Validated Platforms#
This release family of NVIDIA vGPU software provides support for several NVIDIA GPUs on validated server hardware platforms, Red Hat Enterprise Linux with KVM hypervisor software versions, and guest operating systems. It also supports the version of NVIDIA CUDA Toolkit that is compatible with R595 drivers.
Supported NVIDIA GPUs and Validated Server Platforms#
This release of NVIDIA vGPU software on Red Hat Enterprise Linux with KVM 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 Blackwell Architecture#
Note
GPUs based on the NVIDIA Blackwell architecture support SR-IOV.
GPUs based on the NVIDIA Blackwell architecture support manual placement of vGPUs on GPUs in equal-size mode.
GPU |
Mixed vGPU Configuration - Red Hat Enterprise Linux with KVM Releases |
|||||
|---|---|---|---|---|---|---|
Frame Buffer Size (Mixed-Size Mode) |
Series |
Time-Sliced NVIDIA vGPU on Single-Instance GPU[4] |
1:1 MIG-Backed NVIDIA vGPU[5] |
Time-Sliced, MIG-Backed NVIDIA vGPU |
GPU Pass Through |
|
NVIDIA RTX PRO 6000 Blackwell Server Edition |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
|
|
NVIDIA RTX PRO 6000 Blackwell Server Edition liquid cooled |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
|
|
Since 20.2: NVIDIA RTX PRO 5000 Blackwell Workstation Edition 72 GB |
9.6 |
9.6 |
|
|
|
|
NVIDIA RTX PRO 4500 Blackwell Server Edition |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
|
|
GPUs Based on the NVIDIA Ada Lovelace Architecture#
Note
GPUs based on the NVIDIA Ada Lovelace architecture support SR-IOV.
GPUs based on the NVIDIA Ada Lovelace architecture support manual placement of vGPUs on GPUs in equal-size mode.
GPU |
Mixed vGPU Configuration - Red Hat Enterprise Linux with KVM Releases |
|||
|---|---|---|---|---|
Frame Buffer Size (Mixed-Size Mode) |
Series |
NVIDIA vGPU |
GPU Pass Through |
|
NVIDIA L40S |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA L40 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA L20 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA L20 liquid cooled |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA L4 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA L2 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA RTX 6000 Ada |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA RTX 5880 Ada |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA RTX 5000 Ada |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
GPUs Based on the NVIDIA Ampere Architecture#
Note
GPUs based on the NVIDIA Ampere architecture support SR-IOV.
The manual placement of vGPUs on GPUs in equal-size mode is not supported on GPUs based on the NVIDIA Ampere architecture.
GPU |
Mixed vGPU Configuration - Red Hat Enterprise Linux with KVM Releases |
|||
|---|---|---|---|---|
Frame Buffer Size (Mixed-Size Mode) |
Series |
NVIDIA vGPU |
GPU Pass Through |
|
NVIDIA A40[6] |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA A16 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA A10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA A2 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA RTX A6000[6] |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA RTX A5500[6] |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
NVIDIA RTX A5000[6] |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
GPUs Based on the NVIDIA Turing Architecture#
Note
SR-IOV and the manual placement of vGPUs on GPUs in equal-size mode are not supported on GPUs based on the NVIDIA Turing™ architecture.
GPU |
Mixed vGPU Configuration - Red Hat Enterprise Linux with KVM Releases |
|||
|---|---|---|---|---|
Frame Buffer Size (Mixed-Size Mode) |
Series |
NVIDIA vGPU |
GPU Pass Through |
|
Tesla T4 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
10.2, 10.1, 10.0, 9.8, 9.7, 9.6, 9.4, 8.10 |
|
|
Support for a Mixture of Time-Sliced vGPU Types on the Same GPU#
Red Hat Enterprise Linux with KVM supports a mixture of different types of time-sliced vGPUs on the same physical GPU. Any combination of A-series, B-series, and Q-series vGPUs with any amount of frame buffer can reside on the same physical GPU simultaneously. The total amount of frame buffer allocated to the vGPUs on a physical GPU must not exceed the amount of frame buffer that the physical GPU has.
For example, the following combinations of vGPUs can reside on the same physical GPU simultaneously:
A40-2B and A40-2Q
A40-2Q and A40-4Q
A40-2B and A40-4Q
By default, a GPU or GPU instance supports only vGPUs with the same amount of frame buffer and, therefore, is in equal-size mode. To support vGPUs with different amounts of frame buffer, the GPU or GPU instance must be put into mixed-size mode. In mixed-size mode, the maximum number of some types of vGPU allowed on a GPU or GPU instance is less than in equal-size mode. For more information, refer to Virtual GPU Software User Guide.
