Vultr Adds Nvidia A16 GPU for VDI Services

Privately-held cloud computing company Vultr has announced it has added the Nvidia A16 GPU to its array of product offerings to offer virtual desktop infrastructure (VDI) to its customers. The NVIDIA A16 is a purpose-built GPU designed specifically for graphics-rich VDI, claiming that it delivers a user experience nearly indistinguishable from a native PC for remote workers.

Vultr differentiates itself from other cloud service providers by placing emphasis on GPU offerings, and taking full advantage of the Multi-Instance GPU, or MIG features of the A100, which allows for partitioning the GPU into seven virtual GPUs.

Through partitioning similar to hypervisors for CPUs, Vultr offers fractional A100 GPU instances to customers for less than the competition which offer similar services. The company offers a full range of cloud compute, cloud GPU, storage, bare metal, managed databases, Kubernetes, and other cloud-native services powered by GPUs.

The A16 joins the A100 and A40 in the Vultr GPU lineup. “Obviously VDI has been around for a while. And if you asked why don’t more companies use it, frankly, [VDIs] were often not great experiences because they were lagging,” said J.J. Kardwell, CEO of Vultr’s parent company, Constant.

“And now you could have that same experience that’s essentially indistinguishable from having a native PC in front of you. So this is a fantastic product from Nvidia purpose built to support VDI, and this is bringing accelerated computing, right GPU capability to that market,” he added.

Kardwell went so far as to say that it is possible for an end-user to access a virtual desktop that is much more powerful than the physical computer they’re using to access it. For the most part, though, the A16 is meant for general office use for things like browsing, Microsoft Word and Excel, or presentations. If you want to run AutoCAD, that means using a VDI service with A40 acceleration, and Vultr offers that, too.

Cloud GPU instances can be used as remote desktops by connecting to them with clients such as Microsoft Remote Desktop, Parsec, and VNC. Pricing starts at just $21.50 per month, or $0.032 per hour, making these cloud-based virtual desktops an affordable alternative to IT-managed physical personal computers.

“If all you want is basic applications like Excel and browsing the web, you could use the smallest instance and have an accelerated computing experience. And if you are running applications that have more memory needs or more horsepower, you could buy a larger instance to optimize for the exact amount that gives you the experience you want. And that's a big difference from the way GPUs have historically been,” said Kardwell.

There is an issue of latency, which Vultr addresses through its global presence of 30 data centers spread around the world. So clients will be served by the data center physically closest to them.

In addition to individual desktops, Vultr offers shared virtual desktop instances where people in multiple locations could access one desktop at the same time, so they can all work together and see what the other sees. 

Nvidia A16 VDI instances are available through eight of Vultr’s data centers -- Los Angeles, New Jersey, London, Frankfurt, Tokyo, Bangalore, Singapore, and Sydney -- with more to be added later this year.