In 2010 we heard that US Air force used 1760 PlayStation 3 to create a supercomputer. It has a lot of power. The supercomputer has a power of 500 million TFLOPS, using the cluster method. The supercomputer made by Professor Gaurav Khanna of the University of Massachusetts Dartmouth is also famous because he just used 16 PS3 to create a new supercomputer. The US Air force cluster is being used for high-resolution image processing, pattern recognition, and other military use. Building supercomputer using PlayStation 3 is very interesting and it can be great if you also have some spare.
Professor Gaurav Khanna’s used 16 PS3 with Linux OS to do the calculation. The professor created 16 PS3 clusters by the help of the university, donation of Sony, and his own investment. The computer was used for research in black holes. To be specific it is being used to model black hole collision and gravitational waves. Usually, powerful workstation and gaming laptops are expensive. But consoles are not that expensive in comparison. So the use of computational power of PlayStation is actually a smart use. Professor Gaurav Khanna is now on the way to create 220 PS3 cluster supercomputer.
The beloved PlayStation 3 is just awesome because it can give power to both gaming and calculations. Basically, gaming consoles are computers. They gave CPU, GPU, memory, and storage. If we go to a traditional computer, they can be very expensive. Small personal computers can’t handle such a large calculation. Supercomputers are also very expensive for anyone. Using an old but useful piece of technology in such a productive way is a good answer to recycle and reuse concept as well.
Computational Power with Consoles
Professor Khanna said that PS3 worked better than using normal computers in a cluster. The PlayStation 3 actually has powerful octa-core processor in them. This gaming console was released in 2007. Even though new PS4 is out, people still play on the older console. The new generation gaming console from Sony is more powerful. But it turns out that the new PlayStation 4 can’t be used for this purpose.
We can’t blame manufacturers for it. Cluster computing is not the designated use of the console. However, if it would have been possible, we could make more powerful cheap supercomputers. Many universities and colleges in various countries don’t have access to powerful computing. Mainframe and other servers are expensive. Renting is also expensive. Using old consoles like this can solve some of their issues. Institutes can get computational power for cheap and older devices are also utilized. It is not the silver bullet but it can do a lot for many people.
Also Read: US Air Force Space Fence Project
After reading all this and realizing the awesome power of PS3, we think many people want to build their own. The idea is very basic, more console is equal to more power. We know that this gaming console heats up so think of a good cooling system. Professor Khanna used refrigerated shipping containers to keep his supercluster. All you need is a PS3, internet, some skills, and software. These are simple and basic steps to building your own PS3 cluster supercomputer.
How to build a PS3 Supercomputer
This is for educational purpose and we don’t take any responsibility for your device and actions. Please follow these steps to make your own small powerful computer.
- Download Fedora 8 image and burn it on a bootable DVD.
- Download the custom boot image and put it on a USB drive like Pendrive
- Format the PlayStation 3.
- Install the downloaded OS (Fedora 8).
- Setup the SSH, Message Passing Interface and the Cell SDK.
Fedora is a very popular Linux distro and it the OS of choice here. The Fedora 8 came in 2007 and it is the tested one. The newer version might work but there is no guarantee. MPI or Message Passing Interface is one of the important part or standard being used for clustering purpose. It simply helps to connect and multiple computers and enables them to run programs parallelly and exchange messages and instructions.
The detailed instructions are available on the official workshop site of UMass Dartmouth University. If you are a science and computer lover with budget or funding then you should definitely do this. It is one of the best ways to get budget low-cost high-performance computing. Price of HEDT and high-performance computers will come down eventually. Till then we can use methods like these to get the best results from the device we have.