Search found 98 matches

by Raphexon
Sat Jun 22, 2019 11:18 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: CCRL scaling versus human player
Replies: 24
Views: 2866

Re: CCRL scaling versus human player

I asked a friend of mind who is rated not so high, but still not so bad : 1900 FIDE to play against Igel 1.2 and Igel 1.3 with 1 thread, 1 mb of hash and not tablebases and no book. The result is that he had absolutely no chances, but most games were quick time controls. You'd need something really...
by Raphexon
Mon Jun 10, 2019 10:15 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: SPCC: Testrun of Stockfish 190531 finished
Replies: 6
Views: 1029

Re: SPCC: Testrun of Stockfish 190531 finished

I wonder if Allie on a small net vs Leela on the same net would change their ELO relative to each other. Since Allie's search is so different, does she want a bigger or a smaller net (than Lc0?) On cccc Allie plays with the Leelenstein NNs, which have normal 20x256 size. Perhaps, I will try a bigge...
by Raphexon
Sun Jun 09, 2019 1:21 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: SPCC: Testrun of Stockfish 190531 finished
Replies: 6
Views: 1029

Re: SPCC: Testrun of Stockfish 190531 finished

And a big shame that SF doesn't seem to be progressing.
by Raphexon
Sun Jun 09, 2019 1:17 pm
Forum: Computer Chess Club: Tournaments and Matches
Topic: SPCC: Testrun of Stockfish 190531 finished
Replies: 6
Views: 1029

Re: SPCC: Testrun of Stockfish 190531 finished

Thank you for your continued testing. Also I'm wondering if different net sizes behave differently on Allie than on Leela since Allie has a different (and faster) search. Would it be possible to test Allie vs Leela on one of the smaller (distilled) nets too or a 10x128 net? https://github.com/dkappe...
by Raphexon
Mon Jun 03, 2019 5:24 am
Forum: Computer Chess Club: General Topics
Topic: Why are super top professional Go players so stupid?
Replies: 28
Views: 5506

Re: Why are super top professional Go players so stupid?

Good to see that Leela changed from 20x256 to 40x256 net, because at the end bigger is better :mrgreen:. Depends on hardware balance. They don't seem to use CUDA (cuDNN) anywhere, only OpenCL, my speeds are not impressive on my RTX 2070. Might be due to batch sizes used not helping much with cuDNN,...
by Raphexon
Sun Jun 02, 2019 6:19 pm
Forum: Computer Chess Club: General Topics
Topic: Why are super top professional Go players so stupid?
Replies: 28
Views: 5506

Re: Why are super top professional Go players so stupid?

Well there are more possible moves in Go (than in Chess) so naturally it would be logical to assume that a GO player with imperfect knowledge is more likely to blunder than (for example) a Chess player with imperfect knowledge.
by Raphexon
Tue May 28, 2019 2:14 pm
Forum: Computer Chess Club: General Topics
Topic: TCEC S15, END of an ERA event is much more Brutal than I thought!
Replies: 127
Views: 24351

Re: TCEC S15, END of an ERA event is much more Brutal than I thought!

A new possible challenge for LC0. in his latest book Garry Kasparov still states that “Centaur mode” is still the best expression of strength in chess games. I am doubtful after witnessing how LC0 defeated Stockfish. In order to test Kasparov's ipothesis I would suggest the following challenge: GM ...
by Raphexon
Fri May 17, 2019 8:15 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish scaling is now impressive?
Replies: 4
Views: 1309

Re: Stockfish scaling is now impressive?

Jouni wrote:
Fri May 17, 2019 4:28 pm
In CCC 8 final 90 core SF is beating Lc0 now.
Slim margins though.
by Raphexon
Thu May 16, 2019 9:11 am
Forum: Computer Chess Club: General Topics
Topic: TCEC S15, END of an ERA event is much more Brutal than I thought!
Replies: 127
Views: 24351

Re: TCEC S15, END of an ERA event is much more Brutal than I thought!

I do wonder, how strong is the current net on a CPU. On CCRL 40/40 Lc0 0.20.1 w36089 is the 35th strongest single core engine. (SF2.2.1 level) On 40/4 it\s significantly weaker. Wonder how much ELO the latest net (and executable) has in CCRL40/40 conditions. I guess since Lc0 scales pretty well with...
by Raphexon
Sat May 11, 2019 2:23 pm
Forum: Computer Chess Club: General Topics
Topic: A complete 2000 lines of code engine
Replies: 164
Views: 45147

Re: A complete 2000 lines of code engine

Still awesome that even with so few lines, it still would have been the strongest engine back in the late 90s/early 2000s.