Nvidia announces the Grace CPU for data centres

by Mark Tyson on 13 April 2021, 10:11

Tags: NVIDIA (NASDAQ:NVDA), ARM

Quick Link: HEXUS.net/qaeqg2

Add to My Vault: x

At GTC21 Nvidia has announced its first data centre CPU. Dubbed 'Grace', in honour of US computer pioneer Grace Hopper, the new processor is Arm-based and designed to tackle the most complex AI and HPC workloads. In its targeted tasks Nvidia claims Grace "will deliver 10x the performance of today’s fastest servers," when tightly coupled with Nvidia GPUs.

Nvidia is now a three chip company

"Leading-edge AI and data science are pushing today’s computer architecture beyond its limits – processing unthinkable amounts of data," Nvidia CEO Jensen Huang. "Using licensed Arm IP, Nvidia has designed Grace as a CPU specifically for giant-scale AI and HPC. Coupled with the GPU and DPU, Grace gives us the third foundational technology for computing, and the ability to re-architect the data centre to advance AI. Nvidia is now a three-chip company."

In the statement above you may have noticed the reference to 'giant-scale' computing. By focusing on this type of computing it seems that Nvidia has found a worthwhile advantage. Nvidia has found that in workloads such as training next-generation NLP models that have more than 1 trillion parameters a Grace CPU and Nvidia GPU system can be 10x faster than a state-of-the-art Nvidia DGX system which uses an x86 processor.

Other technologies key to accelerated systems powered by Grace include; NVLink to provide a 900GB/s connection between Grace and NVIDIA GPUs (up to 30x the bandwidth of rival servers), a LPDDR5x memory subsystem for speed and efficiency, unified cache coherence, HBM GPU memory.

Two supercomputer labs are already planning to adopt Grace. Both the Swiss National Supercomputing Centre and Los Alamos National Laboratory plan to intro HPE made Grace powered supercomputers in 2023.

Other major GTC day-one announcements

A roadmap of sorts outlining yearly leaps of GPU, CPU and DPU (Data Processing Unit) leaps was revealed yesterday by Nvidia. Between now and 2025 Nvidia plans to refresh Ampere GPUs twice, Bluefield DPUs twice, and intro a second gen Grace CPU. Hopefully it will come up with better names than 'Ampere Next Next' before launch dates.

Nvidia has outlined its "data centre on wheels" – the Drive Atlan autonomous vehicle platform. As you can see from the main image this platform is going to centre on an SoC using Nvidia's newest next gen technologies.

A 4x generational performance improvement is promised with Atlan, and it is claimed to deliver 1,000 TOPS with its Grace-Next CPU, Ampere-Next GPU, Bluefield DPU, other accelerators and high speed I/O. Expect Atlan to sample in 2023, destined for vehicles hitting the roads in 2025.

Nvidia announced eight new Ampere architecture GPUs for designers, architects and engineers. For desktops, the new Nvidia RTX A5000 and RTX A4000 GPUs feature new RT Cores, Tensor Cores and CUDA cores to speed AI, graphics and real-time rendering "up to 2x faster than previous generations". Similarly, Nvidia has prepared the new RTX A2000, RTX A3000, RTX A4000 and RTX A5000 laptop GPUs. Lastly, for data centre based virtualisation there are the new Nvidia A10 GPU and A16 GPUs. These data centre GPUs are claimed to deliver up to 2.5x the virtual workstation performance and up to 2x user density with lower TCO.



HEXUS Forums :: 13 Comments

Login with Forum Account

Don't have an account? Register today!
And it is revealed why Nvidia wants to purchase ARM, they want to be on the frontline of CPU production for datacentre driven deep learning.

Oh what an edge that would be being both the master and consumer of the very resource that is being used.

And this is exactly why Nvidia should not be allowed to acquire ARM, the writing is on the wall that if the acquisition goes through, competitive ARM offerings in the trickle down DL to server space will always be behind Nvidia. If history is anything to go by, if the acquisition were to go through, Nvidia has years where it can stifle competition before it'll get pulled up and many more years before Nvidia will be charged and many more years after that where Nvidia will be forced to do anything.

No licensee should become the owner of the source, not Nvidia or AMD or Intel.
Tabbykatze
And it is revealed why Nvidia wants to purchase ARM, they want to be on the frontline of CPU production for datacentre driven deep learning.

Oh what an edge that would be being both the master and consumer of the very resource that is being used.

And this is exactly why Nvidia should not be allowed to acquire ARM, the writing is on the wall that if the acquisition goes through, competitive ARM offerings in the trickle down DL to server space will always be behind Nvidia. If history is anything to go by, if the acquisition were to go through, Nvidia has years where it can stifle competition before it'll get pulled up and many more years before Nvidia will be charged and many more years after that where Nvidia will be forced to do anything.

No licensee should become the owner of the source, not Nvidia or AMD or Intel.

Couldn't agree more. Unfortunately this is in the hands of people who don't understand how this works at a level beyond “boxes of magic and pixies” because their interest is economics and markets. It worries me that the important, technical details which make this a really terrible idea will be beyond them (much like money beyond “here, have some magic paper” is beyond me).
Looks more like a SoM (System on Module) to me than a CPU, but ok Nvidia, you go ahead and call it a CPU.
Bigger is better, right? Also Good for social media marketing.
We still think that NVidias acquisition of ARM will get denied?
'[GSV
Trig;4289855']We still think that NVidias acquisition of ARM will get denied?

I think it can go one of two ways now.

The US (at least the remnants of Trump but Biden ain't exactly rolling back very quick) would do anything for it to be US controlled because to deny China ARM would be highly destructive to China.

However, there seem to be just as many sensible people going “hold the fawn” and seeing Nvidia as a dangerous owner. I mean, look at the GeForce Partner Program that we still don't know all the details of, Nvidia could easily do that to ARM in far more nuanced ways.

The supposed benefits of Nvidia injecting Mellanox and DL capabilities into ARM are not worth the risk of it being narrowed to an extension of Nvidias empire. Companies tend to hate dealing with Nvidia, Linux foundation is always in tenderhooks, Apple would rather rub their soft appendages through glass, not sure exactly where MS stands but even Nvidias own GPU partners have sour tastes.

ARM is like Mjolnir, you should be worthy to wield it, otherwise it should be left to its own devices.