Ask HN: Low level multiprocessor implementation?

I was reading https://news.ycombinator.com/item?id=8798799 about what languages to learn. I was interested in how Terra leverages lua, a technique I learnt on BBC basic to develop ROMs for the BBC micro. Sort of like a super assembler.

I've been programming for over 40 years, and thought I'd build a framework for the parallela board. I don't see why we have great hardware coming along yet there's no vision in the tool chains. Just C/C++ and openCL. Don't get me wrong, I've done C++ programming since late 80's, it's just I know I can't put a clojure or JavaScript run time on nodes in an epiphany chip.

I'm aware of various activities. As much as I enjoyed my time working with Forth (Jupiter Ace) I'm not sure that developing solutions for the greenArrays chips in ColorForth will become mainstream.

I know there are 'technology clusters' but they are mostly academic talk shops. Here I expect pragmatic answers. I want to do something.

So, my question:

What small systems should I consider to load onto 64 nodes (4 boards) that could enable embedded development with visualisation?