For details of which Ubuntu releases are supported, see Hypervisor Software Releases.

Compatible NVIDIA vGPU Manager and Guest VM Driver Releases

The following combinations of NVIDIA vGPU Manager and guest VM driver releases are compatible with each other.

• NVIDIA vGPU Manager with guest VM drivers from the same release
• NVIDIA vGPU Manager with guest VM drivers from different releases within the same major release branch
• NVIDIA vGPU Manager from a later major release branch with guest VM drivers from the previous branch

The following table lists the specific software releases that are compatible with the components in the NVIDIA vGPU software 14 major release branch.

Incompatible NVIDIA vGPU Manager and Guest VM Driver Releases

The following combinations of NVIDIA vGPU Manager and guest VM driver releases are incompatible with each other.

• NVIDIA vGPU Manager from a later major release branch with guest VM drivers from a production branch two or more major releases before the release of the vGPU Manager
• NVIDIA vGPU Manager from an earlier major release branch with guest VM drivers from a later branch

The following table lists the specific software releases that are incompatible with the components in the NVIDIA vGPU software 14 major release branch.

New Features in Release 14.4

• Security updates - see Security Bulletin: NVIDIA GPU Display Driver - November 2022, which is updated shortly after the release date of this software and is listed on the NVIDIA Product Security page

New Features in Release 14.3

• Security updates - see Security Bulletin: NVIDIA GPU Display Driver - November 2022, which is posted shortly after the release date of this software and is listed on the NVIDIA Product Security page
• Support for non-transparent local proxy servers when NVIDIA vGPU software is served licenses by a Cloud License Service (CLS) instance
• Miscellaneous bug fixes

New Features in Release 14.2

• Security updates - see Security Bulletin: NVIDIA GPU Display Driver - August 2022, which is posted shortly after the release date of this software and is listed on the NVIDIA Product Security page
• Miscellaneous bug fixes

New Features in Release 14.1

• Security updates - see Security Bulletin: NVIDIA GPU Display Driver - May 2022, which is posted shortly after the release date of this software and is listed on the NVIDIA Product Security page
• Miscellaneous bug fixes

Hardware and Software Support Introduced in Release 14.1

• Support for Ubuntu 22.04 LTS as a hypervisor
• Support for Ubuntu 22.04 LTS as a guest OS

New Features in Release 14.0

• Support for NVIDIA GPUDirect^® Storage technology
• Support for GPUDirect technology on all C-series vGPUs on GPUs that support SR-IOV
• Support for GPU System Processor (GSP) in GPU pass through and bare-metal configurations on Linux with vCS
  Note: If you are using a product other than vCS, you must disable GSP as explained in Virtual GPU Software User Guide.
• Enhanced NVIDIA CUDA Toolkit support:
  • NVIDIA CUDA Toolkit profilers can be enabled when unified memory is enabled.
  • Nsight Systems GPU context switch trace is supported.
• Enhancements to the NVIDIA Management Library (NVML) to determine whether a vGPU type supports GPUDirect technology and peer-to-peer CUDA transfers over NVLink
• Addition of RPM and Debian packages for the NVIDIA vGPU software graphics drivers for Linux
• Security updates - see Security Bulletin: NVIDIA GPU Display Driver - February 2022, which is posted shortly after the release date of this software and is listed on the NVIDIA Product Security page
• Miscellaneous bug fixes

Hardware and Software Support Introduced in Release 14.0

• Support for the following GPUs:
  • NVIDIA A2
  • NVIDIA A30X
  • NVIDIA A100X
  • NVIDIA RTX A5500

Features Deprecated in Release 14.0

The following table lists features that are deprecated in this release of NVIDIA vGPU software. Although the features remain available in this release, they might be withdrawn in a future release. In preparation for the possible removal of these features, use the preferred alternative listed in the table.

This release of NVIDIA vGPU software on Ubuntu provides support for several NVIDIA GPUs running on validated server hardware platforms. For a list of validated server platforms, refer to NVIDIA GRID Certified Servers.

The supported products for each type of NVIDIA vGPU software deployment depend on the GPU.

GPUs Based on the NVIDIA Ampere Architecture

GPUs Based on the NVIDIA Turing™ Architecture

GPUs Based on the NVIDIA Volta Architecture

GPUs Based on the NVIDIA Pascal™ Architecture

GPUs Based on the NVIDIA Maxwell™ Graphic Architecture

Supported vGPUs

Only Q-series and C-series time-sliced vGPUs that are allocated all of the physical GPU's frame buffer are supported. MIG-backed vGPUs are not supported.

