- add staged move generation
- fix problems in move validation
- fix problem in draw evaluation
- fix problem in mate distance pruning
- fix possible crash in long games
- fix problem in null moves pruning
- change node counting method (+1 before making a legal move in search)
- change history heuristic array (piece/to)
- disable recapture extension
- use check evasion in the first ply of quiescence
- add evaluation tempo bonus
- adjust mate scores in hash table
- few optimizations
- windows : replace QueryPerformanceCounter() function by GetTickCount()
- use UCI parameters seldepth, lowerbound, upperbound
- perft now return only nodes count (old perft renamed "perftd")
velmarin wrote:
I really like your page, and the engine, and I believe you have a lot to improve.
Thanks,
Yes I have lots of things to improve and bugs to fix,
the main problem is to find time and motivation.
Excuse me, again the translator.
Now read what I did not want to write.
I wanted to say,
you seem to have a good base to work.
A very good base for improvement. Bitboard, speed, ect ..
No critical sovereign actual, Excuseme.
I tested with a mine beta version, with good results, stability, ect.
I liked.
velmarin wrote:
I wanted to say,
you seem to have a good base to work.
A very good base for improvement. Bitboard, speed, ect ..
No critical sovereign actual, Excuseme.
It is what I understood reading your previous message, no problem.