Switching the Mode of a GPU that Supports Multiple Display Modes#
Some GPUs support display-off and display-enabled modes but must be used in NVIDIA vGPU software deployments in display-off mode.
The GPUs listed in the following table support multiple display modes. As shown in the table, some GPUs are supplied from the factory in display-off mode, but other GPUs are supplied in a display-enabled mode.
GPU |
Mode as Supplied from the Factory |
|---|---|
NVIDIA A40 |
Display-off |
NVIDIA L40 |
Display-off |
NVIDIA L40S |
Display-off |
NVIDIA L20 |
Display-off |
NVIDIA L20 liquid cooled |
Display-off |
NVIDIA RTX 5000 Ada |
Display enabled |
NVIDIA RTX 6000 Ada |
Display enabled |
NVIDIA RTX A5000 |
Display enabled |
NVIDIA RTX A5500 |
Display enabled |
NVIDIA RTX A6000 |
Display enabled |
NVIDIA RTX PRO 6000 Blackwell Server Edition |
Display-off |
NVIDIA RTX PRO 6000 Blackwell Server Edition liquid cooled |
Display-off |
A GPU that is supplied from the factory in display-off mode, such as the NVIDIA A40 GPU, might be in a display-enabled mode if its mode has previously been changed.
To change the mode of a GPU that supports multiple display modes, use the displaymodeselector tool, which you can request from the NVIDIA Display Mode Selector Tool page on the NVIDIA Developer website.
Note
Only the GPUs listed in the table support the displaymodeselector tool. Other GPUs that support NVIDIA vGPU software do not support the displaymodeselector tool and, unless otherwise stated, do not require display mode switching.
Hypervisor Software Releases#
This release supports only the hypervisor software releases listed in the table.
Note
If a specific release, even an update release, is not listed, it’s not supported.
Software |
Releases Supported |
Notes |
|---|---|---|
Since 20.2: Red Hat Enterprise Linux with KVM |
10.2 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat Enterprise Linux with KVM |
10.1 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat Enterprise Linux with KVM |
10.0 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Since 20.2: Red Hat Enterprise Linux with KVM |
9.8 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat Enterprise Linux with KVM |
9.7 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat Enterprise Linux with KVM |
9.6 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat Enterprise Linux with KVM |
9.4 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat Enterprise Linux with KVM |
8.10 |
All NVIDIA GPUs that NVIDIA vGPU software supports are supported with vGPU and in pass-through mode. |
Red Hat OpenShift Virtualization (OSV) |
4.16 4.15 |
Supported on only the following GPUs with vGPU and in pass-through mode:
Other GPUs that NVIDIA vGPU software supports are not supported. |
Guest OS Support#
NVIDIA vGPU software supports several Windows releases and Linux distributions as a guest OS. The supported guest operating systems depend on the hypervisor software version.
Note
Use only a guest OS release that is listed as supported by NVIDIA vGPU software with your virtualization software. To be listed as supported, a guest OS release must be supported not only by NVIDIA vGPU software, but also by your virtualization software. NVIDIA cannot support guest OS releases that your virtualization software does not support.
NVIDIA vGPU software supports only 64-bit guest operating systems. No 32-bit guest operating systems are supported.
Windows Guest OS Support#
Note
Red Hat Enterprise Linux with KVM and RHV support Windows guest operating systems under specific Red Hat subscription programs. For details, refer to Certified Guest Operating Systems in Red Hat OpenStack Platform, Red Hat Virtualization, Red Hat OpenShift Virtualization and Red Hat Enterprise Linux with KVM.
NVIDIA vGPU software supports only the 64-bit Windows releases listed as a guest OS on Red Hat Enterprise Linux with KVM. The releases of Red Hat Enterprise Linux with KVM for which a Windows release is supported depend on whether NVIDIA vGPU or pass-through GPU is used.
Note
If a specific release, even an update release, is not listed, it’s not supported.
Windows Guest OS Support in Release 20.2#
Windows Server 2022
Windows 11 25H2 and all Windows 11 releases supported by Microsoft up to and including this release
Windows 10 2022 Update (22H2) and all Windows 10 releases supported by Microsoft up to and including this release
Note
The hardware-accelerated GPU scheduling feature introduced in Windows 10 May 2020 Update (2004) is not supported on GPUs based on the Maxwell architecture and is supported only in pass-through mode on GPUs based on later architectures.