Maximum vGPUs per VM

NVIDIA vGPU software supports up to a maximum of 16 vGPUs per VM on Ubuntu.

Supported Hypervisor Releases

Since 14.1: Ubuntu 22.04 LTS, 20.04 LTS, 18.04 LTS

14.0 only: Ubuntu 20.04 LTS, 18.04 LTS

Supported vGPUs

Only Q-series and C-series time-sliced vGPUs that are allocated all of the physical GPU's frame buffer on physical GPUs that support NVLink are supported.

Supported Hypervisor Releases

Peer-to-Peer CUDA Transfers over NVLink are supported on all hypervisor releases that support the assignment of more than one vGPU to a VM. For details, see Multiple vGPU Support.

Supported Guest OS Releases

Linux only. Peer-to-Peer CUDA Transfers over NVLink are not supported on Windows.

Limitations

• Only direct connections are supported. NVSwitch is not supported.
• Only time-sliced vGPUs are supported. MIG-backed vGPUs are not supported.
• PCIe is not supported.
• SLI is not supported.

Supported vGPUs

All C-series vGPUs on physical GPUs that support single root I/O virtualization (SR-IOV) are supported.

• GPUDirect RDMA technology is supported on both time-sliced and MIG-backed vGPUs that meet these requirements.
• GPUDirect Storage technology is supported only on time-sliced vGPUs that meet these requirements. It is not supported on MIG-backed vGPUs.

Supported Guest OS Releases

Linux only. GPUDirect technology is not supported on Windows.

Supported Network Interface Cards

GPUDirect technology is supported on the following network interface cards:

• NVIDIA ^®ConnectX^®- 7 SmartNIC
• Mellanox Connect-X 6 SmartNIC
• Mellanox Connect-X 5 Ethernet adapter card

Supported vGPUs

Only Q-series and C-series time-sliced vGPUs that are allocated all of the physical GPU's frame buffer on physical GPUs that support unified memory are supported.

Supported Guest OS Releases

Linux only. Unified memory is not supported on Windows.

Limitations

• Only time-sliced vGPUs are supported. MIG-backed vGPUs are not supported.

Description

The NVIDIA hardware-based H.264 video encoder (NVENC) does not support resolutions greater than 4096×4096. This restriction applies to all NVIDIA GPU architectures and is imposed by the GPU encoder hardware itself, not by NVIDIA vGPU software. The maximum supported resolution for each encoding scheme is listed in the documentation for NVIDIA Video Codec SDK. This limitation affects any remoting tool where H.264 encoding is used with a resolution greater than 4096×4096. Most supported remoting tools fall back to software encoding in such scenarios.

Workaround

If your GPU is based on a GPU architecture later than the NVIDIA Maxwell^® architecture, use H.265 encoding. H.265 is more efficient than H.264 encoding and has a maximum resolution of 8192×8192. On GPUs based on the NVIDIA Maxwell architecture, H.265 has the same maximum resolution as H.264, namely 4096×4096.

Description

Issues occur when the channels allocated to a vGPU are exhausted and the guest VM to which the vGPU is assigned fails to allocate a channel to the vGPU. A physical GPU has a fixed number of channels and the number of channels allocated to each vGPU is inversely proportional to the maximum number of vGPUs allowed on the physical GPU.

When the channels allocated to a vGPU are exhausted and the guest VM fails to allocate a channel, the following errors are reported on the hypervisor host or in an NVIDIA bug report:

Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0): Guest attempted to allocate channel above its max channel limit 0xfb
Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0): VGPU message 6 failed, result code: 0x1a
Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0):         0xc1d004a1, 0xff0e0000, 0xff0400fb, 0xc36f,
Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0):         0x1, 0xff1fe314, 0xff1fe038, 0x100b6f000, 0x1000,
Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0):         0x80000000, 0xff0e0200, 0x0, 0x0, (Not logged),
Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0):         0x1, 0x0
Jun 26 08:01:25 srvxen06f vgpu-3[14276]: error: vmiop_log: (0x0): , 0x0

Workaround

Use a vGPU type with more frame buffer, thereby reducing the maximum number of vGPUs allowed on the physical GPU. As a result, the number of channels allocated to each vGPU is increased.

