NVIDIA, Mellanox accelerate SA's supercomputer project

Read time 3min 30sec
Kevin Deierling, chief marketing officer at Mellanox Technologies.
Kevin Deierling, chief marketing officer at Mellanox Technologies.

US-based technology firm NVIDIA's acquisition of Mellanox Technologies will play a significant role in some of the high-performance computing projects under way in SA.

Mellanox Technologies is a California-based supplier of high-performance, end-to-end interconnect solutions for data centre servers and storage systems. In March, the company announced it had reached a definitive agreement under which NVIDIA will acquire Mellanox for approximately $6.9 billion.

Speaking to ITWeb on the side-lines of the NetEvents EMEA IT Spotlight conference in Barcelona last week, Kevin Deierling, chief marketing officer at Mellanox, discussed some of the local projects Mellanox is involved in.

He explained that NVIDIA's recent acquisition of Mellanox, which is still under way, will have a positive impact on two main local projects. Firstly, the Square Kilometre Array (SKA) radio telescope project, an international effort to build the largest and most sensitive radio telescope in the world. Secondly, Mellanox is involved in building the Centre for High-Performance Computing's (CHPC's) most powerful supercomputer in Africa.

NVIDIA, which has had a longstanding partnership with Mellanox, designs and manufactures computer graphics processors, chipsets and related multimedia software. The company operates through two segments: graphics processing units (GPUs) and Tegra processors.

"The acquisition will unite two of the world's leading companies in high-performance computing. Together, NVIDIA's computing platform and Mellanox's interconnect solutions power over 250 of the world's top 500 supercomputers and have, as global customers, every major cloud service provider and computer maker.

"At the moment, we are partners more than anything. We are building two big projects that I'm aware of in SA and one is the SKA where we provide intelligent interconnect solutions towards the project. The enormous amount of data being generated by the telescope requires high-bandwidth, low-jitter networks. In addition, this data has to be analysed. So we're the super-high-speed network that connects the telescope through our Ethernet switches and optical cables," he explained.

The SKA is not a single telescope, but a collection of thousands of antennas to be spread over 3 000km. It will be built in two main phases: phase one in SA and Australia; and phase two expanding into other African countries, with the component in Australia also being expanded.

Once it is completed, the radio telescope will conduct transformational science to improve our understanding of the universe and the laws of fundamental physics, monitoring the sky in unprecedented detail and mapping it hundreds of times faster than any current facility.

In February, the Department of Science and Technology, which is spearheading the project, announced in a statement, that following the successful review of the key infrastructure components, the project will now move on to the bridging phase, which will bring together all the individual detailed designs of elements of the SKA and integrate them on a system level.

The SKA project uses NVIDIA's GPUs, which are also instrumental in building the fastest supercomputer on the African continent, a project spearheaded by the CHPC and originally unveiled in 2016.

"The role of NVIDIA and Mellanox is focused on high-performance computing, artificial intelligence and machine learning technologies used in the supercomputer. Nvidia creates gaming GPUs, and Melanox provides end-to-end InfiniBand and Ethernet interconnect solutions and services," said Deierling.

In addition, server and storage solutions provider Boston has collaborated with Mellanox and the CHPC to host local roadshows this month at the University of South Africa and University of Cape Town, he noted.

"These local events will explore the current state of high-performance computing within academic institutions and discuss the evolution of artificial intelligence from the edge to the data centre, as well as how to utilise private cloud to enable the creation of a secure, scalable, cost-effective and flexible IT infrastructure."

Login with