A pretty sacrifice by M.Tal.

[Dann Corbit]

I have a feeling that the new crafty release will be a big hit.

[MikeB]

I have a feeling that the new crafty release will be a big hit.

We hope so. Also, thanks Dann for your input on the eval cache. Bob had done something very similar, although a little more complicated. As it turned out, the more simple approach was more effective - worth about 1 or 2 elo after 60,000 games of testing. It's now in the master source.

[Dann Corbit]

I have tried all sorts of methods to make the eval cache smaller (partial hash match, etc.) and never got a gain from it. The simplest works best for me too.

It probably adds 100 Elo to TSCP, because it only used the existing hash for the pv and history. My eval hash was the only time saving hash use of SJE's modified version.

A piece list will speed up TSCP a lot more, along with all the modern amenities. But it is kind of like putting lipstick on a pig.

Sungorus is such a better start for a beginner.

[PK]

Sungorus is such a better start for a beginner.

Agreed. Turns out pretty easy to bring to 2700+ Elo on CCRL scale as well:

https://github.com/nescitus/Rodent_II/b ... %201.0.zip (click "Raw" to download)

Moreover, if You compile current source from GitHub, it seems already on par with last released Rodent (1.7) and still missing a couple of important features!

heading towards 2900 regards