The world’s three biggest makers of computer processors – Intel, NVIDIA, and AMD – are showcasing their latest innovations in the supercomputing space at the SC19 supercomputing conference that’s happening this week.
Marketwatch reports that Intel has already come out swinging, showing off a new Graphics Processing Unit (GPU) that’s specifically designed for workloads like AI training, simulation workloads, and High Performance Compute modelling.
Codename: Ponte Vecchio
The GPU, codenamed Ponte Vecchio, uses Intel’s next-gen 7nm manufacturing process, and is slated to launch sometime in 2021. Intel’s mainstream standalone GPUs, mentioned in an earnings call last month, will come out some time next year.
While Intel has built integrated graphics capabilities into its CPUs for years, standalone graphics cards are a completely new venture for the company, and it’s entering a space that has, up to now, been dominated by the other two big players in the space, AMD and NVIDIA.
Supercomputer GPUs, meanwhile, are proving their worth in supercomputers around the world, showing up in more builds than ever before.
Coming in 2021
Marketwatch says these new super-advanced supercomputer GPUs will make their commercial debut in the Aurora, a supercomputer that’s being built by supercomputing legend Cray for the US’s Department of Energy.
If you like impressive numbers, here’s one: the Aurora is anticipated to produce sustained performance of one quintillion calculations per second! That number is absolutely massive: it’s a one followed by eighteen zeroes! And the Aurora will do that once per second! Amazing.
GPUs are essential to supercomputing
The reason for Intel’s interest is that Graphics Processing Units are excellent at processing specific types of workloads, in many cases more efficiently than general processors do. This makes them an essential part of any supercomputing venture and thus a worthwhile venture for Intel.
Says Marketwatch, “The November 2019 list [of the top 500 supercomputers] once again highlights the increasingly important role of accelerators in maintaining performance gains as core deployments continue to slow,” Wells Fargo analyst Aaron Rakers wrote in a note to clients Monday. “Despite the slowing core deployments, the industry was able to maintain performance gains through the expanding use of accelerators, which now make up 27% of all cores deployed versus 24% in November 2018.”
While this market is not quite as big as the other ones Intel plays in, this is still an important move for the company as it deepens institutional knowledge and opens up new avenues for development and revenue.
It also introduces new competition into a tightly-contested space, which can only be good news for the computing industry as a whole. End-customers, meanwhile, can expect better performance from their compute resources in the future on the back of these developments.