Search found 724 matches

by jwes
Thu Mar 22, 2018 1:47 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: trick question
Replies: 25
Views: 5807

Re: trick question

The trick is to use second counter. for IO, I'd use a separate thread (main thread IO, search in a separate thread for the timeout, however, I'd still use polling I use one thread for input which blocks and one thread which waits for an input line, a search result, or a timeout. The search routine ...
by jwes
Thu Mar 22, 2018 6:42 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: trick question
Replies: 25
Views: 5807

Re: trick question

The trick is to use second counter. for IO, I'd use a separate thread (main thread IO, search in a separate thread for the timeout, however, I'd still use polling I use one thread for input which blocks and one thread which waits for an input line, a search result, or a timeout. The search routine ...
by jwes
Fri Mar 09, 2018 9:47 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Looking inside NNs
Replies: 23
Views: 4608

Looking inside NNs

I ran across this which claims to explain the predictions of a NN. I have no idea if it works, but it would be interesting to run it on AlphaZero or giraffe.
code is here
by jwes
Sun Feb 25, 2018 6:35 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: [Discussion] - Measuring move ordering
Replies: 14
Views: 4134

Re: [Discussion] - Measuring move ordering

I am trying to find a formula to measure move-ordering as best as possible and when agreed upon hold an educational (and fun) contest between engine authors. I can think of 2 systems. #1. the percentage when the first move of a move-list in the tree remains the best move at the end of the move-list...
by jwes
Sun Nov 26, 2017 8:42 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: 128-core AMD server
Replies: 40
Views: 8164

Re: 128-core AMD server

jhellis3 wrote:
What about the argument that its time to depth that matters?
TTD is a useless measurement for LazySMP.
So you're saying that LazySMP finds better moves at equal depths rather than finding the same move faster.
by jwes
Sun Nov 19, 2017 9:32 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engines for testing (Linux, fast time control)
Replies: 17
Views: 4393

Re: Engines for testing (Linux, fast time control)

One idea is to use top engines at time odds. This has the benefit of making matches go quicker. For my engine (which is much weaker than Arasan), it also exposes evaluation problems in particular as at time odds of several hundred to one, my engine can outsearch them.
by jwes
Sat Nov 18, 2017 11:19 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Statistical interpretation of search and eval scores.
Replies: 3
Views: 1111

Statistical interpretation of search and eval scores.

My math and stats are pretty rusty, but I seem to recall a good estimate would have an unbiased mean and normally distributed errors. I don't see how scores in centipawns can have these properties, as the true value is win, draw or loss. Can someone help me understand what form of scores would have ...
by jwes
Wed Nov 15, 2017 12:53 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there now coming changes to syzygy databases?
Replies: 18
Views: 4467

Re: Is there now coming changes to syzygy databases?

Is WDL+DTZ+DTM enough to ensure perfect play under the 50 move rule or do we need DTM50? Another idea you may already be doing is splitting tables where one side has two bishops into two parts, one with one bishop of each color and one with both bishops on the same color. This would likely improve c...
by jwes
Tue Nov 14, 2017 10:53 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Is there now coming changes to syzygy databases?
Replies: 18
Views: 4467

Re: Is there now coming changes to syzygy databases?

I have adapted my generator for DTM, but DTM tables cannot replace WDL/DTZ. But they can be used in addition to WDL/DTZ to get mate distances (huge waste of disk space, though). I am not yet finished. Would it be possible to replace some DTZ tables with DTM if those tables have no cursed wins or bl...
by jwes
Thu Nov 09, 2017 3:42 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Move ordering
Replies: 17
Views: 3523

Re: Move ordering

Manik is convinced that there is a bug in the move ordering code of Dorpsgek due to having a comparatively low fail-high on first move rate (I ran a self-play game to depth 8 and got an average of 84.8%), and after a couple of months of not being motivated, I thought I'd ask the CCC. My move orderi...