I process using 6 engines simultaneously.
The engines are Crystal 200804,, MateFinder 10, The Huntsman 1, Stockfish Tactical, SlowChess 2.9, and Houdini 6.0. Each is running 1 thread, PV1, 3Gb of hash, withSyzygy 6-man TBs. The exception being Houdini which is using Nalimov 6-man and PV2 (it can usually find the second best move the fastest).
The test position is a mate in 17 according to every engine but SlowChess 2.9 which thinks it is a mate in 14.
| 8 | |||||||||
| 7 | |||||||||
| 6 | |||||||||
| 5 | |||||||||
| 4 | |||||||||
| 3 | |||||||||
| 2 | |||||||||
| 1 | |||||||||
| a | b | c | d | e | f | g | h |
8/4p2P/8/2p5/p7/K7/2r5/k7 w - - 0 1
This is where SlowChess 2.9 goes off the rails. It believes 4... e5 is correct, while 4...e6 delays the mate by 3 moves.
Once I play out the moves and force 4....e6 to be played, SlowChess finds the correct line, but if I back up a move to 4. Qe4, it still thinks e5 is the best reply.
I am not sure who is supporting SlowChess (open source?), but perhaps whomever "they" are, they can move the engine forward by finding why it plays this move.
Unless there is a newer version that I don't have. If so, let me know and I'll try it.