Windows Guest OS Support in Release 20.1#
Windows Server 2022
Windows 11 25H2 and all Windows 11 releases supported by Microsoft up to and including this release
Windows 10 2022 Update (22H2) and all Windows 10 releases supported by Microsoft up to and including this release
Note
The hardware-accelerated GPU scheduling feature introduced in Windows 10 May 2020 Update (2004) is not supported on GPUs based on the Maxwell architecture and is supported only in pass-through mode on GPUs based on later architectures.
Windows Guest OS Support in Release 20.0#
Windows Server 2022
Windows 11 25H2 and all Windows 11 releases supported by Microsoft up to and including this release
Windows 10 2022 Update (22H2) and all Windows 10 releases supported by Microsoft up to and including this release
Note
The hardware-accelerated GPU scheduling feature introduced in Windows 10 May 2020 Update (2004) is not supported on GPUs based on the Maxwell architecture and is supported only in pass-through mode on GPUs based on later architectures.
Linux Guest OS Support#
NVIDIA vGPU software supports only the 64-bit Linux distributions listed as a guest OS on Red Hat Enterprise Linux with KVM. The releases of Red Hat Enterprise Linux with KVM for which a Linux release is supported depend on whether NVIDIA vGPU or pass-through GPU is used.
Note
If a specific release, even an update release, is not listed, it’s not supported.
Rocky Linux releases that are compatible with supported Red Hat Enterprise Linux releases are also supported as a guest OS.
Linux Guest OS Support in Release 20.2#
Guest OS |
Wayland Support |
X11 Support |
|---|---|---|
Deprecated: CentOS Linux 8 (2105) |
×× |
✓✓ |
Red Hat CoreOS 4.11 |
×× |
✓✓ |
Red Hat Enterprise Linux 10.2 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 10.1 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 10.0 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 9.8 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.7 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.6 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.4 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 8.10 |
×! |
✓! |
✓✓ Supported by NVIDIA vGPU software and enabled by default in the OS.
✓× Supported by NVIDIA vGPU software but disabled by default in the OS and must be enabled.
✓! Supported by NVIDIA vGPU software but requires the alternative display server protocol to be disabled in the OS.
×× Not supported by NVIDIA vGPU software or the OS.
×! Not supported by NVIDIA vGPU software and must be disabled in the OS.
Linux Guest OS Support in Release 20.1#
Guest OS |
Wayland Support |
X11 Support |
|---|---|---|
Deprecated: CentOS Linux 8 (2105) |
×× |
✓✓ |
Red Hat CoreOS 4.11 |
×× |
✓✓ |
Red Hat Enterprise Linux 10.1 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 10.0 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 9.7 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.6 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.4 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 8.10 |
×! |
✓! |
✓✓ Supported by NVIDIA vGPU software and enabled by default in the OS.
✓× Supported by NVIDIA vGPU software but disabled by default in the OS and must be enabled.
✓! Supported by NVIDIA vGPU software but requires the alternative display server protocol to be disabled in the OS.
×× Not supported by NVIDIA vGPU software or the OS.
×! Not supported by NVIDIA vGPU software and must be disabled in the OS.
Linux Guest OS Support in Release 20.0#
Guest OS |
Wayland Support |
X11 Support |
|---|---|---|
Deprecated: CentOS Linux 8 (2105) |
×× |
✓✓ |
Red Hat CoreOS 4.11 |
×× |
✓✓ |
Red Hat Enterprise Linux 10.1 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 10.0 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 and 9 releases |
✓✓ |
×× |
Red Hat Enterprise Linux 9.7 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.6 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 9.4 Not supported on Red Hat Enterprise Linux with KVM hypervisor 8 releases |
×! |
✓! |
Red Hat Enterprise Linux 8.10 |
×! |
✓! |
✓✓ Supported by NVIDIA vGPU software and enabled by default in the OS.
✓× Supported by NVIDIA vGPU software but disabled by default in the OS and must be enabled.
✓! Supported by NVIDIA vGPU software but requires the alternative display server protocol to be disabled in the OS.
×× Not supported by NVIDIA vGPU software or the OS.
×! Not supported by NVIDIA vGPU software and must be disabled in the OS.
NVIDIA CUDA Toolkit Version Support#
The releases in this release family of NVIDIA vGPU software support NVIDIA CUDA Toolkit 13.2.