Description

Issues may occur when graphics-intensive OpenCL applications are used with vGPU types that have limited frame buffer. These issues occur when the applications demand more frame buffer than is allocated to the vGPU.

For example, these issues may occur with the Adobe Photoshop and LuxMark OpenCL Benchmark applications:

• When the image resolution and size are changed in Adobe Photoshop, a program error may occur or Photoshop may display a message about a problem with the graphics hardware and a suggestion to disable OpenCL.
• When the LuxMark OpenCL Benchmark application is run, XID error 31 may occur.

Workaround

For graphics-intensive OpenCL applications, use a vGPU type with more frame buffer.

Description

In pass through mode, all GPUs connected to each other through NVLink must be assigned to the same VM. If a subset of GPUs connected to each other through NVLink is passed through to a VM, unrecoverable error XID 74 occurs when the VM is booted. This error corrupts the NVLink state on the physical GPUs and, as a result, the NVLink bridge between the GPUs is unusable.

Workaround

Restore the NVLink state on the physical GPUs by resetting the GPUs or rebooting the hypervisor host.

Description

Using the frame buffer for the NVIDIA hardware-based H.264/HEVC video encoder (NVENC) may cause memory exhaustion with vGPU profiles that have 512 Mbytes or less of frame buffer. To reduce the possibility of memory exhaustion, NVENC is disabled on profiles that have 512 Mbytes or less of frame buffer. Application GPU acceleration remains fully supported and available for all profiles, including profiles with 512 MBytes or less of frame buffer. NVENC support from both Citrix and VMware is a recent feature and, if you are using an older version, you should experience no change in functionality.

The following vGPU profiles have 512 Mbytes or less of frame buffer:

• Tesla M6-0B, M6-0Q
• Tesla M10-0B, M10-0Q
• Tesla M60-0B, M60-0Q

Workaround

If you require NVENC to be enabled, use a profile that has at least 1 Gbyte of frame buffer.

Description

A VM running a version of the NVIDIA guest VM driver that is incompatible with the current release of Virtual GPU Manager will fail to initialize vGPU when booted on a Ubuntu platform running that release of Virtual GPU Manager.

A guest VM driver is incompatible with the current release of Virtual GPU Manager in either of the following situations:

• The guest driver is from a release in a branch two or more major releases before the current release, for example release 9.4.

  In this situation, the Ubuntu VM’s /var/log/messages log file reports the following error:

  vmiop_log: (0x0): Incompatible Guest/Host drivers: Guest VGX version is older than the minimum version supported by the Host. Disabling vGPU.

• The guest driver is from a later release than the Virtual GPU Manager.

  In this situation, the Ubuntu VM’s /var/log/messages log file reports the following error:

  vmiop_log: (0x0): Incompatible Guest/Host drivers: Guest VGX version is newer than the maximum version supported by the Host. Disabling vGPU.

In either situation, the VM boots in standard VGA mode with reduced resolution and color depth. The NVIDIA virtual GPU is present in Windows Device Manager but displays a warning sign, and the following device status:

Windows has stopped this device because it has reported problems. (Code 43)

Resolution

Install a release of the NVIDIA guest VM driver that is compatible with current release of Virtual GPU Manager.

Description

A single vGPU configured on a physical GPU produces lower benchmark scores than the physical GPU run in pass-through mode.

Aside from performance differences that may be attributed to a vGPU’s smaller frame buffer size, vGPU incorporates a performance balancing feature known as Frame Rate Limiter (FRL). On vGPUs that use the best-effort scheduler, FRL is enabled. On vGPUs that use the fixed share or equal share scheduler, FRL is disabled.

FRL is used to ensure balanced performance across multiple vGPUs that are resident on the same physical GPU. The FRL setting is designed to give good interactive remote graphics experience but may reduce scores in benchmarks that depend on measuring frame rendering rates, as compared to the same benchmarks running on a pass-through GPU.

Resolution

FRL is controlled by an internal vGPU setting. On vGPUs that use the best-effort scheduler, NVIDIA does not validate vGPU with FRL disabled, but for validation of benchmark performance, FRL can be temporarily disabled by setting frame_rate_limiter=0 in the vGPU configuration file.

