Author: Howard Exner
Date: 21:23:12 11/04/99
Go up one level in this thread
On November 04, 1999 at 23:11:16, Dann Corbit wrote:
>On November 04, 1999 at 22:51:11, Howard Exner wrote:
>
>>You mentioned that you run the position for 12 minutes.
>>How long do you run each of the analysis trail variations?
>
>Explanations:
>
>The original analysis for these positions (my first post) was 12 minutes PII 300
>MHz CPU time {standard C.A.P. effort} per position.
>
>The follow-on of the original positions was at one hour on a PIII 500 MHz. You
>can see the time for each row in the 'acs' epd field. The ones at 1000 seconds
>were on a PII 300 MHz machine. The rest of the results were from a PIII 500
>MHz machine.
>
>The analysis trail positions were at about 15 minutes of PII 300 MHz equivalene
>effort per position, and I have analyzed only the moves provided by H.E. and not
>the pv of crafty.
Wow, that's alot of processing. I went through some of the analysis and
it looks like you found some alternative lines that are as good as the ones
proposed by the ECM98 project.( the nxd6 vs Bxe5 alternative both win a piece).
When you complete this analysis of these ten positions posted by Jouni
let me know which ones you think have better or equally good moves as the
original bm. Providing the main variations would be enough - ie: The move X
seems better than the move Y because ... (I'm getting a bit lost by the
many sub lines).
With ECM98 the goal was to remove flawed or murky(bm having no big advantage
over alternative moves) positions as well as toss out the trivial ones(those
moves that programs were finding in a few seconds) from ECM. Some suggested we
take it
one step further and have many programs do another 5-10 second pass on ECM98
and then to toss out the ones that all programs find in that short time. The
result was to have a clean set of positions that would give todays programs
a challenge.
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.