# gbso

An implementation of the stochastic superoptimization algorithm [STOKE](./docs/asplos13.pdf) for the Sharp LR35902 (also known as GB-Z80).

## Setup

Make sure you have [Poetry](https://python-poetry.org/) and Python 3.9+ installed. Then run:

```
$ poetry install
```

## Usage

Input programs are specified as a list of CPU instructions. For a list of instructions available to the processor, take a look at [insn.py](./gbso/cpu/insn.py) or a [standard reference](https://gbdev.io/pandocs/CPU_Instruction_Set.html).

Once you have the program specified, call `gbso.optimize.optimize` on the program. A superoptimized, behaviorally equivalent program will be returned. Additional parameters are available to the optimization function to control search.

For a comprehensive example, see [ex.py](./ex.py).

## TODO

- [ ] Cost penalties for undefined memory/register reads
- [ ] "Improved" equality metric that accounts for correct results in the wrong location
- [ ] Automatic test case generation from counterexamples
- [ ] `DAA` instruction
- [ ] 16-bit memory/register outputs