The ELO tests of version Sting-sf-46x (new changes) seems to be good. I can do the version 4.7 but I first will try to test (only examples) one litle change.
The new version (46x) solves quickly (old dual core) such difficult positions:
8/8/4B3/6p1/7k/8/4pPKP/8 w - - bm Bg4;
8/8/7p/3KNN1k/2p4p/8/3P2p1/8 w - - bm Kc6;
and in the longer time:
1K3kB1/8/6p1/7R/8/4b3/8/6n1 w - - bm Rh8 (1..Kg7 2.Bh7!!); (all without score = zero)
It is important for my main goal draw=zero.
BTW, Peter 64-bit version seems not to be needed.
I am only a worker and I want to give all (users of old computers too) a good toy, the real chess engine.
If I know well, my 32-bit Windows compilations of Sting don't need any libraries to start.
I recommend to use Hash = 1024M. If you want to use more, it should be 4096M minimum (if you may), but I did not test it.
Marek doesn't develop this for Elo. If he did, he'd be rebasing his changes against Stockfish 5, not still using the 2.1.1 codebase.
Sting is usually 50-100 Elo behind Stockfish 2.1.1, so it's still pretty strong, and dare I say it, one of the greatest analysis engines out there (better than Houdini)
Matthew:out
Matthew what makes you say Sting is so good at analysis?
(just asking)
Marek doesn't develop this for Elo. If he did, he'd be rebasing his changes against Stockfish 5, not still using the 2.1.1 codebase.
Sting is usually 50-100 Elo behind Stockfish 2.1.1, so it's still pretty strong, and dare I say it, one of the greatest analysis engines out there (better than Houdini)
Matthew:out
Matthew what makes you say Sting is so good at analysis?
(just asking)
I say it is one of the greatest analysis engines out there, because it is free, portable (I have performed analysis with Sting on a PowerPC G4 - can you say you've done that with Houdini?), and very good at finding good moves.
