Virtual GPU Technology
Try Virtual GPU for Free
Applications
Where to Buy
Certified Servers
Support
NVIDIA Virtual GPU Cloud Gaming
Authorization Required
Not a member? Register Now
Twitter: https://twitter.com/NVIDIAVirt
News: http://nvidianews.nvidia.com/news)
NVIDIA yesterday announced the latest software upgrade for GRID, which moves to version 5 on September 1st
Here's a link to the latest update:
http://nvidianews.nvidia.com/news/nvidia-quadro-virtual-data-center-workstation-software-turns-tesla-gpu-servers-into-powerful-workstations
This upgrade brings the Pascal architecture to GRID and adds support for Compute and Graphics virtualization from the same GPU without the need for a "Mode Switch". There is a rebranding of the vWS license to vDWS (it's full title is: "Quadro Virtual Datacenter Workstation", which is a bit of a mouth full, so vDWS for short). Also there are now enhanced "end to end" monitoring capabilities (including for the first time "application monitoring") which will be a great addition!
There's lots to take in and understand, so have a read through and if you're not already, then make sure you follow the GRID page on Twitter for the latest updates
Regards
That is: M6, M10, M60, P4, P6, P40, P100 & P100x
https://www.wondernerd.net/blog/my-nvidia-grid-5-0-testing/
https://www.wondernerd.net/blog/changing-the-gpu-virtualization-game/
https://www.jitslangedijk.com/the-new-nvidia-grid/
-=Tobias
https://goo.gl/xQWVuU
Is it up to VMware to certify Horizon with the P40 or
is it up to NVIDIA to certify the P40 with Horizon or
both?
Thanks!
I've quoted this from your other post so the different Topics don't get mixed up, as the other Topic is in the Citrix section ...
Unfortunately I'm not going to be able to offer much guidance here, as this is pretty much all down to VMware.
VMware control which GPUs they support on their platform, so any HCL updates will need to come from them. Obviously, NVIDIA and VMware work together to create the integration in the first place though.
I'm not sure why VMware doesn't have any documentation, again, this is a question for them. Obviously, bear in mind that the GRID 5 + P40 combination are not even available to use yet, so maybe VMware are just being a bit slow - What kind of documentation / information are you after?
P40s are available to purchase now, and have been available since the start of 2017. They will be supported, it's just a matter of when VMware announce it on their HCL.
Sorry I can't offer any more than that, I know it's not what you wanted to hear.
Regards
-=Tobias
The "Virtual GPU Software User Guide" section has an abundance of useful information: https://docs.nvidia.com/grid/latest/grid-vgpu-user-guide/index.html
Regards
what do you mean with "encoder capacity"? Not sure if this is of any relevance for the masses.
New schedulers are correctly explained and also how to set them. Maybe you should have a closer look on the documentation provided. Check the vGPU user guide posted from Ben and see A.:
A. Changing vGPU Scheduling Policy...
What information are you missing around CUDA for vGPU profiles?
Best regards
Simon
Yes, check NVML forum (in NVML docs missing any description and I added reference to other docs).
Is "QoS scheduler" presented by marketing only the stupid "fixed/equal share scheduler" ? If yes, it is not QoS ! NVidia should check (under influence of ex Sun Microsystems) - Solaris Resource Manager, Solaris FSS, Solaris Resource Pools (https://gridforums.nvidia.com/default/topic/743/talks-with-the-developers/gpu-scheduler-for-vgpu/) ... Never mind, I do not try to interfere with NVidia's Visions.
Not found in presentation but found in docs.
-=Tobias