ARM AND NVIDIA SITTING IN A TREE, p-l-a-y-i-n-g nicely by, er, getting into cahoots over so-called exascale supercomputing.
Team Green, which has a thing for making lunchbox-sized supercomputers, will be making its CUDA-X AI and high-performance computing (HPC) software work nicely with the ARM ecosystem, which means a load of processors based on CPUs and architectures coming out of the Cambridge chip designer.
Usually, supercomputers are built on the x86 architecture with chips provided by the likes of IBM and Intel. And Nvidia's HPC and AI-centric software stack tends to sit comfortably on x86 machines.
But with plans to get CUDA-X libraries running on ARM-based hardware by 2020, organisations looking to set up non-x86 supercomputers will be able to do so with the knowledge that it'll work nicely with Nvidia's supercomputer software.
"Supercomputers are the essential instruments of scientific discovery, and achieving exascale supercomputing will dramatically expand the frontier of human knowledge," said leather-jacker connoisseur Jensen Huang, who's also founder and CEO at NVIDIA.
"As traditional compute scaling ends, power will limit all supercomputers. The combination of Nvidia's CUDA-accelerated computing and Arm's energy-efficient CPU architecture will give the HPC community a boost to exascale."
For those who don't know, 'exascale' basically means grinding through a quintillion floating point computations (FLOPS) per second. That's basically a massive duck-load of computing power that could be used to come up with smarter AI systems or data and computational heavy research work done by academics and other smart folk.
ARM and Nvidia have held hands in public before with latter using the formers chip architecture and platforms for its AI-powering SoCs and platforms. And the Tegra X1 SoC you'll find in the Nintendo Switch uses ARM Cortex-A57 processors.
But the partnership is set to give Nvidia more access to ARM chip designs, which should benefit Team Green's efforts in creating more mobile SoC.
For everyday folks like ourselves, more access to ARM designs could see Nvidia create some pretty slick mobile hardware. While the supercomputer stuff in interesting, we doubt that'll filter down to us mere mortals anytime soon. µ
And that'll cost you a tenner a month
Watch this space
Hackers could erect man-in-the-middle attacks
Painted into a corner