RDSH HD Media

NVIDIA Virtual GPU Forums Virtual GPU General Discussion
#1

Hi,

We are about to implement RDSH (not VDI) in our network, but I got conflicting info from the internet and local vendors, so I hope someone can answer a few questions.

Our RDSH clients requirements are simple, office and some in-house apps (mostly web-based), but the unclear part is being able to view HD media (through browser or a media player).

  • Is a GPU required for this type of use?

  • If not, Will a GPU offload this type of work from the CPU? or will that not work?

  • If I use a GPU, will a single RDSH VM be able to use multiple GPUs?

  • How can I calculate the required memory (and GPU required) per user?

  • How many users per server?

  • Which GPU do you recommend for this type of use?

  • What type of Licensing is needed?

Setup info:
All clients will be limited to a single monitor and a maximum res of 1920x1080.
All clients will be connected to their RDSH server via 1GB.
All servers will be running as a VM on a Hyper-V 2016 server.

Thanks.

#2

Hi SJ,

The GPU especially helps to increase user experience for browsing and video playback.

  • Is a GPU required for this type of use?

->Not required but desired to increase user experience (better FPS, smooth playback…)

  • If not, Will a GPU offload this type of work from the CPU? or will that not work?

->Yes, especially Encode/Decode of H264 as example can be done in hardware on the GPU which will extremly offload the CPU during video playback

  • If I use a GPU, will a single RDSH VM be able to use multiple GPUs?

->No, not yet but for RDSH this is not necessary. We recommend Tesla M10 to run 4 VMs on a single host.

  • How can I calculate the required memory (and GPU required) per user?

->You need to calculate the framebuffer requirements and we see 8GB FB usage for 30-50 CCUs on RDSH with normal office usage depending on the workload

  • How many users per server?

->4VMs per M10 on a single host = 120 - 150 CCUs

  • Which GPU do you recommend for this type of use?

->Tesla M10

  • What type of Licensing is needed?

-> vApps licensing for CCU

Hope that helps

Simon

#3

Hi Simon,

Thank you for taking the time to answer my questions.

  • Is it required to have "Hardware acceleration" enabled for browsers to take advantage of the GPU?
    I ask because I have problems with "Hardware acceleration" enabled in both Firefox and Chrome.

  • We have 20 branches with 5-30 users each, what GPU do recommend for them?

  • According to this document, the Maximum Resolution for vApps is 1280x1024, what happens when a User with a higher resolution screen connects the RDSH server?

  • How will the CCU be accounted for in the case of RDSH?

  • We would like to use a single instance of HA license servers for both HQ and the branches, but sometimes the link between HQ and a branch gets disconnected for sometime, when will their vGPU stop working?

#4
  • Is it required to have "Hardware acceleration" enabled for browsers to take advantage of the GPU?

-> Sure, otherwise you won’t use the GPU :) Not sure what kind of issues you’re seeing with hardware acceleration. Most likely because you don’t have a GPU atm

  • We have 20 branches with 5-30 users each, what GPU do recommend for them?

->Without knowing the detailed infrastructure this is hard to judge. I’m wondering why there is “local infrastructure” at all. Would expect they connect to a central XenApp farm.

  • According to this document, the Maximum Resolution for vApps is 1280x1024, what happens when a User with a higher resolution screen connects the RDSH server?

->No, our documentation states that the max resolution is for the console session. This has nothing to do with the RDP session. For sure you can use higher resolutions and multi monitor.

  • How will the CCU be accounted for in the case of RDSH?

->Every concurrent user session on RDSH needs a vApps license.

  • We would like to use a single instance of HA license servers for both HQ and the branches, but sometimes the link between HQ and a branch gets disconnected for sometime, when will their vGPU stop working?

->Issue would occur with new established connections. Active running sessions are not affected.

#5

Hi,

  • Actually, the issues I have with "Hardware acceleration" is in my personnel experience using NVIDIA GPU, and with no easy way to debug the issue the only solution was to disable it.

  • Branches will have their own servers, because the links between us is very slow and can sometime get disconnected, if you need more info please let me know, because M10 is too much for them.
    BTW, we will not be using Citrix or VMware.

  • What is considered a console session for a RDSH VM?

  • How will the license server determent how many users are active per RDSH VM? will this be done automatically or do I have to manually assign a number?

Thanks.

#6

Hi SJ,

If you only have a single "VM" as example in your subsidiaries it would make sense to simply use a Quadro P2000 for example.

If you don’t connect via RDP on your RDSH host you’re using the console session (for example with XS VNC console or ESX console). So the console session is restricted to 12x10 resolution to prevent license type misusage.
As you use RDP sessions for your users these sessions are working for sure with higher resolutions.

Currently we are not able to count RDSH sessions so you need to know your current users count and our licensing is EULA based. We will enable RDSH user count in our licensing with a future update release.

Regards

Simon