PlentyChess - A UCI chess engine with a neural network based evaluation.
Building on Windows
For the build to work on Windows, you need to install MinGW as well as the required packages (make, g++, pthread, etc.).
Then you can use the makefile as you would on a unix system.
v.1.0.0:
This is the biggest release of PlentyChess so far, shipping (D)FRC and MultiPV support.
Performance vs. PlentyChess 0.3.0
STC
Elo | 139.43 +- 3.31 (95%)
Conf | 8.0+0.08s Threads=1 Hash=16MB
Games | N: 25004 W: 11963 L: 2435 D: 10606
Penta | [20, 570, 3548, 6590, 1774]
https://openbench.yoshie2000.de/test/807/
LTC
Elo | 137.16 +- 4.44 (95%)
Conf | 40.0+0.40s Threads=1 Hash=64MB
Games | N: 12502 W: 5659 L: 965 D: 5878
Penta | [3, 196, 1697, 3814, 541]
https://openbench.yoshie2000.de/test/808/
Improvements
The following improvements have been made since version 0.3.0:
Much stronger and larger neural network
Various search improvements
Much improved time management
Many speedups
More accurate eval normalization
Added hashfull to the UCI output
Added support for the UCI movestogo parameter
Multiple fixes regarding zobrist keys, as well as other bugfixes
Binary guide
generic: Slowest
avx2: Faster, compatible with most modern CPUs
bmi2: Faster, includes avx2, not recommended for Zen1/2 CPUs
avx512: Fastest, includes bmi2
Comments
Post a Comment