Shulou(Shulou.com)06/02 Report--

This article is translated from an article by Magnar Johnsen, which discusses that in Citrix XenDesktop and VMware Horizon solutions, if you use the NVENC features of NVIDIA GRID card, there will be a significant improvement in performance and user experience for users using 3D software. According to personal understanding, if there is any incorrect understanding, please correct it. Thank you!

The original text is as follows: https://www.virtualexperience.no/2016/09/20/why-nvidia-nvenc-hardware-h-264-encoder-is-important-for-citrix-xendesktop-and-vmware-horizon/

The Nvidia Grid video card itself has an H.264 hardware encoder. With Citrix XenDesktop and VMware Horizon products, you can use this feature to encode H.264 protocol streams in hardware. Why are you so smart? If you want to understand this problem, you must first understand the concept of "click-to-photon time". This time refers to the time when you press the mouse button until you see the screen update. How long the whole process takes depends on your Internet speed, but there are other factors as well. Please refer to the following figure to help understand:

The time to click photons depends on all the factors noted in the image above, as well as the delay of computer hardware. For computer display, there may be a delay of a few milliseconds.

With a local computer as a reference, the average click-to-photon time is about 65 milliseconds, where there is no impact of network and coding / decoding delay. A PCoIP session whose click-to-photon time is about 215ms, so about 150ms is spent on the LAN network. For image applications, moving pictures, flipping through documents and web pages, visible to the naked eye, it is said that as low as 120 milliseconds, adults feel a delay, while young people feel a delay of about 100 milliseconds.

To learn more about click photons, please refer to this link: http://www.virtualexperience.no/2016/03/07/how-to-use-click-to-photon-to-measure-end-to-end-latency/

With the NVIDIA NVENC function, this time will be reduced by about 50 milliseconds. This has been tested in VMWare Blast Extreme, and now Citrix XenDesktop 7.11 can also be implemented. This means that users will get a better user experience, or allow an extra 50 milliseconds delay for your WAN links without reducing the click photon time. Why is hardware coding faster than software coding? Because GPU is parallel processing and CPU is serial processing.

Another benefit is that CPU time can be reduced when the server uses H.264 protocol encoding.

I also wrote another article about how to reduce click photon time when using Autocad for clients that enable H.264 hardware decoding, plus increase FPS speed and mouse polling to shorten click photon time. Article link: http://www.virtualexperience.no/2016/08/09/optimizing-autocad-cursor-lag-on-xendesktop/

Why does adding FPS at the protocol level reduce click-to-photon time? That's because the 16FPS stream has about 1000ms/16 = 62.5ms between each zhen, so this time is equivalent to click photon time. If you increase the FPS to 60, there is only about 16.7ms between frames, so you can almost reduce the 50ms latency here. However, adding FPS requires more bandwidth, so it is not recommended to use WAN links. This also increases the load on H.264 encoding and decoding, which is why GPU hardware encoding and decoding allows you to use more FPS without adding load to the server and client.

The following is a differential dynamic diagram of a 16FPS~60FPS, for your reference, to have a more intuitive understanding of different FPS.

I have experimented with virtual VR, just like HTC Vive and Oculus Rift applications running on remote desktops. In this scenario, you need a shorter click photon time, or photon motion time, to avoid feeling dizzy. To achieve this, I hope that remote protocols will support higher FPS in the future. Of course, you also need more powerful GPU, bottom latency and high bandwidth.

So my advice is: if you want to build a VDI platform for the future and give users a better user experience on mobile content, a GPU-enabled VDI is a must. If you choose a NVIDIA grid card, you can not only get GPU video and image acceleration, but also use NVENC for better scalability and click-to-photon time. At present,

Nvidia Grid M 60 supports 36 H. 264 concurrent streams and M 10 has 28.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.