McCain

[Hurnavich]

Hi,

Any expected updates for McCain i really like this engine

[Scally]

Hi,

Knowing Michael, he’s probably already testing his latest beta/dev of his next update, I’m sure he’ll answer soon.

I love this engine too, with levels it suites Picochess on the Raspberry Pi.

Al.

[MikeB]

Hi,

Any expected updates for McCain i really like this engine

Stockfish 10 is getting primed for release, next McCain update will be available very shortly thereafter. I can tell you that the Stockfish team has made great progress sine SF9 - it should be around +50 Elo stronger than SF9, McCain generally keeps pace, but it's more geared for longer time controls and overnight correspondence chess analysis and almost always plays weaker ( it gets "lucky" once in a while) at the short time controls currently in vogue. I'm not sure if people have noticed, but current dev-SF is much better at analysis than say SF 8. Also, I get tremendous help from Dann Corbit, Lucas Monge, Al Scally, John Stanback with the compiles for the various OS. Thanks to Al Scally, those with DGT/Raspberry Pi setup will get get an engine that crushes any dedicated chess computer available (of course a DGT/Raspberry Pi setup is not a dedicated chess computer, but it certainly looks like one ) .

All the McCain changes are guarded with "#ifdef Maverick" (or #ifndef Maverick) statements so one will be able to easily see the changes from SF 10 when released if so desired.

Just for fun, I will kickoff a short match later tonight vs the current Stockfish-10 release candidate at a much longer time control than what the SF team typically runs. Check back in this thread for the links to follow.

[MikeB]

Just kicked off a 126 game "fun" match with a time control of a 5 min base with 30 second increment between current-dev SF 10 rc ( unofficial release candidate) and cur-dev McCain-X rc ( "X" pronounced as "ten"). No adjudication - every game played to the bitter end ( although fast play is enabled once EGTBs are in play, 6 man syzygy used for both sides). 12 games are running concurrently on a 12 real core Mac Pro with 256M hash and no pondering. Openings will be magic63.epd , handpicked for their unbalanced nature from Kai's openings files.

Results will be reported here as each game is completed:
https://www.dropbox.com/s/xxenip25fz6it ... a.txt?dl=1

PGN file will be updated as each game is completed:
https://www.dropbox.com/s/wzius95cvw8e7hw/all.pgn?dl=1

If you use dropbox on your iPhone, the app will tell you when to refresh the file.

Total match will last about 16 to 18 hours

First updates will start in about an hour , once the first game is completed and then once every game is completed, usually about every 10 minutes or so, but it varies.

Note: Once 115 games are completed, the next update is when all 126 games are completed. It is the design of the script to report games every time a new game is started, once 115 games are completed, all 126 games have been started, hence no update till the end.

[Hurnavich]

Excellent news looking forward to the Mcain release i use for book tuning

[MikeB]

Just kicked off a 126 game "fun" match with a time control of a 5 min base with 30 second increment between current-dev SF 10 rc ( unofficial release candidate) and cur-dev McCain-X rc ( "X" pronounced as "ten"). No adjudication - every game played to the bitter end ( although fast play is enabled once EGTBs are in play, 6 man syzygy used for both sides). 12 games are running concurrently on a 12 real core Mac Pro with 256M hash and no pondering. Openings will be magic63.epd , handpicked for their unbalanced nature from Kai's openings files.

Results will be reported here as each game is completed:
https://www.dropbox.com/s/xxenip25fz6it ... a.txt?dl=1

PGN file will be updated as each game is completed:
https://www.dropbox.com/s/wzius95cvw8e7hw/all.pgn?dl=1

If you use dropbox on your iPhone, the app will tell you when to refresh the file.

Total match will last about 16 to 18 hours

First updates will start in about an hour , once the first game is completed and then once every game is completed, usually about every 10 minutes or so, but it varies.

Note: Once 115 games are completed, the next update is when all 126 games are completed. It is the design of the script to report games every time a new game is started, once 115 games are completed, all 126 games have been started, hence no update till the end.

