Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: monsoon's results for Peter's endgame suite

Author: Robert Hyatt

Date: 16:50:12 12/06/01

Go up one level in this thread


On December 06, 2001 at 19:23:20, Peter McKenzie wrote:

>On December 06, 2001 at 17:07:06, Robert Hyatt wrote:
>
>>On December 06, 2001 at 12:46:54, Ron Murawski wrote:
>>
>>>On December 06, 2001 at 11:54:50, Robert Hyatt wrote:
>>>
>>>>
>>>>
>>>>I was sick last week.  Flu went thru my family.  As a result, I didn't do
>>>>much of anything except teach classes here.
>>>>
>>>>Can you give me a link to the test suite and I'll give it a run.
>>>>
>>>>What time limit do you want?
>>>>
>>>>Bob
>>>
>>>Bob,
>>>
>>>See: http://homepages.caverock.net.nz/~peter/eg_test/eg_test.htm
>>>
>>>You can find all the info there.
>>>
>>>Ron
>>
>>
>>Here are my results...
>>
>>  1      0
>>  2     23
>>  3      0
>>  4      0
>>  5      2
>>  6      1
>>  7     --
>>  8      4
>>  9      0
>> 10     --
>> 11     --
>> 12     --
>> 13     --
>> 14      3
>> 15      1
>> 16     --
>> 17     --
>> 18     17
>> 19     --
>> 20      0
>> 21      0
>> 22      0
>> 23     11
>> 24      0
>> 25     --
>> 26      0
>> 27      1
>> 28      2
>> 29      2
>> 30     14
>> 31      1
>> 32     --
>> 33     --
>> 34     --
>> 35      0
>> 36      0
>> 37     --
>> 38      5
>> 39      1
>> 40     20
>> 41     --
>> 42      0
>> 43     14
>> 44     --
>> 45      8
>> 46      1
>> 47      0
>> 48     --
>> 49      1
>> 50     --
>>
>>-- means not solved in 60 seconds.  All times given are in
>>seconds, with fractional part dropped.
>
>Was that on the Quad?
>Quick count indicates 33/50, best result so far.
>
>cheers,
>Peter


Yes... sorry... quad 700, using 48 megs for hash, 6 megs for phash.  Everything
else as default.

note that if you ever want to run something like that, using Crafty, just type

(assuming the FEN stuff is in a file named "peter.fen" for example)

st=60   (sets time limit to 60 secs per move)
hash=48M   (sets a reasonable hash size)
phash=6M   (ditto)
computer   (sets drawscore=0, reduces some asymmetry)
test peter.fen

and watch it scroll...

If you want to shorten the test, type

test peter.fen 4

that says that if it has the right move (or avoids the bad move) for four
consecutive iterations, stop the search immediately rather than using the full
60 minutes.  I use this for testing WAC, although when I want real results,
I leave the early exit turned off...

Doing this:

test wac.test 2

will run wac at 60 seconds per move and the entire test takes 5 minutes max
as most are found very quickly and the solution "keeps" for 2 consecutive
iterations.  That is much more pleasant for testing/tuning than running for
a full 5 hours at 60 secs/position.




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.