KRPKBP
-
gigabyte137
- Posts: 12
- Joined: Mon Apr 24, 2006 5:04 pm
- Sign-up code: 0
KRPKBP
Has anyone noticed that when in KRPKBP endgame sometimes it doesnt work properly. Meaning normally it shows depth of 1 all the time when in a endgame, with mine sometimes it jumps out of that and into a quick depth search to 40. I have verified with mdsum and wilhelm. Do you think it does that when it sees a promotion to queen and it doesnt have that endgame?
-
guyhaw
- Posts: 489
- Joined: Sat Jan 21, 2006 10:43 am
- Sign-up code: 10159
- Location: Reading, UK
- Contact:
KRPKBP
I know that the KRPKBP EGT has been well used by an experienced endgame player and EGT user, I think with SHREDDER if not other engines, and I've heard of no problems.
The (indirect) interface between the engine you are using and the EGTs can affect behaviour in two ways:
a) if you don't have follow-on EGTs that KRPKBP can convert to (e.g. KQRKBP, KBPKR) then your engine may or may not be intelligent enough to go into evaluate-mode instead, and
b) if your software thinks the EGT is 8-bit based when it is 16-bit based or vice-versa, weird depths can happen. This happened a few years ago, during production of 6-man DTZ and DTZ50 EGTs. A 'wrong cell size' EGT got used, with the result that maxDTx was '1' - and the EGT 'verified' using Nalimov's verifier. Of course, the results were obviously wrong and the reason was traced. This sounds like your 'normally 1' observation.
It would be useful if you said what engine(s) you are using, and gave some positions where you are getting a DTM of '1' reported and where those positions do not enable a conversion into a later endgame.
g
The (indirect) interface between the engine you are using and the EGTs can affect behaviour in two ways:
a) if you don't have follow-on EGTs that KRPKBP can convert to (e.g. KQRKBP, KBPKR) then your engine may or may not be intelligent enough to go into evaluate-mode instead, and
b) if your software thinks the EGT is 8-bit based when it is 16-bit based or vice-versa, weird depths can happen. This happened a few years ago, during production of 6-man DTZ and DTZ50 EGTs. A 'wrong cell size' EGT got used, with the result that maxDTx was '1' - and the EGT 'verified' using Nalimov's verifier. Of course, the results were obviously wrong and the reason was traced. This sounds like your 'normally 1' observation.
It would be useful if you said what engine(s) you are using, and gave some positions where you are getting a DTM of '1' reported and where those positions do not enable a conversion into a later endgame.
g
-
gigabyte137
- Posts: 12
- Joined: Mon Apr 24, 2006 5:04 pm
- Sign-up code: 0
KRPKBP
It is Rybka in Fritz GUI but now I assume it is because I dont have KQRKBP and that is why because when I get into that predicament it goes in and out of 1 ply 65 ply so it seems it has to search with engine when it see a promotion to queen...