Cluster computing with three Raspberry Pi 3 boards

By Russell Barnes. Posted

How to build a cluster computer using three Raspberry Pi boards networked together.

We’re currently enjoying this demonstration on how to quickly turn three Raspberry Pi 3 boards into a cluster computer.

In this video, Kurt Braun shows you how to get the Raspberry Pi boards networked and talking to each other. Kurt uses three Raspberry Pi 3 boards, a WAGO ethernet switch, and a power supply along with an 8GB microSD card for each Raspberry Pi (referred to in the system as a “node”).

It’s a neat solution as well, using a DIN Rail Mount to hold the three Raspberry Pi’s vertically on an equipment rack.
Kurt goes on to explain how to install software to run MPI (Message Passing Interface) applications. He uses MPI For Python, which enables Python programs to exploit multiple processors.

Cluster computing with multiple Raspberry Pi boards

We last looked at building a cluster networked computer back in the Pi Spark Supercomputer Cluster, but this is a much simpler solution.

The world’s fastest computer is currently China’s Sunway TaihuLight with 10,649,600 CPU cores. The twelve cores created by three Raspberry Pi 3 units might seem less impressive, but it’s a great educational environment.

It’s pretty hard to get your hands on high-performance computing (HPC) technology because of the vast expense. But building your own micro version of a cluster computer enables you to explore big data computing using software like Apache Spark on a Raspberry Pi cluster.

A cluster of Raspberry Pi’s is known as a “Bramble” (and sometimes a “Beowulf” cluster) and the biggest bramble ever built is believed to the made by GCHQ, who used 96 Raspberry Pi devices. This is used by GCHQ to teach big data crunching techniques.

Subscribe and never miss an issue

 Get a free Pi Zero W with every twelve-month print subscription

Get a a brand new Raspberry Pi Zero W, a case for it, and a selection of adapter cables with a twelve-month print subscription to The MagPi!