Platform Support
This documents provides an overview of the GPUs and system platform configurations supported.
GPUs
The following NVIDIA datacenter/enterprise GPUs are supported:
Product |
GPU Architecture |
|---|---|
Datacenter A-series Products |
|
NVIDIA A100 |
NVIDIA Ampere |
NVIDIA A100X |
NVIDIA Ampere |
NVIDIA A40 |
NVIDIA Ampere |
NVIDIA A30 |
NVIDIA Ampere |
NVIDIA A30X |
NVIDIA Ampere |
NVIDIA A16 |
NVIDIA Ampere |
NVIDIA A10 |
NVIDIA Ampere |
NVIDIA A2 |
NVIDIA Ampere |
Datacenter T-series Products |
|
NVIDIA T4 |
Turing |
Datacenter V-series Products |
|
NVIDIA V100 |
Volta |
Datacenter P-series Products |
|
NVIDIA Tesla P100 |
Pascal |
NVIDIA Tesla P40 |
Pascal |
NVIDIA Tesla P4 |
Pascal |
RTX-Series / T-Series Products |
|
NVIDIA RTX A6000 |
NVIDIA Ampere |
NVIDIA RTX A5000 |
NVIDIA Ampere |
NVIDIA RTX A4000 |
NVIDIA Ampere |
Quadro RTX 8000 |
Turing |
Quadro RTX 6000 |
Turing |
Quadro RTX 5000 |
Turing |
Quadro RTX 4000 |
Turing |
NVIDIA T1000 |
Turing |
NVIDIA T600 |
Turing |
NVIDIA T400 |
Turing |
The following NVIDIA server platforms are supported:
Product |
Architecture |
|---|---|
Datacenter A-series Products |
|
NVIDIA HGX A100 |
A100 and NVSwitch |
NVIDIA DGX A100 |
A100 and NVSwitch |
Note
The GPU Operator supports NVIDIA A100X/A30X running on the x86 host or on the DPU’s Arm processor.
Note
The GPU Operator supports DGX A100 with DGX OS 5.1+ and DGX A100 with OCP using RHCOS. For installation instructions, see here for DGX OS 5.1+ and here for OCP.
Note
The GPU Operator only supports platforms using discrete GPUs - Jetson or other embedded products with integrated GPUs are not supported.
Kubernetes Platforms
The following Kubernetes platforms are supported:
Kubernetes v1.21+
VMware vSphere with Tanzu
Red Hat OpenShift 4 using Red Hat Enterprise Linux CoreOS (RHCOS) and CRI-O container runtime. See the OpenShift guide for getting started.
Google Cloud Anthos. See the user guide for getting started.
Note
Technical Preview: Red Hat OpenShift 4.10 on ARM Server Base System Architecture (SBSA) systems. Raise issues on GitHub
Note
Note that the Kubernetes community supports only the last three minor releases as of v1.17. Older releases may be supported through enterprise distributions of Kubernetes such as Red Hat OpenShift. See the prerequisites for enabling monitoring in Kubernetes releases before v1.16.
The following table includes the support matrix of the GPU Operator releases and supported kubernetes platforms.