It was apparent this AM something had gone wrong with my most recent updates. After 64 games, these were the results:

Code: Select all

64 of 126 games completed...
Time control: 600 + 30.00 seconds
Threads: 1
Hash: 256
Date: 11/26/18 : 09:38:48
Rank Name                      Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 Stockfish 10rc 64 POPCNT   3123   0.0   37   37    64   36.5  57.0   13    4   47  20.3  73.4  3077 
   2 McCain Xrc                 3077  45.8   37   37    64   27.5  43.0    4   13   47   6.2  73.4  3123 
---------------------------------------------------------------------------------------------------------

So I scrubbed the last several patches I committed , found the error and just ran quick test just to ensure it's back to normal.
This one is obviously one of those "lucky" runs!

Code: Select all

126 rounds and 126 games completed...
Time control: 60 seconds + 3.00 seconds
Target completion: 11/26/18 : 14:40:03
Date: 11/26/18 : 13:50:38
126 game(s) loaded
Rank Name                      Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 McCain Xrc                 3104   0.0   31   31   126   64.5  51.2   19   16   91  15.1  72.2  3096 
   2 Stockfish 10rc 64 POPCNT   3096   8.0   31   31   126   61.5  48.8   16   19   91  12.7  72.2  3104 
---------------------------------------------------------------------------------------------------------
Time to complete: 01:52:36

Will post later tonight when the longer 10 min game with 30 second increment is restarted.

[MikeB]

Code corrections were made and the "fun" match was restarted:
cur-dev-SF-10-rc (release candidate) vs McCain-X rc (X is pronounced ten)
126 game match
Game 10 min plus 30 second increment
Openings: 63magic.epd ( hand selected from Kai's openings)
2010 Mac Pro - 12 core
Target to complete: ~ 17 hours
6 Man Syzygy - SSD
No Adjudications - every game played until the bitter end.
First results expected around 2AM EST

Results:
https://www.dropbox.com/s/jw3sdzvju9fu2 ... a.txt?dl=1

PGN File:
https://www.dropbox.com/s/laey5httvsw1m2g/all.pgn?dl=1

[MikeB]

I killed it here - a little bit worse than expected.

Code: Select all

107 of 126 games completed...
Time control: 600 + 30.00 seconds
Threads = 1
Hash = 256
Date: 11/27/18 : 18:31:07
107 game(s) loaded
Rank Name                      Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 Stockfish 10-RC-f 64 POP   3113   0.0   27   27   107   57.5  53.7   15    7   85  14.0  79.4  3087 
   2 McCain X-RC-f              3087  25.2   27   27   107   49.5  46.3    7   15   85   6.5  79.4  3113 
---------------------------------------------------------------------------------------------------------
107 game(s) loaded

[MikeB]

For those that are interested , kicked yet another 126 game match with a newly tweaked cur-dev-McCain , 10 min base, 30 second increment , first results in an hour:

The STC Match was very close:

Code: Select all

126 rounds and 126 games completed...
Time control: 60 seconds + 3.00 seconds
Target completion: 11/28/18 : 01:08:50
Date: 11/28/18 : 00:30:04
126 game(s) loaded
Rank Name                      Rating   Δ     +    -     #     Σ    Σ%     W    L    D   W%    =%   OppR 
---------------------------------------------------------------------------------------------------------
   1 Stockfish 10-RC-f 64 POP   3101   0.0   32   32   126   63.5  50.4   19   18   89  15.1  70.6  3099 
   2 McCain X-RC-f              3099   2.8   32   32   126   62.5  49.6   18   19   89  14.3  70.6  3101 
---------------------------------------------------------------------------------------------------------
Time to complete: 02:03:15

Results:
https://www.dropbox.com/s/jw3sdzvju9fu2 ... a.txt?dl=1

PGN File:
https://www.dropbox.com/s/laey5httvsw1m2g/all.pgn?dl=1

[Hurnavich]

Excellent results