Search found 360 matches

by noobpwnftw
Sat Feb 24, 2018 3:55 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some opening book design questions
Replies: 39
Views: 11034

Re: Some opening book design questions

Actually, Chess has every necessary parts ready(i.e. results from fishtest) to make a book by brute force. It might practically ultra-weak solve the opening.
by noobpwnftw
Fri Feb 23, 2018 2:23 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some opening book design questions
Replies: 39
Views: 11034

Re: Some opening book design questions

It would then require 12(RCN) * 7 + 6(P) * 10 + 4(B) * 4 + 3(A) * 4 + 4(K) * 2 + 1(turn) = 181 bits or even less if palace states are combined. Although keys(either hash or other formats) are known before probing the book, unlike EGTBs that you won't store keys at all(accessed with data[index] = val...
by noobpwnftw
Fri Feb 23, 2018 6:56 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some opening book design questions
Replies: 39
Views: 11034

Re: Some opening book design questions

1. Probing radix trees does not require comparisons of uncompressed form, so effectively I'm comparing for 7.34 bytes per position on average. 2. Fixed lengths or not do not really matter since you won't be storing your book as "data[index]" anyway. Sorted skip lists would seek just as fast. 3. http...
by noobpwnftw
Thu Feb 22, 2018 4:24 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some opening book design questions
Replies: 39
Views: 11034

Re: Some opening book design questions

Crazy or not, here are some stats on binary encoded FENs from 8.85 billion stored positions in bytes: min-keylen = 17, max-keylen = 45, avg-keylen = 35.19, avg-zkey = 7.34 After compression(as shown by avg-zkey) it is actually less than 8 bytes, yet seeking among keys are close to O(1). You can appl...
by noobpwnftw
Wed Feb 21, 2018 7:42 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some opening book design questions
Replies: 39
Views: 11034

Re: Some opening book design questions

Maybe that is because no one has ever built an opening database large enough for BW symmetry to matter. At least EGTBs implement them, so maybe we should do the same for openings, and doing so is not expensive. BTW, losing a tempo in the openings is never considered bad in Xiangqi, in fact, it is ve...
by noobpwnftw
Wed Feb 21, 2018 6:53 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some opening book design questions
Replies: 39
Views: 11034

Re: Some opening book design questions

If you are meant for Xiangqi: 1. You can have a index scheme that stores only one of the mirrored positions (LR/BW/LRBW). Scores should remain the same with current moving player's perspective. Sample probing code here: https://github.com/noobpwnftw/chessdb/blob/master/db_bench_util/update.php 2. Us...
by noobpwnftw
Thu Nov 23, 2017 9:11 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 128-core AMD server
Replies: 40
Views: 8595

Re: 128-core AMD server

I have a couple of Supermicro boxes that are 4/8 way Xeons. 1. Yes, you can run Windows. 2. Large memory allocation is transparent and far-side NUMA latency is not significant. 3. Most chess engines doesn't spread threads properly across Processor Groups when you have more than 64 cores, including S...
by noobpwnftw
Sun May 28, 2017 10:21 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is AlphaGo approach unsuitable to chess?
Replies: 19
Views: 9986

Re: Is AlphaGo approach unsuitable to chess?

If that is anywhere comparable to top-tier chess engines, they probably have done that already, given the amount of quality games available for training and all that. Eventually what they want is people migrating to TPUs, not beating pro Go players, I assume.
by noobpwnftw
Mon May 22, 2017 11:31 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Probe EGT in quiescence?
Replies: 28
Views: 5273

Re: Probe EGT in quiescence?

Certainly it does contain those basic combinations you've mentioned, they are essential to any bigger tablebases. But there isn't any download page, are you referring to the list pages? I knew you have full basic EGTs since I have visited your page of EGTs, found some and check their sizes already ...
by noobpwnftw
Mon May 22, 2017 8:20 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: How far away are we from deep learning Stockfish, Komodo,...
Replies: 48
Views: 12989

Re: How far away are we from deep learning Stockfish, Komodo

Would it be for example possible to take Stockfishs code and change only the evaluation function into an deep learning evaluation function? And also use cpu + gpu for different tasks? Maybe a code mix of Giraffe and Deep Pink and some improvements would be a better deep learning engine? Why are you...