GPUPerf 3.0 is now available for download
The idea of this tol was actually from another tool called FRAPS, that shows in real-time how many frames per second (FPS) that the GPU produces. It’s displayed as an overlay to applications so you can see the numbers at the same time as you see the applications. What I wanted was to see how many frames per second was delivered to the endpoint when you run a graphical workload in a Citrix session. I was able to extract this information together with some other GPU and Citrix protocol related data and display on the top of the screen the same way as FRAPS. Now if you run both tools you can see the difference in what the GPU produces and what is delivered to the endpoint.
In version 2 of GPUPerf I added support for PCoIP and remote FX too. Now it was possible to actually compare different protocols and get some number to how many FPS is delivered to the endpoint, what bandwidth is required, and what is the CPU and GPU load of your virtual machines.
With Nvidia Grid vGPU it was not possible to get the GPU Utilization. After the September 2016 release of Nvidia Grid drivers a couple of new things was introduced. vGPU monitoring was now possible and with additional information about the NVEnc engine. NVenc is the hardware h.264 encoder that modern remote protocols use for encoding h.264. Now it was time to add this data to GPUPerf. Luckily it was quite easy to implement the new monitoring counters. So I’m very proud to be able to share version 3 of GPU perf with the community. In version 3 you can now monitor in real-time vGPU utilization and NVenc utilization. In addition, I’ve added support for two new protocols, Citrix Framehawk, and VMware blast. I’ve also extracted how much CPU is utilized in the encoding process. This reveals how much CPU you can save by enabling hardware h.264 encoding.
Almost zero cpu used for hdx encoding with HD video in chrome on xendesktop with nvenc enabled pic.twitter.com/P7r3RCTY8t
— Magnar Johnsen (@MagnarJohnsen) October 24, 2016
See how GPUperf3 reveals CPU and GPU enc utilization with Windows 10, Edge, YouTube 4k video over @citrix xendesktop 7.11 and @nvidia nvenc pic.twitter.com/fZRiFvoaRW
— Magnar Johnsen (@MagnarJohnsen) October 24, 2016
In this video you can see how GPUPerf3 reveals information on GPU and CPU when running a Youtube video on XenDesktop.
This next video shows how Autodesk 3DStudio max is using the GPU and CPU on XenDesktop
Here is a complete list of GPUPerf 3 supported metrics.
You can download GPUPerf3 here
GPUPerf is a free community tool developed by and for EUC community. You are using it on your own risk, no support is porvided.
If you like this tool and would like to contribute to the development of it, please donate here:
Pingback: Citrix Policy Settings - Carl Stalhood
Pingback: Detailed Change Log - Carl Stalhood
Hello,
We are using Tesla M60 with XenServer 7, we followed the Citrix and NVIDIA procedures to implements the GPU card but, when the tool to mensure the GPU utilization, the counter GPUUtilization every stay in 0% and my CPU is increased…
We’re using GPU Passthrough to W2k8R2 and NVIDIA drivers are installed on VM…. on device manager, the Tesla M60 is exhibited properly.
Hi.
Have you installed nvidia WMI when you installed the drivers? Have you run the GPUSwitch mode tool to set the M60 in graphics mode?
http://images.nvidia.com/content/pdf/grid/guides/GRID-gpumodeswitch-UserGuide-Apr2016.pdf
Also check this article https://docs.citrix.com/en-us/xenapp-and-xendesktop/7-6/xad-hdx-landing/xad-hdx3dpro-gpu-accel-server.html
Magnar
shame that the tools does not support yet the amd Mxgpu yet.
i’m running a compare between M60 and S7150x2
i would have loved to have the realtime stats with both.
My tools rely on Nvidia API. Intel and AMD has not made it simple to access monitor data. I know tools like process monitor and uberagent can get GPU data for all types of GPU using some undocumented DirectX api, but it requires c++ programming, and I only do .net
All right i had a disscussion with amd today as i borrowed them two of their biggest card for Poc they told me that they work on it to catch the gap with monitoring and vsphere integration.
@Benoit noteris , is there any news regarding AMD’s MxGPU monitoring?
Great tool. How do I interpret the two numbers divied by a slash? For example, FPS: 30/3. Rendered / displayed?
First one is realtime, the other is average