I’m being SUPER careful now because the last two questions seemed to contravene the Discourse rules. I don’t like making waves.
thanks so much for helping out. Here’s my idea ( getting a jetson nano to do this )
from cli ( bash) send in a work load ( could be an entire sub directory or just a file)
take the work load and divide it up into as many work packets as matches the number of gpus
load up each gpu with a module that would perform the work on each packet.
each gpu consumes the relevant packet until complete then checks if all packets consumed
that’s it. In the simple example I have about 5000 text files, each with 5000 lines with about 1000 bytes in each line.
I am looking at DynamicalSystems.jl
I am also looking at some form of finite state machine OF finite state machines.
I am probably going to use JuliaInterpreter,jl with TMUX and possibly emacs ( just want simple)