Its time to plan updating your NVIDIA TESLA M6, M10, M60, P4, P6, P40, P100, V100, T4, RTX6000, RTX8000 with NVIDIA vGPU software 9.1.
NVIDIA have released new drivers for vGPU 9.1. This new-feature branch is supported until June 2020.
I have in this article also included which Public Cloud instance is available with NVIDIA GPUs and which license is BYO or provided by the public cloud provider such as Alibaba, Azure, AWS, GCP, Alibaba.
For a list of validated server platforms, refer to NVIDIA vGPU Certified Servers.
- Citrix XenServer 7.0, 7.1 is not supported with NVIDIA Tesla T4
- NVIDIA vGPU 9.0 is supported with VMware Horizon 7.9 7.8, 7.7, 7.6, 7.5, 7.4, 7.3, 7.2, 7.1, 7.0, 6.2
- NVIDIA vGPU 9.0 is only supported with Citrix Virtual Apps & Desktops 7.15, 7.18, 7 1903, 7 1906 in HDX 3D Pro mode
- If you are a customer using XenServer 7.2, 7.3, 7.4, 7.5 its no longer supported with NVIDIA vGPU 9.1 and should plan upgrading to XenServer 7.6 or Citrix Hypervisor 8.0
This release includes the following software:
- NVIDIA vGPU Manager version 430.46 for the Citrix Hypervisor, VMware vSphere, RHEL KVM, Nutanix AHV
- NVIDIA Windows driver version 431.79
- NVIDIA Linux driver version 430.46
New Features in Release 9.1
- Starting with release 6.7 U3, the assignment of multiple vGPUs to a single VM is supported.
- Support for NVIDIA vComputeServer vGPUs on the following GPUs:
- Quadro RTX 6000
- Quadro RTX 8000
- Security updates
- Miscellaneous bug fixes
Hardware and Software Support Introduced in Release 9.1
- Support for Citrix Virtual Apps and Desktops version 7 1906
Following bugfixes are solved in vGPU 9.1
|Bug ID||Summary and Description|
After the command nvidia-smi -p 0 is run from a guest VM to reset the volatile ECC error count, the VM crashes.
Incorrect ECC error counts are reported for vGPUs on some GPUs when the command nvidia-smi -q is run from a guest VM.
On Linux guest VMs, the license directory /etc/nvidia/license is not deleted when the NVIDIA vGPU softwaregraphics driver is uninstalled.
nvidia-smi vgpu -q shows the incorrect ECC state of a vGPU when ECC is enabled on the physical GPU but disabled on the vGPU from the vGPU VM. This issue occurs because data for the physical GPU host is not being reset and is being reused even after reboot.
The vComputeServer option is missing from the Manage License section in the NVIDIA X Server Settings window.
When a VM configured with vGPU under heavy load is migrated to another host, users’ view sessions may become corrupted after the migration.