To build a CUDA application, the system must have the NVIDIA CUDA Toolkit and the libraries required for linking. For details of the components of NVIDIA CUDA Toolkit, refer to NVIDIA CUDA Toolkit 13.2 Release Notes.
To run a CUDA application, the system must have a CUDA-enabled GPU and an NVIDIA display driver that is compatible with the NVIDIA CUDA Toolkit release that was used to build the application. If the application relies on dynamic linking for libraries, the system must also have the correct version of these libraries.
For more information about NVIDIA CUDA Toolkit, refer to CUDA Toolkit Documentation 13.2.
Note
If you are using NVIDIA vGPU software with CUDA on Linux, avoid conflicting installation methods by installing CUDA from a distribution-independent runfile package. Do not install CUDA from a distribution-specific RPM or Deb package.
To ensure that the NVIDIA vGPU software graphics driver is not overwritten when CUDA is installed, deselect the CUDA driver when selecting the CUDA components to install.
For more information, see NVIDIA CUDA Installation Guide for Linux.
Wayland Display Server Protocol Support#
The Wayland display server protocol is supported with only a subset of supported GPUs, guest OS releases, and display resolutions.
Supported GPUs#
All GPUs supported by NVIDIA vGPU software, except GPUs based on the NVIDIA Turing™ architecture.
Supported Guest OS Releases#
Only Red Hat Enterprise Linux 10 guest OS releases supported by NVIDIA vGPU software.
Supported Display Resolutions#
2560×1600
3840×2160
4096×2160
5120×2880
7680×4320
Limitations with Wayland Display Server Protocol Support#
The following use cases are not supported:
VDPAU
Vulkan video playback
Console VNC
Frame-rate limiting (FRL)
Known Issues with Wayland Display Server Protocol Support#
vGPU Migration Support#
vGPU Migration is supported on all supported GPUs, but only on a subset of supported Red Hat Enterprise Linux with KVM releases and guest operating systems.
Limitations with vGPU Migration Support#
For a host that is running Red Hat Enterprise Linux with KVM 9.4, the following migrations are not supported:
Migration between hosts that are running different versions of the NVIDIA Virtual GPU Manager driver, even within the same NVIDIA Virtual GPU Manager driver branch
Migration to or from a host that is running a version of Red Hat Enterprise Linux with KVM that uses of a vendor-specific VFIO framework
Unless explicitly stated otherwise, these restrictions do not apply to a host that is running Red Hat Enterprise Linux with KVM since 9.6. For example, the following migrations are supported:
Migration between hosts that are running different versions of the NVIDIA Virtual GPU Manager driver
Migration between a host that is running Red Hat Enterprise Linux with KVM 9.6 and a host that is running Red Hat Enterprise Linux with KVM 10.0
vGPU migration is disabled for a VM for which any of the following NVIDIA CUDA Toolkit features is enabled:
Unified memory
Debuggers
Profilers
Supported Hypervisor Software Releases#
Since Red Hat Enterprise Linux with KVM 9.4
Supported Guest OS Releases#
Windows and Linux.
Known Issues with vGPU Migration Support#
Use Case |
Affected GPUs |
Issue |
|---|---|---|
Migration of a VM configured with NVIDIA vGPU software release 20.1 to a host running release 20.0 |
All GPUs that support vGPU Migration |
|
Migration between hosts with different ECC memory configuration |
All GPUs that support vGPU Migration |
Migration of VMs configured with vGPU stops before the migration is complete |
Multiple vGPU Support#
To support applications and workloads that are compute or graphics intensive, multiple vGPUs can be added to a single VM. The assignment of more than one vGPU to a VM is supported only on a subset of vGPUs and hypervisor software releases.
vGPUs that Support Multiple vGPUs Assigned to a VM#
The supported vGPUs depend on the architecture of the GPU on which the vGPUs reside:
For GPUs based on the NVIDIA Volta architecture and later GPU architectures, all Q-series vGPUs are supported.
On GPUs that support the Multi-Instance GPU (MIG) feature, both time-sliced and MIG-backed vGPUs are supported.
For GPUs based on the NVIDIA Pascal™ graphics architecture, only Q-series vGPUs that are allocated all of the physical GPU’s frame buffer are supported.
For GPUs based on the NVIDIA Maxwell™ graphics architecture, only Q-series vGPUs that are allocated all of the physical GPU’s frame buffer are supported.
