A GROUP of engineering students from Bahrain have constructed a supercomputer that is affordable and consumes less power.
As part of their final project, the Bahrain University students – Arsalan Ahmad from Pakistan, Ishaq Khader and Abdullah Alghoul from Jordan, and Muneeb Jaafar from Sudan – developed their own supercomputer.
They even won first place in an exhibition that showcased all the graduation projects for Bahrain University for that semester.
Supercomputers use a cluster of multiple processors that are connected together in parallel combination.
These multiple processors help the computer to complete a large amount of tasks at greater speed, with multiple processors working simultaneously.
On the other hand, normal computers possess only a single processor, which limits them to execute only a single task at a time.
One of the fastest parallel processing supercomputers is from China, named ‘Sunway Taihu Light’ which costs millions of dollars and consumes thousands of kilowatts of power.
Mr Ahmed told the GDN that it took the group eight months to complete the project and they planned to launch a website with details of the supercomputer.
“To achieve the aim of our project, we had to choose a processor that was suitable for cluster implementation,” explained the 24-year-old.
“We finally came to a conclusion that using multiple single chip board computers can prove to be a feasible recourse.
“It took us around eight months to complete the project including the research work, hardware importing, assembling, software accessibility and finally to write the thesis.
“Soon we are hoping to launch a website of our product, where all the details and documentary of the project will be published.”
The students used eight ‘Raspberry Pi 3 model B’ single board computers in a cluster, out of which one chip acted as the master node, which is the main director and scheduler, whereas the rest were slave nodes.
Mr Ahmed said that the basic working principle of the Raspberry Pi cluster was that whenever the master node received a task to be processed, it divided the task into smaller chunks and fed it to the slave nodes.
The slave nodes work on the task independently, and upon completion, they respond back to the master node, this system leads to faster computing and the tasks being executed more quickly while only using 140watts of power.
“It is an affordable parallel processing supercomputer which is an eminently practicable platform for developers of parallel algorithms to test their programs on this device, before implementing them on larger scales,” added Mr Ahmed.
“Moreover, this inexpensive parallel processing device opens up new horizons not only for educational purposes, to learn parallel processing, but the cluster can be modified to incorporate multiple applications like cryptocurrency mining machine, mini-web server, performing scientific simulations or calculating mathematical operations.”
The entire project cost BD270 to put together and the Raspberry Pi cluster is the first of its kind in the GCC.
Supercomputers are used for a wide range of computationally intensive tasks in various fields, including quantum mechanics, weather forecasting, climate research, oil and gas exploration and molecular modelling.
This article provided by NewsEdge.