So, long story (involving cross stitching, my 3 year old daughter, an a good kind of crazy coworker short) I’ve been bitten by the retro gaming bug. It’s a pretty deep hole to fall into. In addition to setting up a veritable shrine to nintendo in the TV room, I’ve inspired me to dust off my NES in an FPGA project knes. After quick look I decided to basically start over. Why? I’m glad you asked.
Code Quality
First lets get this out of the way. I’m not a digital designer. I don’t even play one on TV. So when I took a look at the almost 10 year old code, I was horrified. At the time, I remember being frustrated by the design. Looking at it now, I can’t make sense of it. There are very few comments describing the design intent. The few “unit tests” that did exist, just drove the modules and didn’t test the output. The one diamond in the rough was the 6502 assembly tests and instruction set coverage analyzer. Those parts I’ll keep. At least in spirit.
New Shit’s Come to Light Man!
When I was working on knes, Visual 6502 was just getting started. In the 9 years since. A lot more information about the internals of the 6502 has been researched and published. Also, while I was aware of Beregnyei Balazs’ Schematic I was not aware of Donald F. Hanson’s excellent Block Diagram. That diagram gave me great insight into the actual architecture of the 6502
A New Hope
Last weekend I started over, this time sticking to the architecture shown in the Hanson diagram. I’m trying to comment the code well as well as write good unit tests for the blocks. Here’s the state of the code at the time of writing this post.
I’ve also started vectorizing the Hanson diagram and annotating it with the net names I’m using in the verilog design. Check out the current progress (also at the time of writing this post). Non-gray block are block that are implemented.
To Infinity and Beyond
Let’s see where this takes me! I’m pretty stoked.