hardware : dual xeon e5-2660v3
software : arena 3.5.1
engines :
- Eman 8.20 played with last green net and experience data from 6 learning sessions
- Stockfish 15.0 played with nn-6877cd24400e.nnue net
settings : TC 1m+1s vs different TC, 40 threads, hash 4 GB, partial syzygy 7men
opening list : 50 openings with the most experience data
Code: Select all
# PLAYER : RATING ERROR POINTS PLAYED (%) W D L D(%) OppAvg OppN
1 1m+1s Eman 8.20 nn-ad9b42354671 : 0 ---- 50.5 100 50.5 2 97 1 97.0 -4 1
2 2m+2s Stockfish 15.0 x64 BMI2 : -4 50 49.5 100 49.5 1 97 2 97.0 0 1
White advantage = -3.51 +/- 24.96
Draw rate (equal opponents) = 50.00 % +/- 0.00
Before continuing this experiment, I have to find another list of openings because this one produces too many draws. We see that the average times per move go from simple (1.8s) to double (3.7s). The difference between the average depths (D30 vs D36) is partly erased by the nnue technology.1m+1s Eman 8.20 nn-ad9b42354671 with 2/100 wins (2,00%, D30/131, 1 762 ms) :
1. e4 @ 50,00%
1. g4 @ 50,00%
2m+2s Stockfish 15.0 x64 BMI2 with 1/100 wins (1,00%, D36/245, 3 679 ms) :
1. g4 @ 100,00%
Averages :
233 sec/game
87 plies/game