It’s been a while since I’ve update you guys on my latest project, here we go!
After loading ArchLinux onto all of the nodes, setting up the networking, and getting all the wiring done juuuuusst right, I ran into a problem with getting MPI working.
MPI, or the message passing interface, is what makes a parallel processing cluster truly parallel. Without it, I’ve just built a 12 node stack of Raspberry Pis that can talk to each other, but not truly work together. I’ve got MPI running on the head node, and now I’m trying to use a command called fstab to clone that install directory to each of the other nodes but for some unknown reason its not working. I’ll keep tinkering to get it to work.
On the bright side, I’ve finished my first piece of MPI enabled code! I’ve taken Olber’s Paradox, a really cool physics problem more commonly known as the Robin Hood problem, and written it to run the calculations on multiple nodes. Because ParallePi isn’t fully operational yet, I’ve been testing it on ManeFrame, which is ironic because my future plans were to test software on ParallelPi before I loaded it to ManeFrame, so this is a nice reversal.
I’ll be back next week with some good news hopefully!