You can assign multiple vGPUs with differing amounts of frame buffer to a single VM, provided the board type and the series of all the vGPUs is the same. For example, you can assign an A40-48Q vGPU and an A40-16Q vGPU to the same VM. However, you cannot assign an A30-8Q vGPU and an A16-8Q vGPU to the same VM.
Multiple vGPU Support on the NVIDIA Blackwell Architecture#
Board |
vGPU |
|---|---|
NVIDIA RTX PRO 6000 Blackwell Server Edition |
All Q-series vGPUs |
Since 20.2: NVIDIA RTX PRO 5000 Blackwell Workstation Edition 72 GB |
All Q-series vGPUs |
NVIDIA RTX PRO 4500 Blackwell Server Edition |
All Q-series vGPUs |
Multiple vGPU Support on the NVIDIA Ada Lovelace Architecture#
Board |
vGPU |
|---|---|
NVIDIA L40S |
All Q-series vGPUs |
NVIDIA L40 |
All Q-series vGPUs |
NVIDIA L20 |
All Q-series vGPUs |
NVIDIA L4 |
All Q-series vGPUs |
NVIDIA L2 |
All Q-series vGPUs |
NVIDIA RTX 6000 Ada |
All Q-series vGPUs |
NVIDIA RTX 5880 Ada |
All Q-series vGPUs |
NVIDIA RTX 5000 Ada |
All Q-series vGPUs |
Multiple vGPU Support on the NVIDIA Ampere GPU Architecture#
Multiple vGPU Support on the NVIDIA Turing GPU Architecture#
Board |
vGPU |
|---|---|
Tesla T4 |
All Q-series vGPUs |
Maximum Number of vGPUs Supported per VM#
NVIDIA vGPU software supports up to a maximum of 16 vGPUs per VM.
Hypervisor Releases that Support Multiple vGPUs Assigned to a VM#
All hypervisor releases that support NVIDIA vGPU software are supported.
Peer-to-Peer CUDA Transfers over NVLink Support#
Peer-to-peer CUDA transfers enable device memory between vGPUs on different GPUs that are assigned to the same VM to be accessed from within the CUDA kernels. NVLink is a high-bandwidth interconnect that enables fast communication between such vGPUs. Peer-to-Peer CUDA transfers over NVLink are supported only on a subset of vGPUs, Red Hat Enterprise Linux with KVM releases, and guest OS releases.
vGPUs that Support Peer-to-Peer CUDA Transfers#
Only Q-series time-sliced vGPUs that are allocated all of the physical GPU’s frame buffer on physical GPUs that support NVLink are supported.
Peer-to-Peer CUDA Transfer Support on the NVIDIA Ampere GPU Architecture#
Board |
vGPU |
|---|---|
NVIDIA A40 |
A40-48Q |
NVIDIA RTX A6000 |
A6000-48Q |
NVIDIA RTX A5500 |
A5500-24Q |
NVIDIA RTX A5000 |
A5000-24Q |
Peer-to-Peer CUDA Transfer Support on the NVIDIA Turing GPU Architecture#
Board |
vGPU |
|---|---|
Quadro RTX 6000 |
RTX6000-24Q |
Quadro RTX 6000 passive |
RTX6000P-24Q |
Quadro RTX 8000 |
RTX8000-48Q |
Quadro RTX 8000 passive |
RTX8000P-48Q |
Peer-to-Peer CUDA Transfer Support on the NVIDIA Volta GPU Architecture#
Board |
vGPU |
|---|---|
Tesla V100 SXM2 32GB |
V100DX-32Q |
Tesla V100 SXM2 |
V100X-16Q |
Hypervisor Releases that Support Peer-to-Peer CUDA Transfers#
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.
Guest OS Releases that Support Peer-to-Peer CUDA Transfers#
Linux only. Peer-to-Peer CUDA transfers over NVLink are not supported on Windows.
Limitations on Support for Peer-to-Peer CUDA Transfers#
NVSwitch is not supported. Only direct connections are supported.
Only time-sliced vGPUs are supported. MIG-backed vGPUs are not supported.
PCIe is not supported.
SLI is not supported.
Unified Memory Support#
Unified memory is a single memory address space that is accessible from any CPU or GPU in a system. It creates a pool of managed memory that is shared between the CPU and GPU to provide a simple way to allocate and access data that can be used by code running on any CPU or GPU in the system. Unified memory is supported only on a subset of vGPUs and guest OS releases.
