Fat Titz 2 versus:
1.-EMAN 7.93 (Sfnnv4 net arch. + Eman.exp + no book);
2.-StockfishMZ (Sfnnv5 net arch. + Eman.exp renamed StockfishMZ.exp + no book+ME;PE=0);
3.-StockfishMZ (Sfnnv4 net arch. + Eman.exp renamed StockfishMZ.exp + no book+ME;PE=0).
-GUI: Arena 3.5.1
-Book: SuperGM_4mvs.abk (8 plies book) only for Fat_Titz_2_win64-bmi2
-1 thread-CPU=Intel i5-7400-3GHz (Kaby Lake)
-TBS: 6-men Syzygy bases for both engines
-OS: Windows 10 Home.
+16 -3 =81
+49 Elo points for EMAN 7.93 64-bit BMI2 after 100 games
+18 -5 =77
+49 Elo points for StockfishMZ 170522 Sfnnv5 net architecture+Eman.exp file after 100 games
After the match 3, I would be curious to hear some opinions. I don't think that an extra 1,000 / 10,000 / 100,000 games would fundamentally change the data of the problem (the difference in Elo points between the engines is already stabilizing after a few dozen games).