hello.txt

To the next generation:

This is the wrong branch for the core of camvolution, try the one called "Refactor". I'm pretty sure it contains the correct chisel code,
but I'm not going to bother setting everything up to confirm that. I'm sorry git is as fucked up as it is, but rest
assured that taking shortcuts did in no way end up saving any time...

In order to make sense of daisy (the core processor, see the report) you're better off studying some of the earlier commits.
At the end I was desperate and I was kinda throwing stuff at the wall seeing what would stick. The interesting parts of Daisy
is the core, not the double barrel buffers and flow control etc. You should take a look at one of the commits where the
software emulated processor manages to properly convolute an entire image. You should also take a look at the makeshift
tester which prints processor states (in this branchs final commit there's a file called "scratchpad.scala", look for something
similar in the commit logs if you're interested in how I did testing)

I believe testing in chisel has come a lot further than when I wrote all these shite tests, and I was pretty clueless about
scala, so I think you can do a lot better. I imagine if I made something to parse logs allowing me to step through snapshots
instead of 10 000 lines of log.txt it would have been a lot easier.

The readme for Refactor mentions some C software convolution, it's from a TDT4200 exercise, but it should work well enough.
Don't be dumbasses, don't use C unless you know you need it (hint: you don't)

In short, use this branch for the VHDL HDMI module and use the Refactor (sorry! sorry!) for the chisel code.


You know where we went wrong, it's up to you to find new and exciting ways to go wrong. Good luck!