Topic: Automation / distribution of effort?


I just want to start by saying that the task you've set yourself, and the work you've done thus far, is awesome! I was amazed when I saw's results, and I can't believe I didn't know about this equivalent PSX effort until now. I've being scouring the Wiki and your emu-russia thread (thank you, Google translate ;) ).

That said, it is a huge task. Do you have any plans for automation/distribution of the effort? I saw in your thread a post describing the design for a tool (Chiptunes?) to do distribution, but it doesn't appear to cover automation.

Do you have any hope that a tool (even if only a once-off hack) could be created to, for example, scan the rows of the CPU, identifying and labelling the unique cells it finds? The hi-res images you have seem to indicate this might be feasible.

Also, it seems like hand-drawn traces stored in huge Photoshop files can only go so far. Do you have plans to move to a more specialized format in the future?

I'm looking forward to your future updates!

- hyperbug

Re: Automation / distribution of effort?

Yo cool

Of course we are thinking about automation tool and even tryed Degate (without any success). There are some reasons, it cannot be done easily.

1. Quality of images is not as good. See example of dirty standard cells :
Even the human brain, the most complex device for the analysis is struggling to deal with it (but it still possible). Of course you can skip such cells.

2. Automatic tracing of metall wires cannot be done in some places, because of stitching glitches :

3. Total amount of image data from single chip takes several Gygabytes. Full stitching of huge chips is not impossible because of poor focus on some images and memory limitations.

4. We have to find time to spare for writing such complicated tool. But we have a lot other things to do.
Chiptunes-tool is just experimental tool to test collaboration efforts smile

Re: Automation / distribution of effort?

Do you have plans to move to a more specialized format in the future?

I'm thinking about XML to store vector data for tracing, but again it sense nothing without specialized tool.

And we are using sPlan ( and ummm.. other public places ... smile) for circuits (it can handle custom components in easy way).