Analyzing a one million games database...

[rodolfoleoni]

**ouch!!** It's painful...

I'm running the resulting tree analysis since almost one week. At fixed depth, Baron is taking 5 to 30 secs per position and I could be at about 20% of database.

Gosh!

[marcelk]

**ouch!!** It's painful...

I'm running the resulting tree analysis since almost one week. At fixed depth, Baron is taking 5 to 30 secs per position and I could be at about 20% of database.

Gosh!
:shock:

I have a machine doing just this 24/7. The good thing is that while waiting you have more than enough time to come up with a lot of good filtering ideas :-)

[rodolfoleoni]

**ouch!!** It's painful...

I'm running the resulting tree analysis since almost one week. At fixed depth, Baron is taking 5 to 30 secs per position and I could be at about 20% of database.

Gosh!

I have a machine doing just this 24/7. The good thing is that while waiting you have more than enough time to come up with a lot of good filtering ideas

Being not a programmer, I can use nothing more than Richard Pijl filtering ideas...

In fact, this work derives from the promising experiment I recently ran with Critter session file being used instead of opening books. The Baron facilities allow me to use some shortcuts (e.g. to analyse the tree instead of single games, to handle transpositions, and so on). I don't know about a GUI which has the same features, so the work is being done in console mode. When finished, I should have a learn file to use with experiments.

[Albert Silver]

I'm surprised you started this without knowing beforehand how long it would take. If you got 20 percent done, then you have a very reduced tree IMHO.

[rodolfoleoni]

I'm surprised you started this without knowing beforehand how long it would take. If you got 20 percent done, then you have a very reduced tree IMHO.

I made a similar work some months ago, but with a shallow fixed depth. It lasted 2 days, and my guess was about 30 days this time. The good thing is that I can stop/resume it when I want.

And yes, there are some tricky reductions, you're a good observer.

[jdart]

For what purpose are you analyzing it (for a opening book, or ?).

[rodolfoleoni]

For what purpose are you analyzing it (for a opening book, or ?).

The purpose is to get a learnfile from positions analysis, because I want to test it vs. standard opening books. Recent tests I ran with Critter learn system showed me it could gain superiority vs. traditional opening books, so I want to try it with the Baron.

Meantime, this work serves to score Baron book too, and it could be useful later.