Used Wilhelm to generate all unique move KRkn positions with dm>=35, and 463 positions were found. Checked these positions one by one using various engines, allowing 10s on Pentium IV and finally only 12 were left. And then I tried Delfi 5.1 and it solved all of them! I did not test the Fritz engines, because manual conversion to cbh is laborious and error prone.
Anyway, here are the 12 positions in EPD form:
1n6/8/8/8/8/8/6R1/2K1k3 w - - bm Rg2-d2; dm 37;
3n4/8/8/8/7R/8/8/2K1k3 w - - bm Rh4-d4; dm 37;
8/8/3n4/8/8/8/4R3/2K2k2 w - - bm Re2-e6; dm 35;
8/8/7R/n7/8/8/8/2K2k2 w - - bm Kc1-c2; dm 35;
4n3/7R/8/8/8/8/8/2K2k2 w - - bm Rh7-e7; dm 36;
3R4/8/8/8/8/n7/8/2K2k2 w - - bm Rd8-e8; dm 35;
8/8/8/8/8/8/7k/n1K3R1 w - - bm Rg1-g4; dm 36;
8/3n4/8/8/8/6R1/8/1k1K4 w - - bm Rg3-d3; dm 35;
8/8/8/8/4n3/7R/8/1k1K4 w - - bm Rh3-e3; dm 35;
8/3n4/8/8/8/7R/8/1k1K4 w - - bm Rh3-d3; dm 35;
2n5/8/8/8/8/8/3R4/3K1k2 w - - bm Rd2-e2; dm 36;
4n3/8/8/8/7R/8/8/3K1k2 w - - bm Rh4-e4; dm 36;
Perhaps for the 4 piece positions, we may have to standardise on lesser time?
EGTB Test suite for KRkn
Re: EGTB Test suite for KRkn
I was just going through Aaron Tay's FAQ http://www.horizonchess.com/FAQ/Winboard/egtb.html after a long time and noticed the sentence:
Delfi has internal bitbase tablebases (stronger side wins/draws) for almost all the 4 men tablebases except N vs P and R vs B and KRPKR.
No wonder it is able to solve the above positions instantly.
Delfi has internal bitbase tablebases (stronger side wins/draws) for almost all the 4 men tablebases except N vs P and R vs B and KRPKR.
No wonder it is able to solve the above positions instantly.