this ain't no joke ...

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

this ain't no joke ...

Post by MikeB »

a McBrain update will be released later today - the only ? is which one

the final short test - same short tc that they use on fishtest - the McBrains do very well using normal type openings ( if one has ever looked the opening positions fishtest uses, they are definitely not normal type openings , but it's hard to find fault with their process - so I'm fine with it)

Code: Select all

200 rounds and 2000 games completed...
time control: 950+10
Date: 04/01/17 : 11:49:08
2000 game(s) loaded
Rank Name                      Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 McBrain 2.1f 64 POPCNT     3117   0.0   10   10   800  425.5  53.2  162  111  527  20.2  65.9  3096 
   2 McBrain 2.0 64 POPCNT      3113   4.2   10   10   800  419.5  52.4  164  125  511  20.5  63.9  3097 
   3 McBrain 2.1s 64 POPCNT     3106   6.6   10   10   800  410.0  51.2  140  120  540  17.5  67.5  3098 
   4 Stockfish 010417 64 POPC   3095  11.5   10   10   800  392.0  49.0  127  143  530  15.9  66.2  3101 
   5 Stockfish 8 64 POPCNT      3068  26.5   10   10   800  353.0  44.1  109  203  488  13.6  61.0  3108 
---------------------------------------------------------------------------------------------------------
running the final ltc test now 60 sec with 0.6 inc using the same openings

Code: Select all

80 of 2000 games completed...
time control: 6000+60
Date: 04/01/17 : 12:14:12
80 game(s) loaded
Rank Name                      Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 McBrain 2.1f 64 POPCNT     3101   0.0    2    2    32   18.5  57.8    6    1   25  18.8  78.1  3100 
   2 McBrain 2.0 64 POPCNT      3100   1.5    2    2    32   16.0  50.0    5    5   22  15.6  68.8  3100 
   3 Stockfish 8 64 POPCNT      3100   0.2    2    2    32   15.5  48.4    4    5   23  12.5  71.9  3100 
   4 Stockfish 010417 64 POPC   3100   0.0    2    2    32   15.5  48.4    3    4   25   9.4  78.1  3100 
   5 McBrain 2.1s 64 POPCNT     3099   0.9    2    2    32   14.5  45.3    4    7   21  12.5  65.6  3100 
---------------------------------------------------------------------------------------------------------

you may follow along here ,

https://www.dropbox.com/s/et9gtoldp5mjd ... a.txt?dl=1

updates will be made every 80 games or , about twenty minutes using 10 of the 12 real cores on the Mac Pro

for those that would like to see the pgn
https://www.dropbox.com/s/87ikp9q05yd3v18/all.pgn?dl=1

the pgn file is updated every minute with the latest completed games...

based on prior testing - would not be surprised to see stockfish 8 at the top of the heap when it's done...

The openings used here are the TCEC Superfinal openings from seasons 8 and 9, 100 positions, each engine plays both white and black against the other engines for 800 games , 2000 games in total.
Scacchista1977
Posts: 49
Joined: Thu Nov 10, 2016 11:40 am
Location: Italy
Full name: Aleandro Rossi

Re: this ain't no joke ...

Post by Scacchista1977 »

It looks promising ...
Gusev
Posts: 1476
Joined: Mon Jan 28, 2013 2:51 pm

Re: this ain't no joke ...

Post by Gusev »

Based on my recent brief experience in infinitychess, McBrain is quite useful. What do you think of Raubfisch? It looks like a close relative of McBrain, but is it open-source? I doubt that the second book feature that it provides is particularly valuable, anyways. What would be of value is a C version of McBrain, a la Cfish, IMHO.