: 61
Views: 5678

Re: NNUE Research Project

I am looking for more engines that moved to NNUE. Anyone? The data so far is too less for conclusions. Toga III https://github.com/dkappe/TogaIII/releases Solly, ignore previous post, real numbers - SF12 -> RMS=72.63 | SIM=46.22 SF13 -> RMS=72.20 | SIM=45.65 Note that nets trained on non-stockfish ...