E-Handbook: How to use GPU for rendering graphics in VDI deployments Article 3 of 4

nobeastsofierce - Fotolia

How to handle increasing VDI graphics processing needs

With more applications, including seemingly straightforward ones, such as Microsoft Word, going graphics-intensive, VDI shops must know how to deliver a quality user experience.

It seems reasonable to assume a designer who uses a slew of graphics-intensive applications probably needs a little extra processing power on virtual desktops. But a task worker dealing with Microsoft Word probably doesn't.

That used to be the accepted paradigm with VDI, but the VDI graphics processing landscape has changed. VDI shops must prepare for the fact that users working with mainstream apps need high-level graphics processing. Even apps such as Word use graphics acceleration by default these days.

VDI shops must know why this change has taken hold and where they can turn to make sure users get the experience they expect and require.

How do graphics reach users in VDI deployments in the first place?

In a VDI deployment, a remote display protocol (RDP)  is essentially the go-between that takes the images from the host server and delivers them to the end user's device. Each RDP works a little bit differently, which determines the quality of the image users receive. The graphics processing unit (GPU) takes the burden of rendering images off the CPU. It does so by decoding the information from the host server through rapid mathematical calculations.

Microsoft, VMware and Citrix all offer RDPs. Microsoft's RemoteFX, which is part of Microsoft's Remote Desktop Protocol, delivers Windows desktops to users. It can virtualize physical GPUs when necessary. VMware's proprietary display protocol is Blast Extreme. Its claim to fame is taking VDI graphics processing off the CPU and moving it to the GPU. And Citrix offers HDX, which delivers high-definition images for XenApp and XenDesktop shops. There are also a host of third-party RDPs that VDI shops can consider, including Parallels Client and Xtralogic's browser-based RDP Client.

Why do mainstream apps require graphics acceleration now?

In short, users expect it, and newer applications and OSes require it. Users expect to be able to view videos or load large images without a hitch -- no matter what device they use, no matter where they are and no matter how they access their desktops. A large reason for this is that users work with graphics acceleration in their everyday lives now, whether they know it or not. Their smartphones, tablets and laptops all use graphics acceleration to give them high-quality videos and visuals on the go. Not to mention the fact that most websites require graphics acceleration to function nowadays. And app vendors want to deliver the best possible user experience, so they pretty much all include graphics acceleration in their products.

IT professionals can fight back if they want to. They can, for example, turn graphics acceleration off in Microsoft Word because it might not seem like a big deal for that particular application. If they do, however, the app's performance suffers, which stunts user productivity and causes more problems.

What can VDI shops do?

It's critical for IT pros to be open to change when it comes to GPUs and VDI.

It's critical for IT pros to be open to change when it comes to GPUs and VDI. It's distinctly possible they will have to take a GPU-per-person approach to ensure that every user gets the graphics performance they need, even with mainstream apps. They also must be sure to keep everything updated and test all their apps to make sure everything performs correctly on users' endpoints. In addition, they must be open to adding additional hardware and software that can help their cause.

Virtual GPUs (vGPUs) can help. A vGPU takes all the VDI graphics processing off users' devices and puts it on the host server. All the devices have to do is decode what the server sends them. On the hardware side, Nvidia has several graphics processing card options for IT to pick from. IT should pick a card based on the resources its users need. The M60, for example, is geared toward getting the most out of graphics performance. Nvidia's software options include:

  • Nvidia GRID Virtual Applications: Delivers Windows apps to XenApp and Remote Desktop Session Host shops;
  • Nvidia GRID Virtual PC: Prioritizes the user experience for browsers and videos; and
  • Nvidia Quadro Virtual Data Center Workstation: Allows graphics-intensive apps to function across different device types and locations.

Intel's Graphics Virtualization Technology and Advanced Micro Devices' MxGPU are the other primary players in the vGPU market.

Next Steps

Detailed guide to GPU virtualization

How to choose the right vGPU product

Graphics acceleration brings the speed

Dig Deeper on Virtual desktop delivery tools

Enterprise Desktop
Cloud Computing