Alfil 15.7 64-bit 4CPU Gauntlet

Discussion of computer chess matches and engine tournaments.

Moderator: Ras

User avatar
Graham Banks
Posts: 45320
Joined: Sun Feb 26, 2006 10:52 am
Location: Auckland, NZ

Re: Alfil 15.7 64-bit 4CPU Gauntlet

Post by Graham Banks »

velmarin wrote:Hi, Nathanael


It appears that we have many games finished prematurely,
9 plays, 12 plays without hardly fight.
Image
Any games less than 14 moves per side aren't included in the CCRL database.

I really appreciate Nathanael running these games to help me out with our 4CPU testing.
gbanksnz at gmail.com
User avatar
velmarin
Posts: 1600
Joined: Mon Feb 21, 2011 9:48 am

Re: Alfil 15.7 64-bit 4CPU Gauntlet

Post by velmarin »

Nathanael Russell wrote:
velmarin wrote:Hi, Nathanael


It appears that we have many games finished prematurely,
9 plays, 12 plays without hardly fight.
Image
I see now what you have captured. You have captured the games that have inadvertently "stopped due to an exception" as my Fritz Interface 12 says. For what ever reason, I do not know why this happens. I would say that 30 unfinished games is not bad considering that I was running this gauntlet for 24 hours a day 7 days a week for and managed to yield 290 completed games.
The list is total, where I asked to chessbase sort them by the minimum of moves.
There are quite a few more games, and I find strange that engines end draw up in positions of opening .
It is noted in the pgn that no movements after the opening.

Of course my thanks to their work and C.C.R.L..
User avatar
Nathanael Russell
Posts: 803
Joined: Sat Jan 31, 2015 11:50 pm
Location: Philadelphia, USA

Re: Alfil 15.7 64-bit 4CPU Gauntlet

Post by Nathanael Russell »

velmarin wrote:
Nathanael Russell wrote:
velmarin wrote:Hi, Nathanael


It appears that we have many games finished prematurely,
9 plays, 12 plays without hardly fight.
Image
I see now what you have captured. You have captured the games that have inadvertently "stopped due to an exception" as my Fritz Interface 12 says. For what ever reason, I do not know why this happens. I would say that 30 unfinished games is not bad considering that I was running this gauntlet for 24 hours a day 7 days a week for and managed to yield 290 completed games.
The list is total, where I asked to chessbase sort them by the minimum of moves.
There are quite a few more games, and I find strange that engines end draw up in positions of opening .
It is noted in the pgn that no movements after the opening.

Of course my thanks to their work and C.C.R.L..
As Graham stated, the only way to minimize but not necessarily eliminate the unfinished games "due to an exception" would be to prioritize "late draw adjudication". Initially, the reason why I chose to prioritize "early draw adjudication" was in an effort to eliminate the needless playing through won positions for the purpose of my test. The other reason is changing this one aspect would increase the amount of time required to finish my gauntlets.

Which is the lesser of two evils? Suggestions and/or kind criticism is welcome.
Core_Engine_Tester_CCRL
Image