Author: Robert Hyatt
Date: 13:29:50 02/27/06
Go up one level in this thread
On February 27, 2006 at 15:34:45, Tord Romstad wrote:
>On February 27, 2006 at 15:25:37, Robert Hyatt wrote:
>
>>Can you post a link to the ecmgcp positions you are using?
>
>Sure! I'm too lazy to look for them on the Web, so I just uploaded
>the EPD to my own site:
>
>http://www.glaurungchess.com/ecmgcp.epd
>
>It probably won't stay there forever, so grab it as soon as possible. :-)
>
>I find ECMGCP to be much more useful than WAC, which is simply
>too easy on modern hardware. Most of the ECMGCP positions are
>of medium difficulty, and are solved in 2-10 seconds by strong
>programs on modern hardware. There are also a few really difficult
>ones thrown in.
>
>Tord
I still use WAC a lot because they are so easy. Lets me run a test that takes
60 seconds for the entire test, as a quick evaluation of some changes. Then I
can do things like this in the c-shell:
foreach ext (0 .25 .50 .75 1.0)
foreach reduce (0 .25 .50 .75 1.0 1.25 1.50)
foreach something else ()
crafty ext=$ext reduce=$reduce something_else=$something_else <wac.run
end
end
end
And run a _bunch of different settings on the same problem set. in fiddling with
reductions, my parameters were varied on 3 settings, 4 different parameters,
giving 3^4 test runs (81). If the thing takes 30 minutes a test, that takes
forever to get any initial feedback. With WAC, I can give it say 1 minute for
the whole test and get something back in under 2 hours.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.