GPUPerf – realtime GPU and HDX info

[wr_row][wr_column span=”span12″][wr_text]

At Citrix Synergy 2014 i showed a demo of Oculus Rift on HDX3Dpro, and I used a small tool to show real-time info about GPU usage and HDX info overlaid on the screen. This is a simple tool I developed in visual studio, which pulls information from NVidia and Citrix WMI counters. To use this tool you must have NVidia and HDX WMI counters installed. It’s designed to work with XenDesktop 7.x and NVidia Grid/Quadro

The tool will these show real-time counters, updated every 1 second:

  • HDX FPS
  • HDX Latency
  • HDX Input session bandwidth
  • HDX Output session bandwidth
  • GPU Computing usage
  • GPU Memory
  • GPU Temp
  • GPU Power

Combined with fraps you will be able to compare the app GPU FPS with HDX FPS.

This tool will not work with vGPU, only for physical GPU or GPU passthough.

Server hosted machines (XenApp) is not working properly with this version of GPUPerf.

You can download this tool for free here:

DISCLAIMER: This tool is completly unsupported, the use of this tool is on your own risk. The data is based on what Nvidia and Citrix wmi counters provide, I’m not responsible for the validity of the data provided by this tool.


1374 Downloads

Here are some video’s I made that shows this tool in use.

[/wr_text][/wr_column][/wr_row]

7 thoughts on “GPUPerf – realtime GPU and HDX info

  1. Our satellite links are routinely 1200-2000ms (below 60S). Were not attempting to run XD, but we want to upgrade our XA to 7.5. I am a little concerned about the removal of speedscreen and wondered how HDX deals with this? The apps in question are fairly simply data entry ones.

    • 1200-2000ms is really high when working with Citrix. I would not recommend to work with Citrix over such latency. I don’t know what happens with speedscreen in 7.5, but I know that when the Framehawk technology is integrated with xenapp in the future, it will help a lot on high latency connections. You should be able to get sat links with lower latency than that.

  2. Pingback: How Nvidia vGPU is sharing compute power and how to monitor and manage it. | Virtual eXperience

  3. Pingback: How Nvidia vGPU is sharing compute power and how to monitor and manage it. (updated) | Virtual eXperience

  4. Pingback: How Nvidia vGPU is sharing compute power and how to monitor and manage it. (updated) | Virtual Experience

  5. Pingback: I’m 100% sure that 100% is not 100% | Virtual eXperience

  6. I’m very interested in how you got this working, I’m currently trying to get the new Rift 2 to work on some Amazon instances (G2) with the GPU passthrough, and XenDesktop 7.8… would you mind giving a few us a hand, or at least some pointers?!

Leave a Reply

Your email address will not be published. Required fields are marked *

*