Daniel Homan wrote:I just put the sources for EXchess v6.10 up on my website...
It has a number of minor improvements over v6.02 in king safety, some piece and mobility value optimization via CLOP, and a few minor search improvements.
I also revised the 'default' score values in the source code to match the score.par values which have been learned and hand optimized over a long period of time... this way, if score.par is deleted by accident a new version with the 'best' values will be automatically created.
EXchess v7.11 is now available from my website...
No promises on total strength increase, but my tests show it in the ballpark of +40-50 elo on one core compared to v7.03.
Lots of smallish changes, hence the small version number increase, including...
-- Improvement to work-sharing in lazy smp implementation, motivated by
ABDADA discussion by Daniel Shawul on talkchess.com, but using a data
structure rather than hash table to control this
-- Addition of a pawn hash move to pawn hash table
-- Changed from one to two layers of history based reductions
-- Use null move to detected threatened pieces when apparently far
ahead but null move is refuted by a capture
-- Various improvements to King Safety
-- Aging history table between moves rather than initializing each time
-- Using piece-ID rather than from-square to index history and reply
-- Padding hash table entries to give identical sizes on 32 and 64
-- Some parameter adjustments with CLOP
-- Various minor bug fixes