GPU powered VDI – Virtual Desktops with NVIDIA GRID

German Version

User demands on the systems have risen continuously in recent years. In addition, more and more software products – especially the operating systems (Windows) themselves – no longer calculate the graphic representation on the CPU (Central Processing Unit). Instead, they calculate it on the GPU (Graphics Processing Unit) optimized for this purpose. If no GPU is available, the calculation continues to be performed on the CPU. Especially when using virtual desktops – i.e. desktops that are provided in the data center – the lack of GPU is often a clear disadvantage compared to classic PCs. In addition, the load on the corresponding servers is getting significantly higher and the number of users per physical server decreases. The costs per user increase. The industry recognized this disadvantage several years ago and introduced the GPU Pass-through technology. A physical graphics card is assigned to a virtual machine. This can then use the full power of the graphics card, but each virtual machine requires its own graphics card. The costs per user increase. To resolve this issue, NVIDIA has released GRID technology. This makes it possible for several virtual machines to share a graphics card, but still maintain full performance – similar to what already happens in all virtual environments with the CPU.

You will learn what to consider when planning a (v)GPU environment and learn step by step how to deploy it with Citrix and NVIDA. This is followed by monitoring the environment and correcting known errors.

Contents (excerpt):

Plan
(v)GPU Types
License Variants
HDX / HDX 3D Pro

Do
Graphics Cards
License Server
Virtual Delivery Agent

Check
GPU Monitoring
NVIDIA-SMI
GPU Profiler

Act
System Load
3Dconnexion Devices
Troubleshooting

Print-Version:
ISBN (10): 1983043311
ISBN (13): 978-1983043314
Amazon