Hello Garbor,SzG wrote:Just watched this hooror ending. It was played under WinBoard and, unfortunately, I haven't saved the debug file.
...
it seems like a problem in the time management function (again). I managed to reproduce the blunder Qe6+?? on
[D] 8/3k4/1Q6/p1p5/P3q3/1P5K/4P3/8 b - - 6 67
It appears the engine ran out of time before finishing the 40 moves cycle , causing it to make the move instantly. Unfortunately as you can see the horrible Qe6+ is selected as the best move if the engine has no time to search in this position:
Code: Select all
RedQueen. Copyright (C) 2008-2012 Ben-Hur Carlos Vieira Langoni Junior.
uci
id name RedQueen 1.1.1
id author Ben-Hur Carlos Vieira Langoni Junior
option name Hash type spin default 128 min 1 max 4096
option name Threads type spin default 8 min 1 max 16
option name Ponder type check default true
option name Positional_Evaluation_Weight type spin default 100 min 1 max 200
option name Tactical_Evaluation_Weight type spin default 100 min 1 max 200
option name Use_Custom_Engine_Settings? type check default true
option name Custom_Settings_File_Path type string default parameters.txt
option name Clear Hash type button
uciok
setoption name Threads value 1
position fen 8/3k4/1Q6/p1p5/P3q3/1P5K/4P3/8 b - - 6 67
go depth 0
bestmove e4e6
go depth 2
info depth 1 seldepth 1
info depth 1 score cp -50 time 1 nodes 239 nps 239 pv e4e6 b6e6
info nodes 239 time 1 nps 239 hashfull 0
info depth 2 seldepth 4
info depth 2 score cp -49 time 2 nodes 504 nps 504 pv e4e3 h3h4 e3e4
info nodes 504 time 2 nps 504 hashfull 0
info depth 2 seldepth 4
info depth 2 score cp -49 time 2 nodes 504 nps 504 pv e4e3 h3h4 e3e4
info nodes 504 time 2 nps 504 hashfull 0
bestmove e4e3 ponder h3h4
Time management related functions will be my priority to the next release.
Thanks for testing it.