ChessUSA.com TalkChess.com
Hosted by Your Move Chess & Games
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   RegisterRegister 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

A new chess engine : m8 (comming not so soon)
Post new topic    TalkChess.com Forum Index -> Computer Chess Club: Programming and Technical Discussions Flat
View previous topic :: View next topic  
Author Message
Peter Österlund



Joined: 19 Apr 2010
Posts: 537
Location: Sweden

PostPost subject: Re: A new chess engine : m8 (comming not so soon)    Posted: Sun Feb 01, 2015 7:38 pm Reply to topic Reply with quote

mar wrote:
mathmoi wrote:
Isn't lazy SMP the technique that use the TT as a communication device between threads that search the same tree?

Basically yes, but it has been improved since the old days (also we're using larger TTs today):
resync on each ID iteration, early termination when one of the helpers finishes and run each other thread on depth+1 (as proposed by Dan Homan).

Quote:

My understanding was that it gave little speedup passed 2 CPU. Am I missing something?

Where did you read that? :shock: There's evidence that it works and in fact it works very well.
If claim sun in blue will you believe it? If I keep repeating it over and over again? Or will you trust what you see?

I have tested this algorithm using Cheng 0.38 on a 16 core Dell PowerEdge T620 computer. Both hyperthreading and turbo boost are enabled. The computer runs Fedora 19. Here are the results, computed by bayeselo:
Code:
Cheng 4c vs Cheng 1c, 1+0.08:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_4c     68    8    8  1457   70%   -68   35%
   2 cheng4_038       -68    8    8  1457   30%    68   35%

Cheng 4c vs Cheng 1c, 8+0.64:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_4c     66   10   10  1011   71%   -66   42%
   2 cheng4_038       -66   10   10  1011   29%    66   42%

Cheng 8c vs Cheng 4c, 1+0.08:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_8c     28    8    8  1475   59%   -28   44%
   2 cheng4_038_4c    -28    8    8  1475   41%    28   44%

Cheng 16c vs Cheng 8c, 1+0.08:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_16c     4    8    8  1400   51%    -4   46%
   2 cheng4_038_8c     -4    8    8  1400   49%     4   46%

Cheng 16c vs Cheng 8c, 2+0.16:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_16c    13    8    8  1433   54%   -13   50%
   2 cheng4_038_8c    -13    8    8  1433   46%    13   50%

Cheng 16c vs Cheng 8c, 4+0.32:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_16c     6    8    8  1151   52%    -6   53%
   2 cheng4_038_8c     -6    8    8  1151   48%     6   53%

Cheng 16c vs Cheng 8c, 180+1:
Rank Name             Elo    +    - games score oppo. draws
   1 cheng4_038_16c    17   20   20   184   56%   -17   53%
   2 cheng4_038_8c    -17   20   20   184   44%    17   53%

Even at hyper bullet speed (1s+0.08s/move) it scales well up to 8 cores. +130 elo from 1 to 4 cores, and +56 elo from 4 to 8 cores. At 16 cores it seems hyper bullet speed is too fast for the algorithm to be effective, but it still seems to work well, +34 elo, at 3m+1s/move time control, even though the number of games (184) is too low to be really sure.

Feeding all 16 vs 8 cores games into bayeselo gives +16 elo and LOS=99.9%, so it is pretty clear that the algorithm works also for 16 cores even if it is unclear exactly how well it works.

Some other results for comparison, even though the low number of games and different 1 core ratings make it hard to draw any definite conclusions:
Code:
Texel 16c vs Texel 8c, 180+1:
Rank Name          Elo    +    - games score oppo. draws
   1 Texel16c       21   19   18   190   58%   -21   69%
   2 Texel8c       -21   18   19   190   42%    21   69%

Texel 16c vs Texel 16c_nonuma, 180+1:
Rank Name          Elo    +    - games score oppo. draws
   1 Texel16c        7   19   19   175   53%    -7   71%
   2 Texel16cnn     -7   19   19   175   47%     7   71%

