Null Move Pruning Test Results
Revision as of 21:27, 17 January 2019 by GerdIsenberg (talk | contribs)
Home * Engine Testing * Null Move Pruning Test Results
Home * Search * Selectivity * Pruning * Null Move Pruning * Test Results
Here you can find some experimental test results on Null Move Pruning by varying R on Depth and other features.
Increasing R With Depth
Mark Lefler
One scheme tested by Mark Lefler on 25 October 2007 is as follows:
R=3; if (side to move has 2 or more pieces) and (depth>7) then R=4;
In a 100 game 1/1 match between Now with and without this change, the change was worth just over 50 Elo. I am now testing against other programs. Followup testing showed only about a 20 Elo gain, with about 100 games against a variety of opponents. Within the error of testing.
See also
Forum Posts
- verified null move by Robert Hyatt, CCC, June 21, 2009
- my first cluster test result with R by Daniel Shawul, CCC, January 29, 2010
- value of LMR and null-move by Robert Hyatt, OpenChess Forum, July 12, 2010
- Extended Null-Move Reductions by Alvaro Cardoso, CCC, August 20, 2010 » Null Move Reductions