Performance and enterprise-grade manageability are core to the NVIDIA GRID experience. Our recent August 2016 release continued to raise the bar on enterprise-grade manageability and user experience with the industry’s first vGPU monitoring.
Now, with Citrix, we’re continuing to provide more options for improved user experience and management. The latest developments in Citrix XenDesktop 7.11, combined with NVIDIA GRID accelerated H.264 encoding (called NVENC), deliver improved performance, decreased latency, and increased scalability.
Raising the Bar on Performance
Thanks to offloading encoding, NVENC reduces Citrix XenDesktop host CPU utilization by over 36%.1. Jason Kyungho Lee of the NVIDIA Performance Engineering team, was able to measure a 12% FPS (frames-per-second) increase compared to CPU-based H.264 encoding, with almost no increase in network bandwidth over XenDesktop 7.8. Tests also showed NVENC delivers a more consistent XenDesktop 7.11 user experience than with traditional CPU-based encoding.
The improvements come in part from the impact of the Citrix software graphics process (CTXGFX.exe) offloaded from CPU to NVENC. Our testing reveals a dramatic increase to host scalability by 30%, based on 22% less VM memory usage and 93% less VM CPU usage – all accomplished without impacting performance.
So now you can turn the “knobs” on your infrastructure to meet your desired goal: delivering an upgraded experience to your existing end users or adding even more users without taking a hit on performance. You get twice the bang for the buck: XenDesktop performance, and host scalability.
End users will also see a dramatic improvement to perceived latency. Our internal testing showed a consistent 55ms of latency reduction in XenDesktop 7.11 when compared with 7.8. This is the difference between a virtual desktop or app that feels “spongy” versus one that feels “snappy,” like you would expect with a local, physical PC experience.
Unlike many optimizations that force trade-offs in other design considerations, NVENC offers its benefits with no such compromise. Latency improvement offers users better experience or they can use the performance gain to extend their distance from data center, working further away (think global offshoring scenarios, distributed teams, etc.). Offloading from CPU to GPU allows IT administrators to increase density without sacrificing performance or to offer more performance at the density they have today.
Extending the Scale and Reach of XenDesktop 7.11
Citrix XenDesktop 7.11, combined with our NVIDIA Tesla M10 GPU and NVIDIA GRID software, delivers the ability to host up to 128 users in a typical 2U server, without sacrificing performance or user experience. Organizations can now deliver and manage superior, native-like user experience to all user types from designers to office workers.
GPU-Powered Monitoring Insights = Better User Experience
With the latest NVIDIA GRID August 2016 Release, Citrix users also get NVIDIA GRID Monitoring. NVIDIA GRID Monitoring provides IT with new visibility into their environment and allows them to intelligently design, manage, and support their end user’s experience, ensuring delivery of native-like virtual desktops from the data center with enterprise-grade manageability.
Windows Server 2016: Day One Support for Discrete Device Assignment
More good news – NVIDIA GRID comes with Windows Server 2016 day one support. Citrix deployments on Hyper-V can immediately benefit from support for Discrete Device Assignment (DDA) also known as GPU-passthrough, delivering workstation performance from Microsoft Hyper-V based data centers. Learn more about NVIDIA GRID and Microsoft solutions.
1- Based on internal NVIDIA Performance Engineering testing.