According to my database, 46.Qb1 wins for Black as well as Qc1, Qd2, Qe2, Kb3 and Kb1, why shouldn't all of these, if one does so?Spliffjiffer wrote: ↑Sat Jan 22, 2022 5:19 pm i cant await wonders from "Osters"-SF...and i do not expect wonders from You...my amount of RAM is maybe too small to get a "root-investigation" of this position...BUT i will try it now again !
if you find interest and time i would be eagerly interested in what u and your system can find out...would be nice to see which moves might be as good as 46.Qb1, Peer...but 46.Qb1 !? is loosing for black
White's not in any hurry and not in Zugzwang, so first move doesn't matter so much as for the necessary maneuvers with King, Queen and Knight. There's more than one order and more then one direction the Knight can take attacking two of the three fields a5, c5, e5.
Tried new SF dev. by going into one of my stored winning lines, evals get >5 sooner then with Crystal, but yet at a short ponder- time started with empty hash of about 3 minutes, at position after 46.Qb6, SF dev. still sticks to 46...Nxb6? with an eval of >4, so the title of the thread is still true at least for the second part of it.
And at Backward it's still difficult to keep the evals at root position >4, so Nxb6 is still a candidate as good as some others for the engine then again. Returning to root positions from winning lines means almost no special hash- entries to 46.Nxb6, so this one move is still to be pondered about almost as long as started with empty hash, that's the old problem of the nodes of not enough hash for all of the candidates to compare, even after Forward- Backward, always the latest lines gone backward are the most hash- filling ones.
I'd still rather take Crystal for this one position, evals don't get as high as quickly as with SF dev. by going into the winning lines, but they (the evals) are more reliable as for winning at all, cause the taking of the Queen at once is evaluated as drawn. If the fortress is searched out here, it probably will be later on too, if there still is one.
With some more hardware- time free, maybe I'll give it one more try later on to find out, how long which way to win takes regards...
Edit: see your latest posting now, probably meaning, you already gave up with SF dev. too again, nevermind, guess it can't work without fortress detection code of the kind of Crystal's.
...i understand!?....do not take too much time for an ivestigation...its much easier when we investigate ten years later probably...one head one ass lol...yes, best i found for this pos was: Crystal4 !...currently trying with Shash but its not toooo promising....thx Pete