Note
Unified memory is disabled by default. If used, you must enable unified memory individually for each vGPU that requires it by setting a vGPU plugin parameter. NVIDIA CUDA Toolkit profilers are supported and can be enabled on a VM for which unified memory is enabled.
vGPUs that Support Unified Memory#
On GPUs that support the MIG feature and on which this feature is enabled, only Q-series MIG-backed vGPUs that occupy an entire GPU instance are supported. All other MIG-backed vGPUs are not supported.
On single-instance GPUs, only Q-series vGPUs that are allocated all of the physical GPU’s frame buffer on physical GPUs that support unified memory are supported.
Unified Memory Support on the NVIDIA Blackwell GPU Architecture#
Board |
vGPU |
|---|---|
NVIDIA RTX PRO 6000 Blackwell Server Edition |
NVIDIA RTX PRO 6000 Blackwell DC-96Q |
Since 20.2: NVIDIA RTX PRO 5000 Blackwell Workstation Edition 72 GB |
NVIDIA RTX PRO 5000 Blackwell WE-72Q |
NVIDIA RTX PRO 4500 Blackwell Server Edition |
NVIDIA RTX PRO 4500 Blackwell DC-32Q |
Unified Memory Support on the NVIDIA Ada Lovelace GPU Architecture#
Board |
vGPU |
|---|---|
NVIDIA L40 |
L40-48Q |
NVIDIA L40S |
L40S-48Q |
NVIDIA L20 |
L20-48Q |
NVIDIA L4 |
L4-24Q |
NVIDIA L2 |
L2-24Q |
NVIDIA RTX 6000 Ada |
RTX 6000 Ada-48Q |
NVIDIA RTX 5880 Ada |
RTX 5880 Ada-48Q |
NVIDIA RTX 5000 Ada |
RTX 5000 Ada-32Q |
Unified Memory Support on the NVIDIA Ampere GPU Architecture#
Board |
vGPU |
|---|---|
NVIDIA A40 |
A40-48Q |
NVIDIA A16 |
A16-16Q |
NVIDIA A10 |
A10-24Q |
NVIDIA A2 |
A2-16Q |
NVIDIA RTX A6000 |
A6000-48Q |
NVIDIA RTX A5500 |
A5500-24Q |
NVIDIA RTX A5000 |
A5000-24Q |
Guest OS Releases that Support Unified Memory#
Linux only. Unified memory is not supported on Windows.
Limitations on Support for Unified Memory#
Only time-sliced Q-series and C-series vGPUs that are allocated all of the physical GPU’s frame buffer on physical GPUs that support unified memory are supported. Fractional time-sliced vGPUs are not supported.
NVIDIA GPU Operator Support#
NVIDIA GPU Operator simplifies the deployment of NVIDIA vGPU software with software container platforms on immutable operating systems. An immutable operating system does not allow the installation of the NVIDIA vGPU software graphics driver directly on the operating system. NVIDIA GPU Operator is supported only on specific combinations of hypervisor software release, container platform, and guest OS release.
For more information, refer to Using NVIDIA vGPU in the NVIDIA GPU Operator documentation.
NVIDIA Deep Learning Super Sampling (DLSS) Support#
NVIDIA vGPU software supports NVIDIA DLSS on NVIDIA RTX Virtual Workstation.
Supported DLSS versions: 2.0. Version 1.0 is not supported.
Supported GPUs:
NVIDIA L40
NVIDIA L40S
NVIDIA L20
NVIDIA L20 liquid cooled
NVIDIA L4
NVIDIA L2
NVIDIA RTX 6000 Ada
NVIDIA RTX 5880 Ada
NVIDIA RTX 5000 Ada
NVIDIA A40
NVIDIA A16
NVIDIA A2
NVIDIA A10
NVIDIA RTX A6000
NVIDIA RTX A5500
NVIDIA RTX A5000
NVIDIA RTX PRO 6000 Blackwell Server Edition
NVIDIA RTX PRO 6000 Blackwell Server Edition liquid cooled
Since 20.2: NVIDIA RTX PRO 5000 Blackwell Workstation Edition 72 GB
NVIDIA RTX PRO 4500 Blackwell Server Edition
Tesla T4
Note
NVIDIA graphics driver components that DLSS requires are installed only if a supported GPU is detected during installation of the driver. Therefore, if the creation of VM templates includes driver installation, the template should be created from a VM that is configured with a supported GPU while the driver is being installed.
Supported applications: only applications that use nvngx_dlss.dll version 2.0.18 or newer