# echo "frame_rate_limiter=0" > /sys/bus/mdev/devices/vgpu-id/nvidia/vgpu_params

For example:

# echo "frame_rate_limiter=0" > /sys/bus/mdev/devices/aa618089-8b16-4d01-a136-25a0f3c73123/nvidia/vgpu_params

The setting takes effect the next time any VM using the given vGPU type is started.

With this setting in place, the VM’s vGPU will run without any frame rate limit.

The FRL can be reverted back to its default setting as follows:

1. Clear all parameter settings in the vGPU configuration file.

   # echo " " > /sys/bus/mdev/devices/vgpu-id/nvidia/vgpu_params

   Note: You cannot clear specific parameter settings. If your vGPU configuration file contains other parameter settings that you want to keep, you must reinstate them in the next step.

2. Set frame_rate_limiter=1 in the vGPU configuration file.

   # echo "frame_rate_limiter=1" > /sys/bus/mdev/devices/vgpu-id/nvidia/vgpu_params

   If you need to reinstate other parameter settings, include them in the command to set frame_rate_limiter=1. For example:

   # echo "frame_rate_limiter=1 disable_vnc=1" > /sys/bus/mdev/devices/aa618089-8b16-4d01-a136-25a0f3c73123/nvidia/vgpu_params

Description

If all GPUs in the platform are assigned to VMs in pass-through mode, nvidia-smi will return an error:

[root@vgx-test ~]# nvidia-smi
Failed to initialize NVML: Unknown Error

This is because GPUs operating in pass-through mode are not visible to nvidia-smi and the NVIDIA kernel driver operating in the Ubuntuhost.

To confirm that all GPUs are operating in pass-through mode, confirm that the vfio-pci kernel driver is handling each device.

# lspci -s 05:00.0 -k
05:00.0 VGA compatible controller: NVIDIA Corporation GM204GL [Tesla M60] (rev a1)
               Subsystem: NVIDIA Corporation Device 113a
               Kernel driver in use: vfio-pci

Resolution

N/A

Description

A licensed client of NVIDIA License System (NLS) fails to acquire a license with the error The allowed time to process response has expired. This error can affect clients of a Cloud License Service (CLS) instance or a Delegated License Service (DLS) instance.

This error occurs when the time difference between the system clocks on the client and the server that hosts the CLS or DLS instance is greater than 10 minutes. A common cause of this error is the failure of either the client or the server to adjust its system clock when daylight savings time begins or ends. The failure to acquire a license is expected to prevent clock windback from causing licensing errors.

Workaround

Ensure that system clock time of the client and any server that hosts a DLS instance match the current time in the time zone where they are located.

Status

Not a bug

Ref. #

3859889

Description

The NVIDIA vGPU software graphics driver fails to load on hypervsiors based on Linux with KVM. This issue affects UEFI VMs configured with a vGPU or pass-through GPU that requires a large BAR address space. This issue does not affect VMs that are booted in legacy BIOS mode. The issue occurs because BAR resources are not mapped into the VM.

Workaround

1. In virsh, open for editing the XML document of the VM to which the vGPU or GPU is assigned.

   # virsh edit vm-name

   vm-name

   The name of the VM to which the vGPU or GPU is assigned.

2. Declare the custom libvirt XML namespace that supports command-line pass through of QEMU arguments.

   Declare this namesapce by modifying the start tag of the top-level domain element in the first line of the XML document.

   <domain type='kvm' xmlns:qemu='http://libvirt.org/schemas/domain/qemu/1.0'>

3. At the end of the XML document, between the </devices> end tag and the </domain> end tag, add the highlighted qemu elements.

   These elements pass the QEMU arguments for mapping the required BAR resources into the VM.

   </devices>
     <qemu:commandline>
       <qemu:arg value='-fw_cfg'/>
       <qemu:arg value='opt/ovmf/X-PciMmio64Mb,string=262144'/>
     </qemu:commandline>
   </domain>

4. Start the VM to which the vGPU or GPU is assigned.

   # virsh start vm-name

   vm-name

   The name of the VM to which the vGPU or GPU is assigned.

Status

Not an NVIDIA bug

Ref. #

200719557

Description

VP9 and AV1 decoding with web browsers are not supported on Microsoft Windows Server 2019 and later supported releases. This issue occurs because starting with Windows Server 2019, the required codecs are not included with the OS and are not available through the Microsoft Store app. As a result, hardware decoding is not available for viewing YouTube videos or using collaboration tools such as Google Meet in a web browser.

