Stock drew in KBBK !?

[Onno Garms]

I am not inclined to analyze this, but I had similar problems in early versions of Onno. Onno could not even win KRK. It always showed mate scores but never actually mated. After this was reported, I found that some more engines have trouble mating in elementary endgames without tablebases.

In Onno this was due to the TT replacement scheme. I never figured out what exactly was going on, but I fixed that by storing in the TT twice. Once with normal replacement scheme and once I hashed the pair (position,depth). When probing, I check both entries. When the depth specific entry is better (i.e. larger min_value, smaller max_value, or a move where the general entry has move_none), I return the depth specific entry.

Unfortunately this made Onno play worse in general positions, so I activate this scheme only when KRK is on the board...

I'm not really proud on that solution (which I have not fully understood), but it's the best I know to do about that problem.

[Onno Garms]

StockFish 2.0 could easily win KBBK or KBNK,

If this is the case, I'd recommend to examine the changes that have been done in the TT scheme. (There are at least some changes related to using the TT in PV.)

[zamar]

If this is the case, I'd recommend to examine the changes that have been done in the TT scheme. (There are at least some changes related to using the TT in PV.)

Yes, this is most likely the reason. We made a change which made stockfish to remember previous thinking sessions, so it became more useful for analysis, but the change created this bug. We will fix this issue in next relase.

[Don]

... (both Wine and Arena are extremely lame and buggy).

If this is so, why do you use them?

Both Fruit and Stockfish are available as native Linux versons, I think, and you could run them under a natively running XBoard..

XBoard is also very fast in comparison. Arena is very nice but it doesn't feel crisp and responsive. It's sort of has the refresh speed of a web browser and not a desktop application.