So I have followed this tutorial on how to make a Raspberry PI cluster computer. It works great, I can parallel my processes with MPI, but I'm also wondering if this is possible. Currently, all four of my Pis are running the same thing at the same time, but it doesn't look like they're helping each other out with computing power. If they are, and I'm confused about how this whole thing works, forgive me. But to me, it looks like pi@client is running the same thing, as is pi@node01, pi@node02, pi@node03, etc. Which is fine, unless I want to pool that power. Any clarification and help would be great, thanks.

NOTE: only the green activity light on the client is flashing

MPIExec Example: mpiexec -f machinefile -n 4 echo Test

EDIT: If I change my -n tag in mpiexec, will that run it as a singular process as in all computers are contributing, or only one is doing the work?

  • I skimmed message passing thing MPI. I have been googling for a long while and often heard that message passing is usually the bottle neck. For example, RpiZ AI using camera is "not working", because the message passing through camera and USB (Rpi USB2, which is ridiculously slow). MPI via Ethernet is perhaps fast, but why not use just one Rpi and python multrprocessing to spread the concurrent jobs over 4 CORES? – tlfong01 May 14 '19 at 2:15
  • If you refere to "this tutorial" then please add a reference to the tutorial, ex. url or such. – MatsK May 14 '19 at 4:03
  • @MatsK Oh no, sorry, I thought it was linked – Widdles May 14 '19 at 12:01
  • @tlfong01 And it looks like it is just multiprocessing. I don't see any collaboration, but keep in mind everything has worked according to Tinkernut's instructable – Widdles May 14 '19 at 12:04
  • You might let us know what kind of concurrent processing you are doing. Is it some kind of matrix/vector computing, multi dimension, sparse or else, can data processing be pipelined, or all elements must in memory at the same time. Is multiprocessing multi core most important, or big memory data important. Have you compared and contrasted various storage and transfer methods: RAM, FRAM, SSD, USB3/2, Cat5/6, Wireless N, WiFi 802.11ac, SATA2/3, GPIO DMA, etc, etc, ... – tlfong01 May 15 '19 at 5:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.