Search found 64 matches

by ymatioun
Fri Dec 23, 2016 10:09 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: On-line engine blitz tourney November
Replies: 88
Views: 14140

Re: On-line engine blitz tourney November

Well, i don't think we can have the tournament without you, so i guess tomorrow is off the table.
by ymatioun
Fri Dec 23, 2016 8:02 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: On-line engine blitz tourney November
Replies: 88
Views: 14140

Re: On-line engine blitz tourney November

Have the server problems been fixed? Will there be December tourney? There is not much time left in a month - the only Saturday other than tomorrow is the 31st, when i imagine many people will not be available. So tomorrow may be our last chance.

Youri.
by ymatioun
Mon Sep 12, 2016 11:15 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Best move statistics
Replies: 22
Views: 4475

Re: Best move statistics

I have done something similar, but limited to Q search only. Then you only have good or equal captures. I constructed a table indexed by attacker and by victim, and use that for move ordering instead of the usual MVV/LVA. That approach helped, but only slightly.
by ymatioun
Wed Sep 07, 2016 1:00 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Baffling multithreading scaling behavior
Replies: 41
Views: 7992

Re: Idea: Try a parallel perft()

why don't you profile your slow run - using "very sleepy" profiler, for example. When i had a similar problem, profiler showed that a lot of time was spent writing in memory shared by all threads (to update total node count and other search stats). Once i removed that code, NPS scaling became almost...
by ymatioun
Fri Aug 26, 2016 10:03 pm
Forum: Computer Chess Club: General Topics
Topic: Fizbo 1.8 released
Replies: 6
Views: 1794

Fizbo 1.8 released

Version 1.8 of Fizbo chess engine has just been released. It can be downloaded from https://sites.google.com/site/fizbochessengine/ Here are release notes: 1. added BMI2 executable. 2. reworked and improved pondering logic, fixing pondering time-out bug. Now the engine should never time out, with or...
by ymatioun
Thu Aug 25, 2016 10:22 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Value of IID
Replies: 6
Views: 1643

Re: Value of IID

I tried implementing IID in Fizbo on several occasions, and never got any improvement.
by ymatioun
Wed Aug 17, 2016 11:54 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: perpetual check position
Replies: 27
Views: 3869

Re: perpetual check position

I just searched this position with Fizbo to depth 39/seldepth 67, and it still does not see a draw. It looks to me that the depth for a forced repetition draw here s really very high (black king can go on most squares in files f-g-h twice before it happens). I don't have any special logic to recogni...
by ymatioun
Wed Jun 29, 2016 10:45 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engine testing hardware.
Replies: 11
Views: 2817

Re: Engine testing hardware.

i run 8000 games overnight on my only computer. This way it does not interfere with anything. I just back into time limits that will keep all cores busy for 8-9 hours.
by ymatioun
Sun Jun 26, 2016 12:19 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Transposition Table - Updating entries
Replies: 3
Views: 976

Re: Transposition Table - Updating entries

I had the same TT replacement logic, and tried changing it to "always replace"; now searching fine #70 position to depth 40 takes only 78 msec, when it used to take 17 sec - speed-up of over 200! I'll test this change further before accepting it (and i don't expect a large ELO gain), but it is truly...
by ymatioun
Sun May 29, 2016 12:47 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: An old dilemma
Replies: 17
Views: 2886

Re: An old dilemma

I have see this, too. So some time ago i switched from self-play to gauntlets as a primary way to evaluate engine strength. This approach should correlate well with CCRL rankings.