Version

This issue affects Microsoft Windows Server releases starting with Windows Server 2019.

Status

Not an NVIDIA bug

Ref. #

200756564

Description

Applications running in a VM request memory to be allocated and freed by the vGPU manager plugin, which runs on the hypervisor host. When an application requests the vGPU manager plugin to free previously allocated memory, some of the memory is not freed. Some applications request memory more frequently than other applications. If such applications run for a long period of time, for example for two or more days, the failure to free all allocated memory might cause the hypervisor host to run out of memory. As a result, memory allocation for applications running in the VM might fail, causing the applications and, sometimes, the VM to hang.

Workaround

If an application or a VM hangs after a long period of usage, restart the VM every couple of days to prevent the hypervisor host from running out of memory.

Status

Resolved in NVIDIA vGPU software 14.1

Ref. #

200724807

Description

If GPU System Processor (GSP) firmware is enabled with an unsupported product in a GPU pass through or bare-metal configuration on Linux, the VM or bare-metal host should fail to acquire a license. This behavior is implemented to prevent the VM or host from being in an unsupported configuration. However, only license acquisition from a networked legacy NVIDIA vGPU software license server fails. For all other license deployments, the VM or host acquires a license and is in an unsupported configuration.

GSP is supported only for vCS in GPU pass through and bare-metal deployments on Linux. If you are using any other product in a GPU pass through or bare-metal deployment on Linux, you must disable the GSP firmware.

The GSP firmware might be enabled with an unsupported product in a GPU pass through or bare-metal configuration on Linux. In this situation, the following error message is written to the licensing event log file when the VM or host attempts to acquire a license:

Invalid feature requested for the underlying GSP firmware configuration.
Disable GSP firmware to use this feature.

For the location of the licensing event log file, refer to Virtual GPU Client Licensing User Guide.

Workaround

Ensure that the GSP firmware is disabled as explained in Virtual GPU Software User Guide.

Status

Resolved in NVIDIA vGPU software 14.1

Ref. #

3523478

Description

After the NVIDIA vGPU software graphics driver for Linux is upgraded from a Debian package, the driver is not loaded into the VM.

Workaround

Use one of the following workarounds to load the driver into the VM:

• Reboot the VM.
• Remove the nvidia module from the Linux kernel and reinsert it into the kernel.

  1. Remove the nvidia module from the Linux kernel.

     $ sudo rmmod nvidia

  2. Reinsert the nvidia module into the Linux kernel.

     $ sudo modprobe nvidia

Status

Not a bug

Ref. #

200748806

Description

If a proxy is set with a system environment variable such as HTTP_PROXY or HTTPS_PROXY, a licensed client might fail to acquire a license.

Workaround

Perform this workaround on each affected licensed client.

1. Add the address of the NVIDIA vGPU software license server to the system environment variable NO_PROXY.

   The address must be specified exactly as it is specified in the client's license server settings either as a fully-qualified domain name or an IP address. If the NO_PROXY environment variable contains multiple entries, separate the entries with a comma (,).

   If high availability is configured for the license server, add the addresses of the primary license server and the secondary license server to the system environment variable NO_PROXY.

2. Restart the NVIDIA driver service that runs the core NVIDIA vGPU software logic.

   • On Windows, restart the NVIDIA Display Container service.
   • On Linux, restart the nvidia-gridd service.

Status

Closed

Ref. #

200704733

Description

Desktop session connections fail for a 2Q, 3Q, or 4Q vGPU that is configured with four 4K displays and for which the NVIDIA hardware-based H.264/HEVC video encoder (NVENC) is enabled. This issue affects only Teradici Cloud Access Software sessions on Linux guest VMs.

This issue is accompanied by the following error message:

This Desktop has no resources available or it has timed out

This issue is caused by insufficient frame buffer.

Workaround

Ensure that sufficient frame buffer is available for all the virtual displays that are connected to a vGPU by changing the configuration in one of the following ways:

• Reducing the number of virtual displays. The number of 4K displays supported with NVENC enabled depends on the vGPU.

  vGPU	4K Displays Supported with NVENC Enabled
  2Q	1
  3Q	2
  4Q	3

• Disabling NVENC. The number of 4K displays supported with NVENC disabled depends on the vGPU.

  vGPU	4K Displays Supported with NVENC Disabled
  2Q	2
  3Q	2
  4Q	4

• Using a vGPU type with more frame buffer. Four 4K displays with NVENC enabled on any Q-series vGPU with at least 6144 MB of frame buffer are supported.

