For those that haven't been following the NVIDIA GRID Social Media or News sites:
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
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
For anyone interested, I've attached a complete list of GRID 5 vGPU profiles. This covers every supported GRID enabled GPU:
That is: M6, M10, M60, P4, P6, P40, P100 & P100x
This is indeed a big announcement with many changes, including support for the new P6, the introduction of QoS to the Pascal GPU line, a reworked virtual Workstation (Quadro Viirtual Datacenter Workstation), enhanced monitoring capabilities, plus changes to the licensing server and delivery, and more. For a couple of blogs out already that cover more in detail, see the following with undoubtedly more to follow:
[url]https://www.wondernerd.net/blog/my-nvidia-grid-5-0-testing/[/url]
[url]https://www.wondernerd.net/blog/changing-the-gpu-virtualization-game/[/url]
[url]https://www.jitslangedijk.com/the-new-nvidia-grid/[/url]
We all waiting for new updates, hope they will improve our products!! too bad i have to spend my money on [url=https://rxcoupons.org/]drug coupons[/url] now. i would wish so much to get a new nvidia product. my pc is already getting too slow...
We all waiting for new updates, hope they will improve our products!! too bad i have to spend my money on drug coupons now. i would wish so much to get a new nvidia product. my pc is already getting too slow...
When should we expect to see support for the P40 with VMware Horizon 7.x?
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!
Hi TOuimet
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 ...
[quote="TOuimet"]BJones,
Thank you for the additional information above about the P40. We have a desire to utilize the P40 cards in Dell PowerEdge R740 servers running VMware vSphere 6.5 and Horizon 7.x. However, I am finding NO documentation on the P40 specifically related to VMware Horizon and additionally the P40 is not listed on VMware's HCL. When do you expect further information and support to be released for the P40 and VMware?
Thank you![/quote]
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
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 ...
TOuimet said:BJones,
Thank you for the additional information above about the P40. We have a desire to utilize the P40 cards in Dell PowerEdge R740 servers running VMware vSphere 6.5 and Horizon 7.x. However, I am finding NO documentation on the P40 specifically related to VMware Horizon and additionally the P40 is not listed on VMware's HCL. When do you expect further information and support to be released for the P40 and VMware?
Thank you!
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.
There have been quite a few GRID-related announcements at VMworld over the last couple of days, so perhaps documentation will be forthcoming. Have you asked for help on either Twitter or LinkedIn, where you are likely to get some attention?
There have been quite a few GRID-related announcements at VMworld over the last couple of days, so perhaps documentation will be forthcoming. Have you asked for help on either Twitter or LinkedIn, where you are likely to get some attention?
Anyone requiring detailed information about GRID 5.0 should look here: https://docs.nvidia.com/grid/latest/index.html
The "[b]Virtual GPU Software User Guide[/b]" section has an abundance of useful information: https://docs.nvidia.com/grid/latest/grid-vgpu-user-guide/index.html
Regards
[i]My opinion - Nvidia fails again in communication:
[list]
[.]There is no documentation for "encoder capacity" API.[/.]
[.]There are the schedulers "Best Effort Scheduler" and "Equal Share Scheduler" mentioned in docs but no information how to set/manage them.[/.]
[.]"QoS scheduler" is unknown. (all previous is about NVML - [url]https://devtalk.nvidia.com/default/topic/1023524/system-management-and-monitoring-nvml-/-vgpu-management-qos-api-/[/url])[/.]
[.]CUDA is enabled in all Pascal P*-*Q vGPU profiles. ([url]https://gridforums.nvidia.com/default/topic/769/general-discussion/cuda-opencl-support/post/4801/#4801[/url])[/.]
[/list]
[/i]
Hi mcerveny,
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
Hi mcerveny,
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?
[quote="sschaber"]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.[/quote]
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 ([url]https://gridforums.nvidia.com/default/topic/743/talks-with-the-developers/gpu-scheduler-for-vgpu/[/url]) ... [i]Never mind, I do not try to interfere with NVidia's Visions.[/i]
[quote="sschaber"]What information are you missing around CUDA for vGPU profiles?[/quote]
Not found in presentation but found in docs.
sschaber said: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.
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.
sschaber said:What information are you missing around CUDA for vGPU profiles?
Check the recent VMware conference for announcements regarding support for newer GRID hardware and modules. As Ben indicated above, it's up to VMware to certify any releases that support their product lines.
Check the recent VMware conference for announcements regarding support for newer GRID hardware and modules. As Ben indicated above, it's up to VMware to certify any releases that support their product lines.
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