Eelco de Groot wrote:I do not understand all of Crafty's output, I interpret the numbers between brackets as seconds but in the last lines there is a number of 1.04 seconds, if that is for 18 plies that would be very fast even for Crafty
It is seconds, i dont know why it says 1 second at the fail high(maybe its something with the copying from the scid interface ?)
Also note that in your output, it found the correct move at depth 17, not depth 18...
Yes it was a typo
Ive used Crafty almost exclusively analysing my own games and is quite happy with the improvements it have been showing(just wanted to tell)
But is it an eval issue that it sticks to Nd5 for so long ?
Other engines(glaurung 2.2 and Rybka) seems quite much faster in finding Rdf1.
I must say that I too have been greatly enjoying Crafty 22.8. It's quite strong (not top tier by ANY means but still very strong and useful for analysis of games) and when compiled with -DSKILL option, you can set skill=10 (or whatever level works for you) and get a decent game against it without getting brutally defeated by a full strength engine.
royb wrote:I must say that I too have been greatly enjoying Crafty 22.8. It's quite strong (not top tier by ANY means but still very strong and useful for analysis of games) and when compiled with -DSKILL option, you can set skill=10 (or whatever level works for you) and get a decent game against it without getting brutally defeated by a full strength engine.
Eelco de Groot wrote:I do not understand all of Crafty's output, I interpret the numbers between brackets as seconds but in the last lines there is a number of 1.04 seconds, if that is for 18 plies that would be very fast even for Crafty
It is seconds, i dont know why it says 1 second at the fail high(maybe its something with the copying from the scid interface ?)
Also note that in your output, it found the correct move at depth 17, not depth 18...
Yes it was a typo
Ive used Crafty almost exclusively analysing my own games and is quite happy with the improvements it have been showing(just wanted to tell)
But is it an eval issue that it sticks to Nd5 for so long ?
Other engines(glaurung 2.2 and Rybka) seems quite much faster in finding Rdf1.
The question is, what are the eval numbers for Nd5 vs Rf1? If they are close, then it is an eval issue. I have seen several positions posted here where Crafty finds the correct move far faster than (say) Rybka. But the "find" is not tactical in nature, it is positional. Every program evaluates things from king safety to pawn structure and passed pawns differently, and those little differences can make a big difference in positions where the best move is only separated from another move by a very small positional score...
