Subscribe to:

Subscribe to :: ::

NVidia Intends To Stop Data Centers From Using Consumer-Grade Graphics Cards

January 10, 2018 by   | Category: Computing

nVidia Has said “NO” to businesses wanting to uses its consumer-grade GeForce graphics cards in their data centers.

By changing its licensing agreements on its software packages and drivers, Nvidia has moved to push companies away from GeForce cards to its Tesla graphics accelerators.

The increase in performance of consumer GPUs, alongside their relatively wallet-friendly prices, have made GeForce cards more appealing to businesses running data centres.

Previously, the artificial imitation of double-precision floating point capabilities of GeForce GPUs made Nvidia’s enterprise-grade Quadro workstations and Tesla products more appealing to companies looking for good performance with data centre workloads reliant on common general-purpose GPU acceleration.

But the shifts in the workloads have meant half precision not double precision is more appealing, which bypasses the issue of limited performance and has seen companies turn towards cheaper consumer GPUs.

Naturally, Nvidia doesn’t want big businesses buying its cheaper cards when it has enterprise-grade tech to shift.

So a change in licensing terms looks to shift businesses back to Tesla GPUs.

“No Datacenter Deployment. The software is not licensed for data centre deployment,” Nvidia’s updated driver licence agreement notes, laying down the law.

However, it adds: “Except that blockchain processing in a data centre is permitted”, which suggests Nvidia still wants people to use its cards for mining cryptocurrency

How Nvidia will enforce this new licensing agreement will have to be seen, but we expect that it will be a major pain in the posterior for companies making use of GeForce cards in their data centres, potentially resulting in them pulling a load of graphics cards from their server arrays when Nvidia’s legal eagles come swooping in.

For people making use of Nvidia GPUs in their personal computing, this move will mean very little but should help prevent companies from snapping up GeForce cards and potentially driving prices up. And it will prevent Nvidia from throttling performance of GeForce GPUs to limit their data centre appeal.

That being said the strength of the parallel processing found in GPUs for crunching machine and deep learning algorithms means Nvidia GPUs have a strong appeal for developers working on artificial intelligence and smart systems.

As such, this could be the reason why the previously enthusiast-level GeForce Titan GPUs have dropped the GeForce branding, seen pricing hit a hefty $3500, and get marketed for AI-powering use.

While Nvidia’s graphics tech all share a common architecture, it would appear that the previously blurred lines between enterprise and consumer-grade cards are becoming more tightly defined for better or worse.



Tell us what you're thinking...
and oh, if you want a pic to show with your comment, go get a gravatar!

You must be logged in to post a comment.