Need Program For Comparing Hacked vs. Official Crafty

[jack512]

What is a good program for comparing two versions of Crafty (hacked vs. official)? I am running on Ubuntu Linux.

Also, what is considered a good comparison run? Obviously there is a tradeoff between short time control, which allows more games, vs. a longer time control representing tournament conditions. All I have is my 2 core 2.4GHz PC with 4GB memory.

[bob]

What is a good program for comparing two versions of Crafty (hacked vs. official)? I am running on Ubuntu Linux.

Also, what is considered a good comparison run? Obviously there is a tradeoff between short time control, which allows more games, vs. a longer time control representing tournament conditions. All I have is my 2 core 2.4GHz PC with 4GB memory.

Do you mean compare by playing games? xboard is the easiest option using match mode.

Time control depends on what you "hacked". I like Larry K's ratio of inc = 1/60th of base time. So 1min + 1 sec, or 30sec + .5sec, etc...

[jack512]

Yes, compare by playing games is what I meant. Thx, I will get xboard.

My hack is to use a different opening book, and play this against the opening book that comes with Crafty. Out of OB, each side uses the official Crafty engine.

I'm not sure what you mean by inc. If I set time control with the Crafty command "time 90/30/sd/60;" like club play, I'm afraid it might take forever to do enough games to be statistically significant. On the other hand setting it too quick might not be realistic.
Thx, Jack Fishburn

[jack512]

oops I meant "time 30/90/sd/60" (30 moves in 90 minutes then 60 minutes for rest of game)

[Robert Pope]

My hack is to use a different opening book, and play this against the opening book that comes with Crafty. Out of OB, each side uses the official Crafty engine.

I'm not sure what you mean by inc. If I set time control with the Crafty command "time 90/30/sd/60;" like club play, I'm afraid it might take forever to do enough games to be statistically significant. On the other hand setting it too quick might not be realistic.
Thx, Jack Fishburn

Since you aren't changing the engine itself, you should be able to get by with a very fast time control. Like game in 5 seconds, not anything near club play or blitz.

[bob]

Yes, compare by playing games is what I meant. Thx, I will get xboard.

My hack is to use a different opening book, and play this against the opening book that comes with Crafty. Out of OB, each side uses the official Crafty engine.

I'm not sure what you mean by inc. If I set time control with the Crafty command "time 90/30/sd/60;" like club play, I'm afraid it might take forever to do enough games to be statistically significant. On the other hand setting it too quick might not be realistic.
Thx, Jack Fishburn

You can give xboard most any time control. You might try something like:

-mps 60 -tc 0:10 -inc 0.3

that gives 60 moves every 10 seconds with a 0.3 second increment...

[hgm]

oops I meant "time 30/90/sd/60" (30 moves in 90 minutes then 60 minutes for rest of game)

Note that this is not a possible time control for WB engines. The only supported TC are N moves / M minutes repeating ('classical'), or M minutes base time + I seconds per move ('incremental' or 'sudden death').

XBoard's internal TC handling is already capable of doing multi-session time controls, but there just isn't a way to tell that to the engines.

[Evert]

My hack is to use a different opening book, and play this against the opening book that comes with Crafty. Out of OB, each side uses the official Crafty engine.

I'd just like to point out that using a different opening book does not constitute "hacking" of a chess engine. You're just comparing the relative quality of the opening books.

[jack512]

In the xboard I downloaded to Ubuntu Linux, the manpage says -inc disables -mps:

"-inc or -timeIncrement seconds. If this option is specified, `movesPerSession' is ignored. Instead, after each player's move, `timeIncrement' seconds are added to his clock."

I will try -tc 0:10 -inc 0.3