Are you talking about finding the Mate in 11, or just finding Qh5? Either way, this is not a hard position to solve at all. Even if I turn off EVERYTHING (hash, history, aspiration, SEE, pruning, null move, IIR, etc.), my mediocre Myrddin finds Qh5 in 1.6 seconds at depth 5, although it's a LONG way from finding mate.Chessnut1071 wrote: ↑Sat Sep 30, 2023 1:26 am FEN[226] = "6r1/p1pq1p1p/1p1p1Qnk/3PrR2/2n1P1PP/P1P5/4R3/6K1 w - - 0 1"; // f5h5 e5h5 g4g5 h5g5 h4g5 h6h5 e2h2 h5g4 h2g2 g4h5 f6f3 h5h4 f3g3 h4h5 g2h2 d7h3 g3h3 g6h4 h3h4 h5g6 h4h6
Using the FEN above, I tested the solution with and without the TT. Results as follows:
1) With TT: time to solution: 161.085 seconds; nodes searched = 310,019,642
2) Without TT: time to solution = 1244.05 seconds; nodes searched = 1,556,880,385
The TT table reduced time to solution by 80% and the number of nodes searched by 87%.
My TT table includes 1) hash key 2) success/failure 3) ply. If a collusion occurs, per your suggestion, I just throw ignore the hash key. Also, I have enough space so the keys can't overlap. I use checkmate solutions instead of perft, and have found at least 80% reduction in time and about 92 % reduction in nodes searched. However, the TT gets much more efficient at higher ply levels. The one thing I found is you need larger hash table for deep ply solutions. Without a TT the time to solution on some problems could be months or even years. I can't even imagine running without a TT unless the solution is only a few ply deep.
If I turn off only hash, it finds Mate in 12 in 19 seconds and Mate in 11 in 67 seconds.
If I enable hash, it finds Mate in 11 in 0.2 seconds.
So there must be several methods of improving the time to find mates that you are not making use of.