Hi guys, i might be posting this wrong...
I seriously need some feedback on the nvidia M10 2b profile challenges we are seeing.
We are running 14.2 drivers on esxi and golden image.
We are running IC Clones - Non persistent.
Our golden image is running with 4 CPUs, 2 cores pr socket.
10 GB memory
2 and 4 dsplays with 256 MB memory
When this image are running on a Thin OS 9.3.0 on Dell Wyse optiplex 3000, with two and three monitors, the whole system is lagging. Streams are really really bad, and slows down windows big time.
The CPU is running 75-95%, but it seems that the GPU offload is working, since the GPU in the OS is running about 25-45%.
But, I have users that only have 3-4 chrome applications running, spread on two or three monitors, with realtime monitoring of some statistics on our companys website (braodcasting company)
I have tried to deploy a desktop pool without GPU, and it is running much better, especially streaming live streams. The CPU are running 95-100%, and slows down the user experience in windows 10, but the streaming seems to be working alright. I do not see lag in the stream, even if i try to do other thins on the VDI. Its just slow, but working. Not a solution though. Why does the streaming seems to work much better without GPU?
I have tried a lot of different settings.
Enabled 3d rendering, with 4 displays and 256mb video memory.
Disabled 3D rendereing
Enabled and disabled hardware acceleration in chrome...
Why does the GPU works better without GPU for the streaming part? it is very confusing.
The Horizon Performance tracker tells me:
Client to remote session UDP
Remote session til client: UDP
Connection server: TCP
Encoder name: Nvidia NvEnc H264 4:4:4
The problem might be at the client side. If your thin client doesn't support H264 4:4:4 hardware decoding, it will use software decoding which may take a lot of CPU to decode. Can you change to H264 4:2:0 and if the system is still lagging, please check CPU usage of client as well
I give me up to 60fps. But i see that the Tracker reports TCP while we are connected to the local vcenter:
But when i connect to the other site, it reports UDP and works smoother.