Way back in 2012, when the original Raspberry Pi was introduced, it managed to shake the whole computing industry because of its tiny form-factor and unlimited potential for experimentation with its usage and programming.
The Raspberry Pi, since then, has become a staple product for those wishing to explore different ways of computing and programming. It has allowed people to come up with creative and custom solutions to their problems and then program it all themselves.
These days, people use a Raspberry Pi for all sorts of stuff like Web servers, Wi-Fi extenders, security cameras and, to automate their daily lives. However, we’re not quite done exploring the Raspberry Pi yet as different companies continuously try to innovate new uses for the little beast.
The tiny computer has shown a ton of potential in terms of managing to produce some serious processing power as multiple Raspberry Pis can be linked together to form a bigger ‘Pi Cluster’ with combined computing power.
So, consequently, one of the newer uses of the Raspberry Pi has been trying to figure out different ways to extract the maximum computing power out of Pi Clusters. This has resulted in some seriously impressive and innovative Pi Clusters that can potentially change computing as we know it. The Los Alamos Lab, for example, managed to conjure up a Raspberry Pi Supercomputer in 2017 using a Pi Cluster consisting of over 750 Raspberry Pis, giving it over 3,000 cores to operate with.
Oracle has built the world’s most powerful Raspberry Pi Supercomputer
After Los Alamos Lab, Oracle has decided to step up and flex its own version of a massive Pi Cluster Supercomputer. The Oracle Raspberry Pi Supercomputer consists of a network of 1,060 Raspberry Pi 3 B+ devices, which were launched last year. Oracle is flaunting it’s “extremely large take on a mobile device” as the “world’s largest Raspberry Pi cluster”.
The Pi Cluster, therefore, has a whopping 4,240 cores that It can use for processing. While the original design for this Oracle Pi cluster consisted of 1,024 Raspberry Pis in a UK-styled telephone booth, the design that is on display actually has 1,060 Raspberry Pi nodes in a glass housing instead of the telephone booth.
The Raspberry Pi 3 B+ devices being used in the Pi Cluster were launched last year and cost $35. In such a measly price, the Raspberry Pi 3 B+ contains a fairly powerful 1.4 GHz 64-bit ARM based Cortex A-53 Quad-core processor. The 3 B+ offers a 15% performance gain over the previous Model B. Oracle also released the specifications for the whole Pi Cluster setup. These included 22 network switches, 49 custom 3D-printed holders for the Raspberry Pi modules, 18 USB power supplies, 5 six-foot racks, and, obviously, a plethora of networking and power cables.
As far as the operating system is concerned, the Oracle Pi Cluster is booted via a Supermicro 1U Xeon server and runs on the newly announced Oracle Autonomous Linux. The Oracle Autonomous Linux is the company’s latest autonomous operating system and is based on the Oracle Linux (which also powers the Oracle Cloud and Engineered Systems).
How much does the Oracle Supercomputer cost?
If you’re wondering whether you could build a supercomputer of your own using cheap Raspberry Pi clusters or not, the answer is that you probably could if you’re rich. While a single Raspberry Pi 3 B+ costs only $35, getting 1,060 of these bad boys together would cost over $37,000.
Furthermore, this doesn’t even take the additional costs of putting all these things together with all the extra equipment into consideration. However, $37,000 is still rather dirt cheap considering the sheer processing potential of this massive Pi Cluster.
Many companies have been keen on utilizing the Raspberry Pi for different purposes. Building a supercomputer out of Pi Clusters is certainly one of the most impressive feats in the world of computing.
As time goes by, more companies are producing their own take on this concept and want to continue breaking barriers. Los Alamos Lab currently has a plan of scaling its version of the supercomputer up to 40,000 cores, which would make it over 10 times bigger than the previous effort. It would be interesting to see whether these newer concepts actually turn out to be useful computing products or are just companies flexing on each other by slapping a bigger number on their credentials.