Supermicro introduced an Arm-based server – the ARS-210M-NR – in December 2022 Servethehome took it for a spin, load it with four Nvidia A16 cards. The test server was equipped with a 128-core amp 3GHz Altra Max Arm processor, 16 DDR4 memory modules (512GB total memory), two 25GbE-capable SPF28 ports, 16 2.5-inch bays and a pair of 2kW power supplies , all nicely assembled in a tool-less enclosure.
The icing on the cake is the quartet of Ampere-based Nvidia A16 graphics cards. Each card features 64GB of GDDR6 ECC memory and four GPUs sitting on a passively-cooled two-slot, full-height, full-length board. The A16 isn’t the most powerful data center card Nvidia has; that would be the A100; However, it hits a sweet spot for service providers looking to get the maximum number of concurrent users on a single board.
In terms of raw performance, each of these GPUs has half the CUDA cores of the Geforce RTX 3050, so yes, they should be able to run the legendary Crysis game without too much trouble, meaning you’ve got 16 in theory could be iterations of one of the most exhausting games ever run concurrently. You only need to create 16 virtual machines and download 16 copies of crisis.
No pricing was given, but given that the A16 costs around $3,000 and a fully loaded server With similar credentials costing around $10,000, you would likely pay upwards of $22,000 for such a system.
Power to the cloud
Well, who would need such a debauchery of power? First of all, cloud gaming is now mainstream (although Google exited earlier this year when it was mothballed stages) and it’s the ARS-210M-NR that makes it a reality.
But beyond that, it’s about VDI (Video Desktop Infrastructure) for businesses and businesses. The rise of Hybrid work acted as a catalyst for the launch of Virtual Workplace (and Virtual PC) as To work from home became a reality for millions of users. And only one of these servers can accommodate many VMs (virtual machines); a recent test by a cloud computing company Nutanix showed that two Nvidia A16 graphics cards can run 128 virtual machines, which means four cards should be able to run 256.
Of course, you may need to put in a more powerful CPU than the Altra Max, and also more memory (Nutanix used 1.5TB of RAM and two Intel Xeon Gold 6354, each with 18 physical cores). Increasing SaaS support for virtual GPUs in a variety of applications (eg Photoshop or chrome) explains why dense servers with very high GPU counts are becoming more common.
Other applications like web hosting can benefit from a high CPU core count, which is why VPS (Virtual Private Servers) are now affordable with the price gap dedicated servers and Raw metal solutions remain high.
The 2U server was running on Servethehome Ubuntu and while the experience wasn’t as smooth as it could be, it was definitely worth the effort. Arm is slowly becoming a formidable opponent for both Intel’s Xeon and AMD’s Epyc, and Amazon is preparing the fourth generation of it graviton CPU family, there’s never been a better time to try Arm.
