Since you’re here...

We hope you will consider supporting us today. We need your support to continue to exist, because good entries are more and more work time. Every reader contribution, however big or small, is so valuable. Support "Chess Engines Diary" even a small amount– and it only takes a minute. Thank you.
============================== My email: jotes@go2.pl



Chess engine: Velvet 2.0.0

on

Velvet - UCI Chess Engine
Author: Martin Hornet, Rating JCER=3059

 It is based upon my previous web-based engine Wasabi, which can be played here.
Installation
x86_64-modern - recommended for recent CPUs from 2013 onwards (requires a CPU with support for the BMI1 instruction sets)
x86_64-popcnt - for older 64-Bit CPUs, which support the POPCNT instruction, but not BMI1
x86_64-vintage - for older 64-Bit CPUs, which support neither POPCNT nor BMI1

v.2.0.0
The major feature of this new release is the addition of a neural network for position evaluation 
The neural network has been trained from a set of 800 million chess positions from Velvet self-play games.
For this I created two new tools:
gensets: Extracts and labels chess positions from games that Velvet plays against itself
trainer: Trains a neural network using the labeled chess position sets as training data
Changes
Added neural network based evaluation
Optimized slider attack generation using magic bitboards
Refactored and simplified time management
Additional optimizations and minor bug-fixes

Comments

Post a Comment