Status

Not an NVIDIA bug

Ref. #

200701959

Description

The names of vGPUs that reside on the NVIDIA A100 80GB GPU are incorrectly shown as Graphics Device by the nvidia-smi command. The correct names indicate the vGPU type, for example, A100DX-40C.

$ nvidia-smi
Mon Jan 25 02:52:57 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 460.32.04    Driver Version: 460.32.04    CUDA Version: 11.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  Graphics Device     On   | 00000000:07:00.0 Off |                    0 |
| N/A   N/A    P0    N/A /  N/A |   6053MiB / 81915MiB |      0%      Default |
|                               |                      |             Disabled |
+-------------------------------+----------------------+----------------------+
|   1  Graphics Device     On   | 00000000:08:00.0 Off |                    0 |
| N/A   N/A    P0    N/A /  N/A |   6053MiB / 81915MiB |      0%      Default |
|                               |                      |             Disabled |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Status

Open

Ref. #

200691204

Description

After a Teradici Cloud Access Software session has been idle for a short period of time, the session disconnects from the VM. When this issue occurs, the error messages NVOS status 0x19 and vGPU Message 21 failed are written to the log files on the hypervisor host. This issue affects only Linux guest VMs.

Status

Open

Ref. #

200689126

Description

NVIDIA GPU Operator doesn't support vGPU deployments on GPUs based on architectures before the NVIDIA Turing™ architecture. This issue is caused by the omission of version information for the vGPU manager from the configuration information that GPU Operator requires. Without this information, GPU Operator does not deploy the NVIDIA driver container because the container cannot determine if the driver is compatible with the vGPU manager.

Status

Open

Ref. #

3227576

Description

The nvidia-smi command shows 100% GPU utilization for NVIDIA A100, NVIDIA A40, and NVIDIA A10 GPUs even if no vGPUs have been configured or no VMs are running. A GPU is affected by this issue only if the sriov-manage script has not been run to enable the virtual function for the GPU in the sysfs file system.

[root@host ~]# nvidia-smi
Fri Dec 23 11:45:28 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 510.108.03   Driver Version: 510.108.03  CUDA Version:  11.6     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  A100-PCIE-40GB      On   | 00000000:5E:00.0 Off |                    0 |
| N/A   50C    P0    97W / 250W |      0MiB / 40537MiB |    100%      Default |
|                               |                      |             Disabled |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Workaround

Run the sriov-manage script to enable the virtual function for the GPU in the sysfs file system as explained in Virtual GPU Software User Guide.

After this workaround has been completed, the nvidia-smi command shows 0% GPU utilization for affected GPUs when they are idle.

root@host ~]# nvidia-smi
Fri Dec 23 11:47:38 2022
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 510.108.03   Driver Version: 510.108.03  CUDA Version:  11.6     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|                               |                      |               MIG M. |
|===============================+======================+======================|
|   0  A100-PCIE-40GB      On   | 00000000:5E:00.0 Off |                    0 |
| N/A   50C    P0    97W / 250W |      0MiB / 40537MiB |      0%      Default |
|                               |                      |             Disabled |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                                  |
|  GPU   GI   CI        PID   Type   Process name                  GPU Memory |
|        ID   ID                                                   Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Status

Open

Ref. #

200605527

Description

The amount of frame buffer listed in a guest VM by the nvidia-smi command for vGPUs on GPUs that support Single Root I/O Virtualization (SR-IOV) is incorrect. Specifically, the amount of frame buffer listed is the amount of frame buffer allocated for the vGPU type minus the size of the VMMU segment (vmmu_page_size). Examples of GPUs that support SRIOV are GPUs based on the NIVIDIA Ampere architecture, such as NVIDA A100 PCIe 40GB or NVIDA A100 HGX 40GB.

For example, frame buffer for -4C and -20C vGPU types is listed as follows:

• For -4C vGPU types, frame buffer is listed as 3963 MB instead of 4096 MB.
• For -20C vGPU types, frame buffer is listed as 20347 MB instead of 20480 MB.

Status

Open

Ref. #

200524749

Description

Upgrading the NVIDIA vGPU software graphics driver in a Linux guest VM with multiple vGPUs might fail. This issue occurs if the driver is upgraded by overinstalling the new release of the driver on the current release of the driver while the nvidia-gridd service is running in the VM.

