Search found 1142 matches

by Gian-Carlo Pascutto
Sat Jan 18, 2020 8:06 am
Forum: Computer Chess Club: General Topics
Topic: The Stockfish of shogi
Replies: 18
Views: 2092

Re: The Stockfish of shogi

I don't speak Japanese but their architecture seems to be: W1 = 125388 x 256 W2 = 512 x 32 W3 = 32 x 32 W4 = 32 x 1 And I can make some educated guesses: they exploit the fact that W1 doesn't change much to compute the result of that layer incrementally, which is coincidentally the heaviest layer. T...
by Gian-Carlo Pascutto
Tue Dec 10, 2019 5:28 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: PEXT/PDEP are even slower than you think on Zen
Replies: 12
Views: 1451

Re: PEXT/PDEP are even slower than you think on Zen

Another consideration is noise, the 180W thermals from the 10980XE can be cooled down very quietly with a Noctua NH-D15, I have two systems with these running and I can hardly hear that they are turned on. For a TR3-3970X you will start at 280W (probably even more), a 64 core TR3 will be like 500W,...
by Gian-Carlo Pascutto
Mon Dec 09, 2019 5:06 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: PEXT/PDEP are even slower than you think on Zen
Replies: 12
Views: 1451

PEXT/PDEP are even slower than you think on Zen

https://twitter.com/trav_downs/status/1 ... 7962364928

It was already known they were slow, but I didn't see anyone point out before that how slow they get actually depends on the argument.
by Gian-Carlo Pascutto
Wed Nov 27, 2019 7:19 am
Forum: Computer Chess Club: General Topics
Topic: w49921 net
Replies: 29
Views: 2204

Re: w49921 net

RubiChess wrote:
Wed Nov 27, 2019 5:46 am
Remember that they still use last official SF10 for testing
They do not.
by Gian-Carlo Pascutto
Tue Nov 26, 2019 8:44 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: CCRL 40/40, 40/4 and FRC lists updated (23rd November 2019)
Replies: 22
Views: 2483

Re: CCRL 40/40, 40/4 and FRC lists updated (23rd November 2019)

Jouni wrote:
Mon Nov 25, 2019 5:15 pm
Chessbase don't waste time https://en.chessbase.com/post/optimizing-fat-fritz .
I am sure Albert Silver (who wrote the article) is very proud he wrote the strongest engine in the world.
by Gian-Carlo Pascutto
Mon Nov 25, 2019 1:40 pm
Forum: Computer Chess Club: General Topics
Topic: Anyone heared of MuZero ?
Replies: 3
Views: 485

Re: Anyone heared of MuZero ?

>Assuming the "steps" they refer to are games

It's training steps of 2048 minibatches. I don't see any mention of games - they run the processes asynchronously. But it seems they generated about 20 billion positions. So, probably quite a bit more than a million games.
by Gian-Carlo Pascutto
Mon Nov 25, 2019 11:59 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: CCRL 40/40, 40/4 and FRC lists updated (23rd November 2019)
Replies: 22
Views: 2483

Re: CCRL 40/40, 40/4 and FRC lists updated (23rd November 2019)

https://www.sp-cc.de/lc0-testing.htm

This shows Fat Fritz being 40-50 Elo weaker? Rather incompatible results, unless 49921 is a very weak net or something.
by Gian-Carlo Pascutto
Wed Nov 20, 2019 3:27 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: UCI Win/Draw/Loss reporting
Replies: 65
Views: 8089

Re: UCI Win/Draw/Loss reporting

You can use the logit of the probability of winning, and multiply it by a constant so that it looks like centipawns. I do this for my program, and it works very well. I use this, which is the same thing right? https://www.chessprogramming.org/Pawn_Advantage,_Win_Percentage,_and_Elo It's convenient ...
by Gian-Carlo Pascutto
Tue Oct 29, 2019 8:08 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: UCI Win/Draw/Loss reporting
Replies: 65
Views: 8089

Re: UCI Win/Draw/Loss reporting

I missed this discussion. So what is agreed up on: info wdl 200 600 200 format for uci, and what for xboard? Thanks. I don't think anything has been agreed upon, but Fritz 17 is now confirmed to ship with a modified lc0 that has the patch pointed out in the original post: https://github.com/DanielU...
by Gian-Carlo Pascutto
Tue Oct 29, 2019 8:03 am
Forum: Computer Chess Club: General Topics
Topic: Some words on Leela Go Zero
Replies: 4
Views: 1363

Re: Some words on Leela Go Zero

GPU: RTX 2070 May network versus elfv2: 10:10 October network versus elfv2: 12:8 So, the "700 Elo points" in self ratings is hugely inflated, and the progress is probably no more than 100 Elo points in 5 months, quite modest for Go. It's hard to say anything from such a short match, no? I think the...