The task is too much for one :uxn:? Ask a friend for help.
It's crazy how good the stack architecture maps on parallelism. Also, it's crazy how self-modification throws a wrench in parallelism.
The task is too much for one :uxn:? Ask a friend for help. It's crazy how good the stack architecture maps on parallelism. Also, it's crazy how self-modification throws a wrench in parallelism. 10 comments
[DATA EXPUNGED]
@neauoire Yep, there's a good reason highly parallel code tends to prefer pure functions and read-only data. @chigo non-destructive operations are linear in a way that allows you to jump at any point and reduce it. @chigo A different way to looking at this is "graph reduction", but most of uxn's opcode fail at this, but still an interesting idea if you're interested in parallel computing, is to have a stack of interaction between nodes(operations between registers?), and reduce for as long there are reduceable items on the stack. If your entire system is based on the IN six reduction rules, you have something like linear combinators. |