Search found 361 matches

by Raphexon
Thu Apr 15, 2021 12:11 pm
Forum: Computer Chess Club: General Topics
Topic: Speed of Lc0 in the new Intel GPU's?
Replies: 7
Views: 182

Re: Speed of Lc0 in the new Intel GPU's?

Does anyone have any news about the speed of Lc0 in the new Intel GPU's? Pathetic is the answer you are looking for. Unless your GPU is GTX 1030 or something from an early last decade. Than CPU performance might be in the same order of magnitude (but still slower). Is that because they dont have te...
by Raphexon
Thu Apr 15, 2021 10:26 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Fairy-Max and chess variants rules generalization question to HGM
Replies: 11
Views: 321

Re: Fairy-Max and chess variants rules generalization question to HGM

On a tangent, it's a shame the Japanese Chu-Shogi players never invented a mnemonic pieceset like HGM's.
Maybe it would have never died out.
by Raphexon
Thu Mar 25, 2021 6:34 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Why doesn't this seemingly over-powered technique work?
Replies: 5
Views: 753

Re: Why doesn't this seemingly over-powered technique work?

it seems like NNUE uses the normal Stockfish calculation to look several moves deep and then uses a neural net type evaluation instead of a static evaluation (do correct me if I am wrong here). This gives it a clear edge over normal Stockfish, since it effectively gets a few more moves of 'free' de...
by Raphexon
Tue Mar 23, 2021 9:49 am
Forum: Computer Chess Club: General Topics
Topic: Marvin 5.00 - Ethereal 12.75, 1:0 mate after 33!! ... great game!
Replies: 17
Views: 1174

Re: Marvin 5.00 - Ethereal 12.75, 1:0 mate after 33!! ... great game!

Looks like Stockfish 13 would have also played b7-b5 until depth 19. (2 million nodes)

But even weaker engine seem to find it after a few seconds.
So not a very hard position, but one that will induce blunders at low tc.
by Raphexon
Fri Mar 12, 2021 7:59 pm
Forum: Computer Chess Club: General Topics
Topic: stockfish 10 vs. Mephisto III S Glasgow
Replies: 126
Views: 22739

Re: stockfish 10 vs. Mephisto III S Glasgow

Comparing nodes 1:1 between two engines is also false. As it stands, SFdev is the strongest entity for a Pentium II (Or any CPU with MMX or better), and that's a CPU from 1997. SF-HC would be the strongest on the older 32 bit CPUs. (going back all the way to 1985) On an i386 SF11 should still be abl...
by Raphexon
Wed Mar 10, 2021 4:13 pm
Forum: Computer Chess Club: General Topics
Topic: A random walk down NNUE street ….
Replies: 46
Views: 4679

Re: A random walk down NNUE street ….

Mike, a few thoughts 1) There’s not a lot of benefit to d11 at this stage. I’d just go with d6 or d8 until your net has figured out piece values. 2) generate 1b positions per generation. More is better. 3) start with lambda 0.0 (all result) and gradually add more eval. Not sure how fast to turn the...
by Raphexon
Tue Mar 09, 2021 3:32 pm
Forum: Computer Chess Club: General Topics
Topic: Cutochess cli
Replies: 39
Views: 3131

Re: Cutochess cli

I have a question. What is the difference between -tb PATHS and -tbpieces N. https://github.com/cutechess/cutechess/blob/master/projects/cli/res/doc/help.txt#L101 -tb PATHS Adjudicate games using Syzygy tablebases. PATHS should be semicolon-delimited list of paths to the compressed tablebase files....
by Raphexon
Tue Mar 02, 2021 4:11 pm
Forum: Computer Chess Club: General Topics
Topic: Stockfish node speed mystery
Replies: 7
Views: 929

Re: Stockfish node speed mystery

Or maybe hybrid eval gets called more with more cores.
by Raphexon
Sun Feb 28, 2021 3:31 pm
Forum: Computer Chess Club: General Topics
Topic: Rybka vs Fat Fritz controversies
Replies: 33
Views: 3503

Re: Rybka vs Fat Fritz controversies

I think for a while, everyone thought that chess engine progress will improve ... but it will improve mostly due to the improvement in hardware running bean counter engines. Rybka changed that thinking and it made clear that the software side of chess engines is where the major gain in ELO would co...
by Raphexon
Wed Feb 17, 2021 5:54 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Reinforcement learning project
Replies: 48
Views: 8451

Re: Reinforcement learning project

Aren't the repetition rules for Janggi really complicated (or even subjective in nature)?
How are those coded?