Blog

In 2013 I made a quick prototype based on an idea I had for a mobile phone game. It was a sliding puzzle game with a difference. Each tile in the puzzle would have it’s own physical layout and physics properties and the goal would …

Right now the Monkeys write purely random, and mostly garbage, Z80 code which is fed into a genetic algorithm which then shapes it into something resembling a real solution to a problem. There were several reasons why I wanted to add support for a real-world …

So far the monkeys have been able to produce simple code to solve simple problems using a very simple machine language of just 24 instructions. I want to see what happens if we give them a more complex, real world processor to play with. Processor …

The monkeys can now count successfully to 8 using a conditional loop, thanks to the genetic algorithm generator. Next, I want to see if they can solve some new problems. It’s often the first program you write when learning how to code, so let’s try …

While working on some new exercises for the monkeys, I realised we’d need some new functionality in the virtual machine. This would then need to be exposed to C# via the C++/CLI interface and then displayed in the WPF-based UI. For me, these latter steps …

This Monkey’s Gone To Seven We left our monkeys with the ability to count to 3 in a mere 20 minutes. I’m proud of them but it’s not going to earn them any bananas. The next step is to count to 8. I think this …

Lazy Monkeys Left overnight, our monkeys are able to count to 3. It’s not terribly impressive, really, so let’s see what we can do to reduce that time. Multicore Monkeys The current system, being completely random, is perfect for parallelization. We have a single unit, …

Now we have a system that can run code and examine the results of that execution, but I’m still concerned that this whole concept could fall at the first hurdle, so it’s time to test out the code generation. First Exercise – Write non-zero byte …