GPU: To Use Or Not To Use?
Anyone who uses Virtual Desktop Infrastructure (VDI) will agree that the technology is several things rolled into one. For starters, it effectively delivers applications and desktops to users, facilitates remote access, streamlines IT administration and boosts productivity. Furthermore, it facilitates BYOD programs and is a proactive disaster recovery solution for workplaces.
While virtualization has delivered on these promises, there was previously room for improvement in a key area: graphics performance. Why is this so? Back then, users mostly worked with applications that provided basic functionality. Today, industry innovations and digital transformation trends are changing the way we work. Applications are constantly evolving to deliver graphically immersive interfaces and high levels of user interactivity. At the same time, the graphics demands of operating systems such as Windows 10 are increasing. This results in a greater need for excellent graphics resources which every IT infrastructure should be capable of supporting.
Luckily, we have remote display protocols that support lower bandwidth, higher bandwidth and multimedia redirection. Hypervisors also feature complete GPU support which can be used on a 1:1 basis or shared across multiple virtual desktops. The purpose of all this is to improve graphics performance and end UX. GPU virtualization is at the core of this because it allows virtual machines to receive GPU benefits like a physical desktop. Essentially, graphics processing which is generally managed by the CPU is offloaded to the server. This supports a greater number of users and boosts overall application performance.
While VDI graphics cards contribute to successful VDI, you must know why they are necessary before making an investment decision. Let’s look at some ways in which GPU technology ensures that your VDI efficiency is top-notch.
Uncompromised User Experience
Conduct a user-based situation analysis. GPU is optional if your staff includes task and knowledge workers as they work with basic and mid-level operational applications. Here’s the tricky side of things. RAM and CPU resources are consumed at high rates in most virtualized workplaces where GPU support is necessary. Those that do not leverage this platform experience slower performance and reduced feature sets. Moreover, maximizing performance in high-latency environments, video-conferencing scenarios and 2D/3D graphics are major challenges where power users are concerned. Fortunately, many VDI providers support NVIDIA virtual GPU technology to accommodate such circumstances.
Keeping this in view, the main goal should be to balance added workloads and improve the user VDI experience. This can be done by simply delegating the heavy lifting to GPUs to take some pressure off CPUs. Let’s consider two common examples. vGPU is crucial for video editors who run 3D imaging apps like CAD or for visualizers dealing with tri-dimensional images. These users cannot expect solid virtualization performance without some form of processing offload. For a better understanding of how exactly graphics virtualization helps such users, let’s consider a couple of scenarios.
Case In Point: Architectural Firm
Consider an architectural company where team members spend more than half of their working days working outside the physical office. They expect a feature-rich and engaging experience anywhere, anytime and on every device, be it a PC, tablet or smartphone. The company’s current setup produces high physical footprint and there is a lack of visibility into staff’s resource demands. This means more telephone calls to the organization’s IT help desk. Administrators find themselves constantly troubleshooting issues and staff members are unhappy and unproductive. Finally, the company considers implementing VDI as a solution to offer designers and engineers remote access to 3D apps. Assume that these groups make up around 30% of the total user base which justifies the move to VDI.
The management installs NVIDIA K1 and K2 boards to enable IT to run around 8-12 users on each shared GPU board. This system not only boosts flexibility and consolidation. VDI deploys the image and delivers desktops to users with 3D accelerated efficiency, wherever they may be. So, the company lower its IT demands and delivers better UX by boosting the basic performance of its VDI setup. From a business perspective, this leads to measurable progress as satisfied employees make it easier to set deliverables by the numbers.
The company can further do more with less by implementing ClearCube’s SmartVDI solution. This is a host platform that seamlessly transitions users from utilizing PCs to using highly available virtual desktops which deliver superior UX. SmartVDI desktops have the ability to manage advanced configurations with similar personalization features as desktop computers. The solution’s openstack architecture consolidates the system into existing VDI setups to serve all user groups. This includes power users that require uncompromised graphics quality and even general audiences like task and knowledge users.
Case In Point: Windows 10
Imagine that your company implements virtualization technologies and you decide to take your IT infrastructure to the next level. You migrate to Windows 10 to keep up-to-date with fast-paced modern work environments. Here’s the question. What happens if your VDI setup does not include NVIDIA GRID vGPU support for power users? To begin with, there will be high CPU usage due to increased graphics consumption. This results in lower server density and increased cost per user. Other factors include scaled-down application functionality and compromised experience on devices like Thin Clients without virtual machine GPU acceleration.
Leveraging NVIDIA GRID™ facilitates a smooth transition and allows you to make the most of the new OS. To start with, this leads to better scalability and performance. This is because the strain is taken off the CPU by offloading graphics workloads to the GPU. Secondly, the application experience is rich and engaging as applications are delivered exactly the way they are designed. This not only facilitates a consistently great experience across all devices. You get to promote mobility and improve teamwork and collaboration. Also, when IT operations are streamlined, tech support needs reduce and the VDI acceptance rate increases. This delivers a native desktop experience on Thin Clients and there is lower latency on work activities. Examples include streaming videos for training and conferencing. As a whole, you get to optimize your VDI setup at a lower cost without sacrificing UX.
Spice It Up With Blade PCs & Hypercube FX
ClearCube offers Blade PCs which deliver powerful desktop to data center performance to fulfill the demands of power users. Their high-density rackmount platforms provide comprehensive 1:1 PCoIP/GPU options for high-precision and data-sensitive applications. Let’s consider an example for CAD/CAM engineers. The M1034W workstation supports PCoIP host-side rendering with CPU/GPU processing that provides fast performance across applications. This Blade PC also integrates NVIDIA Quadro P6000 that features 24 GB of GDDR5 GPU memory with ultra-fast bandwidth. Users can easily utilize this model to create and render advanced models and compute massive databases.
You are also recommended to use Hypercube FX to expand Blade PC efficiencies. This is a 4U chassis with modules that connect to R3092D Blades for better GPU computation and fast storage in the datacenter. Configurations of eGPU and DAS modules provide robust resources for research, engineering, AI and clustered applications. As a whole, the Hypercube FX is a flexible graphics acceleration solution which you can customize to meet multiple use cases.
Please visit our website for further details on how these solutions can help make the most of your VDI environment.