Damir wrote:Thanks Michael. Keep up the good work.
Did you try and test it against the latest Stockfish ?
Not the release version. The testing I did previously showed it might be about the same. The biggest reason for the increase are the commits the SF development team made and me doing less damage on adding the added features. Objective is really to keep pace with Stockfish , but with added analysis features not readily available in the Stockfish releases. Of course you also have Matefinder , and Sugar , which all have added functionality - my objective was to try keep incorporate some of their fuctuonality into one engine as a user option , bypassing the need to load a different engine and of course Brainfish has the best opening book specifically designed for the Stockfish engine.
I did add a tweak to the smp mode , which narrows the search a tad for every thread added (CPU), allowing it to get deeper faster. Perhaps 1 or 2 ELO for every thread added. You can see the code for that tweak in the Search::Init function that sets up the LMR table search.c. Very time consuming to test , but I ran tests at 3, 4 , 6 and 12 core and the result was quite linear of 1-2 added ELO above what SMP typically delivers. Say if going from 1 CPU to 4 CPU adds 100 ELO , this tweak will take it to 3 to 6 ELO higher (3 x 1 to 3 X 2) or 103 to 106 - my guess is that is probably near the midpoint or 104.5.
Vinvin wrote:In Arena, the field "Current move" is not displayed
Go into your UCI Options and check mark the option "ShowInfo" that should display the "current move" field. It is off by default, only because that is my preference. One of the perks when you modify code, you can modify it to suit your own preferences. On a 12 Core Mac Pro , there is just way to much output being displayed when running on all 12 cores ( for me anyway 😉).