Crafty-22.9 is available

[Charles B.]

How to prevent Crafty 22.9 from ever resigning?

Setting resign= to a very high number in the .rc file looks to have no effect in preventing the engine from resigning.

[Dann Corbit]

This is all very interesting but this latest version of Crafty on my Quad is still being hammered by the ancient Fritz 7 (running on only 1 cpu) on my tactics test...

Are you sure that the answers in your tactics test are correct?

I have found that most test sets are full of flaws.

[Werewolf]

ok I was a little hasty. After completing the test suite Crafty 22.9 (64bit, 4 cores) got 27/50 same as Fritz 7 but behind Fritz 9 (32/50), Zappa Mexico II (41/50) and Rybka 3 (45/50).

[Werewolf]

This is all very interesting but this latest version of Crafty on my Quad is still being hammered by the ancient Fritz 7 (running on only 1 cpu) on my tactics test...

Are you sure that the answers in your tactics test are correct?

I have found that most test sets are full of flaws.

Yes they are correct. Rybka 3 gets 45/50 and would get more if I gave it enough time.

[Werewolf]

This is all very interesting but this latest version of Crafty on my Quad is still being hammered by the ancient Fritz 7 (running on only 1 cpu) on my tactics test...

Is your tactics test secret?

No, I'm happy to post it here for people to view. Not sure how to attach a file on this site...

[Dann Corbit]

Unless it has thousands of lines, just paste it into the text box with either "Code" or "List" tags from above.

[bob]

ok I was a little hasty. After completing the test suite Crafty 22.9 (64bit, 4 cores) got 27/50 same as Fritz 7 but behind Fritz 9 (32/50), Zappa Mexico II (41/50) and Rybka 3 (45/50).

As I said, that is not that surprising. What I am working on is playing actual games, which is a different task from solving tactical test positions... I'm not surprised that the 22.x versions don't do so well since the extension code has been removed, etc...

[bob]

How to prevent Crafty 22.9 from ever resigning?

Setting resign= to a very high number in the .rc file looks to have no effect in preventing the engine from resigning.

works for me. to never resign, just use "resign=0" and it will never resign. That doesn't mean a GUI won't resign for it, but Crafty won't itself...

[Denis P. Mendoza]

Bob,

Thanks again for the new update. Just saw at Leo's forum release of Crafty 22.10src. For early testing, some may try this MSVC (32/64 bit) pgo builds.

I could only test it in 2-cores, but I noticed (or I wasn't aware maybe) that it switches from 1 thread to 2 threads while waiting for its turn and when it's time to move.


Get it here: http://www.computerchess.info/tdbb/phpB ... ?f=9&t=273

Denis

[bob]

Bob,

Thanks again for the new update. Just saw at Leo's forum release of Crafty 22.10src. For early testing, some may try this MSVC (32/64 bit) pgo builds.

I could only test it in 2-cores, but I noticed (or I wasn't aware maybe) that it switches from 1 thread to 2 threads while waiting for its turn and when it's time to move.

Do not see how that could happen, unless you have "smpnice=0" in the crafty.rc/.craftyrc file. Otherwise when it is not thinking it should use nothing, when it is thinking or pondering it should use whatever you told it to use with the mt=N command...

Get it here: http://www.computerchess.info/tdbb/phpB ... ?f=9&t=273

Denis