Workaround

1. Stop the nvidia-gridd service.
2. Try again to upgrade the driver.

Status

Open

Ref. #

200633548

Description

On Linux, the frame rate might drop to 1 frame per second (FPS) after NVIDIA vGPU software has been running for several minutes. Only some applications are affected, for example, glxgears. Other applications, such as Unigine Heaven, are not affected. This behavior occurs because Display Power Management Signaling (DPMS) for the Xorg server is enabled by default and the display is detected to be inactive even when the application is running. When DPMS is enabled, it enables power saving behavior of the display after several minutes of inactivity by setting the frame rate to 1 FPS.

Workaround

1. If necessary, stop the Xorg server.

   # /etc/init.d/xorg stop

2. In a plain text editor, edit the /etc/X11/xorg.conf file to set the options to disable DPMS and disable the screen saver.

   1. In the Monitor section, set the DPMS option to false.

      Option "DPMS" "false"

   2. At the end of the file, add a ServerFlags section that contains option to disable the screen saver.

      Section "ServerFlags"
          Option "BlankTime" "0"
        EndSection

   3. Save your changes to /etc/X11/xorg.conf file and quit the editor.

3. Start the Xorg server.

   # etc/init.d/xorg start

Status

Open

Ref. #

200605900

Description

The ECC memory settings for a vGPU cannot be changed from a Linux guest VM by using NVIDIA X Server Settings. After the ECC memory state has been changed on the ECC Settings page and the VM has been rebooted, the ECC memory state remains unchanged.

Workaround

Use the nvidia-smi command in the guest VM to enable or disable ECC memory for the vGPU as explained in Virtual GPU Software User Guide.

If the ECC memory state remains unchanged even after you use the nvidia-smi command to change it, use the workaround in Changes to ECC memory settings for a Linux vGPU VM by nvidia-smi might be ignored.

Status

Open

Ref. #

200523086

Description

After the ECC memory state for a Linux vGPU VM has been changed by using the nvidia-smi command and the VM has been rebooted, the ECC memory state might remain unchanged.

This issue occurs when multiple NVIDIA configuration files in the system cause the kernel module option for setting the ECC memory state RMGuestECCState in /etc/modprobe.d/nvidia.conf to be ignored.

When the nvidia-smi command is used to enable ECC memory, the file /etc/modprobe.d/nvidia.conf is created or updated to set the kernel module option RMGuestECCState. Another configuration file in /etc/modprobe.d/ that contains the keyword NVreg_RegistryDwordsPerDevice might cause the kernel module option RMGuestECCState to be ignored.

Workaround

This workaround requires administrator privileges.

1. Move the entry containing the keyword NVreg_RegistryDwordsPerDevice from the other configuration file to /etc/modprobe.d/nvidia.conf.
2. Reboot the VM.

Status

Open

Ref. #

200505777

Description

When GPU performance is being monitored, host core CPU utilization is higher than expected for moderate workloads. For example, host CPU utilization when only a small number of VMs are running is as high as when several times as many VMs are running.

Workaround

Disable monitoring of the following GPU performance statistics:

• vGPU engine usage by applications across multiple vGPUs
• Encoder session statistics
• Frame buffer capture (FBC) session statistics
• Statistics gathered by performance counters in guest VMs

Status

Open

Ref. #

2414897

Description

Because of a known limitation with NvFBC, a frame capture while the interactive logon message is displayed returns a blank screen.

An NvFBC session can capture screen updates that occur after the session is created. Before the logon message appears, there is no screen update after the message is shown and, therefore, a black screen is returned instead. If the NvFBC session is created after this update has occurred, NvFBC cannot get a frame to capture.

Workaround

Press Enter or wait for the screen to update for NvFBC to capture the frame.

Status

Not a bug

Ref. #

2115733

Description

When the scheduling policy is fixed share, GPU engine utilization can be reported as higher than expected for a vGPU.

