Hi,
to tidy up the certificate situation at edx.org a bit I somehow happened into the cs50x course, and to finish it off I submitted a small 6502 emulator project complete with stdio support and firmware written in assembler, so it can be run in a terminal window as just another command line tool. Now I had the admittedly quite silly idea to see what can be done in direction of machine learning with such a minuscule device. We are talking an 8-bit CPU here, with very few registers, and a 16 bit address space that is good for directly addressable 64KB of memory. To nudge it a bit more in direction to practicability, I am almost finished with extending it to W65C02 mode, which makes it emulate the CMOS-version with better interrupt handling and power saving modes. I think I also have a silicon version lying around somewhere in a DIL packaging. They are really cheap, so copying it in hardware should be really easy. It may have access to external equipment through an IO address space with a nominal clock rate around roughly a MHz (for now, this should be made configurable). Now the goal for this exercise is really to see how far I come with the application part and how much the inference logic can really be shrunk down (Probably ogling the Annual Mad Scientist Award here :-). I am preparing some standard rom module to experiment with the virtual machine itself and then later plan to prepare some core ROM functionality in Assembler and C. It probably is a complete waste of time, but much fun. If anyone wants to take a look, it is available at GitHub - asgalon/haiku6502: A 6502 emulator with stdio/ncurses interface , very much a construction site with lots of building material strewn around. Hard hat required.
Well I thought I might as well announce its existence here. It comes with an Educational Community License, for what it’s worth. Still waiting for any kind of feedback or sign of life here after two weeks, so if this does not provoke at least one reaction I am off for good, declaring this community demised