Hey everyone!

First off, I'm extremely sorry for the lack of communication here. We're decently active in the discord channel with updates, but I sometimes forget that not everyone is there. So here is an update.

A while back we posted an update that the AI was able to make quantum moves and that it would be seen in the next patch. This is still true. The AI has been able to make quantum moves for a while now in internal test versions, but it is incredibly slow. At the time it was thought that just some performance optimizations would fix this and we would be able to release the update after a couple of weeks of profiling.

Profiling and optimization did help, but not nearly enough (think 30 seconds to 1 minute per move for complex positions). After a few months of going round and round with optimizing and rethinking and rewriting large parts of the engine structure it has come to a critical point. What the engine really needs is ground up overhaul. Everything from how the board state is represented internally to how move outcomes are calculated needs to be redone and optimized.

To that end I've started a ground up rewrite of the engine side of things in C++. This will give us access to some of the language's innate better performance, and allow us to use a lot of the optimization techniques already developed for other fast Chess engines written in C++. As a nice side effect this will also open the door for running the engine as a standalone which could make a future mobile version easier to implement.

Going forward I'll try to be more diligent about posting updates here.

Thank you everyone for your continued support,
-Chris

Hey everyone!

First off, I'm extremely sorry for the lack of communication here. We're decently active in the discord channel with updates, but I sometimes forget that not everyone is there. So here is an update.

A while back we posted an update that the AI was able to make quantum moves and that it would be seen in the next patch. This is still true. The AI has been able to make quantum moves for a while now in internal test versions, but it is incredibly slow. At the time it was thought that just some performance optimizations would fix this and we would be able to release the update after a couple of weeks of profiling.

Profiling and optimization did help, but not nearly enough (think 30 seconds to 1 minute per move for complex positions). After a few months of going round and round with optimizing and rethinking and rewriting large parts of the engine structure it has come to a critical point. What the engine really needs is ground up overhaul. Everything from how the board state is represented internally to how move outcomes are calculated needs to be redone and optimized.

To that end I've started a ground up rewrite of the engine side of things in C++. This will give us access to some of the language's innate better performance, and allow us to use a lot of the optimization techniques already developed for other fast Chess engines written in C++. As a nice side effect this will also open the door for running the engine as a standalone which could make a future mobile version easier to implement.

Going forward I'll try to be more diligent about posting updates here.

Thank you everyone for your continued support,
-Chris