For example, GPU engine usage for six P40-4Q vGPUs on a Tesla P40 GPU might be reported as follows:

[root@localhost:~] nvidia-smi vgpu
Mon Aug 20 10:33:18 2018
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 390.42                 Driver Version: 390.42                    |
|-------------------------------+--------------------------------+------------+
| GPU  Name                     | Bus-Id                         | GPU-Util   |
|      vGPU ID    Name          | VM ID    VM Name               | vGPU-Util  |
|===============================+================================+============|
|   0  Tesla P40                | 00000000:81:00.0               |  99%       |
|      85109      GRID P40-4Q   | 85110    win7-xmpl-146048-1    |     32%    |
|      87195      GRID P40-4Q   | 87196    win7-xmpl-146048-2    |     39%    |
|      88095      GRID P40-4Q   | 88096    win7-xmpl-146048-3    |     26%    |
|      89170      GRID P40-4Q   | 89171    win7-xmpl-146048-4    |      0%    |
|      90475      GRID P40-4Q   | 90476    win7-xmpl-146048-5    |      0%    |
|      93363      GRID P40-4Q   | 93364    win7-xmpl-146048-6    |      0%    |
+-------------------------------+--------------------------------+------------+
|   1  Tesla P40                | 00000000:85:00.0               |   0%       |
+-------------------------------+--------------------------------+------------+

The vGPU utilization of vGPU 85109 is reported as 32%. For vGPU 87195, vGPU utilization is reported as 39%. And for 88095, it is reported as 26%. However, the expected vGPU utilization of any vGPU should not exceed approximately 16.7%.

This behavior is a result of the mechanism that is used to measure GPU engine utilization.

Status

Open

Ref. #

2227591

Description

The command nvidia-smi vgpu -m shows that vGPU migration is supported on all hypervisors, even hypervisors or hypervisor versions that do not support vGPU migration.

Status

Closed

Ref. #

200407230

Description

If the Luxmark application is run on a Linux guest VM configured with NVIDIA vGPU that is booted without acquiring a license, a segmentation fault occurs and the application core dumps. The fault occurs when the application cannot allocate a CUDA object on NVIDIA vGPUs where CUDA is disabled. On NVIDIA vGPUs that can support CUDA, CUDA is disabled in unlicensed mode.

Status

Not an NVIDIA bug.

Ref. #

200330956

Description

On Red Hat Enterprise Linux 6.8 and 6.9, and CentOS 6.8 and 6.9, a segmentation fault in DBus code causes the nvidia-gridd service to exit.

The nvidia-gridd service uses DBus for communication with NVIDIA X Server Settings to display licensing information through the Manage License page. Disabling the GUI for licensing resolves this issue.

To prevent this issue, the GUI for licensing is disabled by default. You might encounter this issue if you have enabled the GUI for licensing and are using Red Hat Enterprise Linux 6.8 or 6.9, or CentOS 6.8 and 6.9.

Version

Red Hat Enterprise Linux 6.8 and 6.9

CentOS 6.8 and 6.9

Status

Open

Ref. #

• 200358191
• 200319854
• 1895945

Description

By default, the Manage License option is not available in NVIDIA X Server Settings. This option is missing because the GUI for licensing on Linux is disabled by default to work around the issue that is described in A segmentation fault in DBus code causes nvidia-gridd to exit on Red Hat Enterprise Linux and CentOS.

Workaround

This workaround requires sudo privileges.

If you are licensing a physical GPU for vCS, you must use the configuration file /etc/nvidia/gridd.conf.

1. If NVIDIA X Server Settings is running, shut it down.

2. If the /etc/nvidia/gridd.conf file does not already exist, create it by copying the supplied template file /etc/nvidia/gridd.conf.template.

3. As root, edit the /etc/nvidia/gridd.conf file to set the EnableUI option to TRUE.

4. Start the nvidia-gridd service.

   # sudo service nvidia-gridd start

When NVIDIA X Server Settings is restarted, the Manage License option is now available.

Status

Open

Description

NVIDIA vGPU software licenses remain checked out on the license server when non-persistent VMs are forcibly powered off.

The NVIDIA service running in a VM returns checked out licenses when the VM is shut down. In environments where non-persistent licensed VMs are not cleanly shut down, licenses on the license server can become exhausted. For example, this issue can occur in automated test environments where VMs are frequently changing and are not guaranteed to be cleanly shut down. The licenses from such VMs remain checked out against their MAC address for seven days before they time out and become available to other VMs.

Resolution

If VMs are routinely being powered off without clean shutdown in your environment, you can avoid this issue by shortening the license borrow period. To shorten the license borrow period, set the LicenseInterval configuration setting in your VM image. For details, refer to Virtual GPU Client Licensing User Guide.

Status

Closed

Ref. #

1694975