Komodo 8 16c vs Komodo 8 8c, 180+1:
Rank Name          Elo    +    - games score oppo. draws
   1 komodo8_16c    12   18   18   196   55%   -12   78%
   2 komodo8_8c    -12   18   18   196   45%    12   78%

_________________
Peter Osterlund - peterosterlund2@gmail.com
http://hem.bredband.net/petero2b
Back to top
View user's profile Send private message Send e-mail Visit poster's website
Display posts from previous:   
Subject Author Date/Time
A new chess engine : m8 (comming not so soon) Mathieu Pagé Sun Feb 01, 2015 3:22 am
      Re: A new chess engine : m8 (comming not so soon) Graham Banks Sun Feb 01, 2015 3:53 am
      Re: A new chess engine : m8 (comming not so soon) Ferdinand Mosca Sun Feb 01, 2015 5:36 am
            Re: A new chess engine : m8 (comming not so soon) Evert Glebbeek Sun Feb 01, 2015 9:02 am
                  Re: A new chess engine : m8 (comming not so soon) Mathieu Pagé Sun Feb 01, 2015 3:20 pm
                        Re: A new chess engine : m8 (comming not so soon) Evert Glebbeek Sun Feb 01, 2015 11:23 pm
            Re: A new chess engine : m8 (comming not so soon) Mathieu Pagé Sun Feb 01, 2015 3:06 pm
                  Re: A new chess engine : m8 (comming not so soon) Ferdinand Mosca Mon Feb 02, 2015 12:58 pm
                        Re: A new chess engine : m8 (comming not so soon) Reinhard Scharnagl Mon Feb 02, 2015 1:37 pm
      Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Sun Feb 01, 2015 8:48 am
            Re: A new chess engine : m8 (comming not so soon) Mathieu Pagé Sun Feb 01, 2015 3:16 pm
                  Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Sun Feb 01, 2015 5:00 pm
                        Re: A new chess engine : m8 (comming not so soon) Peter Österlund Sun Feb 01, 2015 7:38 pm
                              Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Sun Feb 01, 2015 8:14 pm
                                    Re: A new chess engine : m8 (comming not so soon) Lucas Braesch Mon Feb 02, 2015 10:16 am
                                          Re: A new chess engine : m8 (comming not so soon) Lucas Braesch Mon Feb 02, 2015 10:39 am
                                          Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Mon Feb 02, 2015 11:57 am
                                                Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Mon Feb 02, 2015 12:09 pm
                                                      Re: A new chess engine : m8 (comming not so soon) Matthew Lai Mon Feb 02, 2015 1:10 pm
                                                            Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Mon Feb 02, 2015 3:44 pm
                                                                  Re: A new chess engine : m8 (comming not so soon) Matthew Lai Mon Feb 02, 2015 4:42 pm
                                                                        Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Mon Feb 02, 2015 6:28 pm
                                                                              Re: A new chess engine : m8 (comming not so soon) Matthew Lai Mon Feb 02, 2015 6:40 pm
                                                                                    Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Mon Feb 02, 2015 6:56 pm
                                                                                          Re: A new chess engine : m8 (comming not so soon) Matthew Lai Mon Feb 02, 2015 7:05 pm
                        Re: A new chess engine : m8 (comming not so soon) Mathieu Pagé Sun Feb 01, 2015 9:39 pm
                              Re: A new chess engine : m8 (comming not so soon) Martin Sedlak Mon Feb 02, 2015 8:41 am
      Re: A new chess engine : m8 (comming not so soon) Ray Banks Sun Feb 01, 2015 9:34 am
      Re: A new chess engine : m8 (comming not so soon) H.G.Muller Sun Feb 01, 2015 4:36 pm
      Re: A new chess engine : m8 (comming not so soon) Volker Annuss Sun Feb 01, 2015 10:36 pm
            Re: A new chess engine : m8 (comming not so soon) Álvaro Begué Sun Feb 01, 2015 11:03 pm
Post new topic    TalkChess.com Forum Index -> Computer Chess Club: Programming and Technical Discussions

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum




Powered by phpBB © 2001, 2005 phpBB Group
Enhanced with Moby Threads