TalkChess.com

Thank you for testing! Finally Marvin is above 3000.

/Martin

A new version of Marvin is now available. I estimate that it's about 50 elo stronger than the previous version. Release Notes SEE improvements More aggressive SEE pruning Improved time management Added counter move and follow-up move history heuristic Store position evaluation in transposition table...

Thank you for testing!

Marvin 3.5.0 is now available. It's about 30 elo stronger than 3.4.0, and maybe 20 elo stronger than the version that played in TCEC S17. Most of the work for this release have been focused on the evaluation so that's where most of the improvements comes from. For the next release I plan to focus mo...

I have just released Marvin 3.4.0. My tests show a 100 elo improvement compared to version 3.3.2, but that's probably a bit optimistic. Most of the improvements comes from more aggressive LMP and improvements to history heuristic. I have also done some refactoring of the move generation and move sel...

After some additional testing I found two more issues with the book code and I think there are still more that I haven't found yet. I clearly haven't tested this code enough. I'm working on a few other things at the moment so it will take some time, but I will at least make sure to fix this before t...

I checked the code and if it gets to a position that's not in the book then it will stop checking the book for the rest of the game. So it will miss cases where the game transposes back to a position that is in the book. Could this explain the problems you're seeing?

Ingen har testat senaste releasen än, men jag gissar på ~2960 med 4 kärnor.

I have just done a second bug fix release, Marvin 3.3.2. This release fixes an issue where Marvin would sometimes fail to win in a
mate-in-x position.

Thanks to the user NajdorfB on GitHub for reporting the issue.

A new fixed version, Marvin 3.3.1, is now available.

https://github.com/bmdanielsson/marvin- ... tag/v3.3.1