GPU Operator Release |
Kubernetes |
OpenShift |
Anthos |
|---|---|---|---|
1.10 |
v1.21+ |
4.9 and 4.10 |
Supported |
1.9 |
v1.19+ |
4.8 and 4.9 |
Supported |
1.8 |
v1.18+ |
4.7, 4.8 and 4.9 |
Supported |
1.7 |
v1.18+ |
4.5, 4.6 and 4.7 |
Supported |
1.6 |
v1.16+ |
4.5, 4.6 and 4.7 |
Supported |
1.5 |
v1.13+ |
4.4.29+, 4.5 and 4.6 |
Supported |
1.4 |
v1.13+ |
4.4.29+, 4.5 and 4.6 |
Supported |
1.3 |
v1.13+ |
4.4.29+, 4.5 and 4.6 |
Supported |
1.2 |
v1.13+ |
Not supported |
Supported |
1.1.7 |
v1.13+ |
4.1, 4.2, 4.3, and 4.4 |
Supported |
1.1 |
v1.13+ |
Not supported |
Not supported |
1.0 |
v1.13+ |
Not supported |
Not supported |
GPU Operator Release |
Kubernetes |
OpenShift |
Anthos |
|---|---|---|---|
1.10 |
v1.21+ |
4.9 and 4.10 |
Not Supported |
1.9 |
v1.19+ |
4.8 and 4.9 |
Not Supported |
1.8 |
v1.18+ |
4.7 and 4.8 |
Not Supported |
1.7 |
v1.18+ |
4.6, 4.7 and 4.8 |
Not Supported |
1.6 |
v1.16+ |
4.6 and 4.7 |
Not Supported |
1.5 |
v1.13+ |
4.6 |
Not Supported |
GPU Operator Release |
Kubernetes |
OpenShift |
Anthos |
vSphere with Tanzu |
Release |
|---|---|---|---|---|---|
1.10.1 |
v1.21+ |
4.9.9+ and 4.10 |
Not Supported |
Supported |
2.0 |
1.10.1 |
v1.21+ |
Not Supported |
Not Supported |
Supported |
1.2 |
1.9.1 |
v1.21+ |
Not Supported |
Not Supported |
Supported |
1.1 |
1.8.1 |
v1.21+ |
Not Supported |
Not Supported |
Not Supported |
1.0 |
Note
The GPU Operator versions are expressed as x.y.z or <major, minor, patch> and follows the semver terminology.
Only the most recent release of the GPU Operator is maintained through z patch updates. All prior releases of the GPU Operator are deprecated (and unsupported) when a new x.y version of the GPU Operator is released.
The product lifecycle and versioning are subject to change in the future.
Deployment Scenarios
The GPU Operator has been validated in the following scenarios:
Bare metal
GPU passthrough virtualization
NVIDIA vGPU
Linux distributions
The following Linux distributions are supported:
Ubuntu 18.04.z, 20.04.z LTS
DGX OS 5.1+
Red Hat Enterprise Linux CoreOS (RHCOS) for use with OpenShift 4.9 and 4.10
CentOS 7
The following Linux distributions are supported:
Ubuntu 20.04.z LTS
Red Hat Enterprise Linux CoreOS (RHCOS) for use with OpenShift 4.9 and 4.10
The following Linux distributions are supported:
Ubuntu 20.04.z LTS
Red Hat Enterprise Linux CoreOS (RHCOS) for use with OpenShift 4.9 and 4.10
In addition, the following container management tools are supported:
Helm v3
Docker CE 19.03+
containerd 1.4+
CRI-O with OpenShift 4 using Red Hat Enterprise Linux CoreOS (RHCOS)
Supported Platforms with NVIDIA AI Enterprise
The following platforms are supported. Refer to the NVIDIA AI Enterprise Documentation for more detailed information.
Ubuntu 20.04.z LTS bare metal
Red Hat OpenShift 4.9.9+ and 4.10 with RHCOS on bare metal
Red Hat OpenShift 4.9.9+ and 4.10 with RHCOS on VMware vSphere 7.0 Update 2+
VMware vSphere 7.0 Update 2+ with Ubuntu 20.04 guest operating systems
VMware vSphere with Tanzu (7.0 U3c) with Ubuntu 20.04 guest operating systems
Supported NVIDIA vGPU Products
NVIDIA vGPU 12.0+ with the following software products
NVIDIA Virtual Compute Server (C-Series)
NVIDIA RTX Virtual Workstation (vWS)
Supported Hypervisors with NVIDIA vGPU
The following Virtualization Platforms are supported. Refer to the NVIDIA vGPU Documentation for more detailed information.
VMware vSphere 7
Red Hat Enterprise Linux KVM
Red Hat Virtualization (RHV)
Note
The GPU Operator deploys the NVIDIA driver as a container. In this environment, running on desktop environments (e.g. workstations with GPUs and display) is not supported.
GPU Operator Component Matrix
Release |
NVIDIA Driver |
NVIDIA Driver Manager for K8s |
NVIDIA Container Toolkit |
NVIDIA K8s Device Plugin |
NVIDIA DCGM-Exporter |
Node Feature Discovery |
NVIDIA GPU Feature Discovery |
NVIDIA MIG Manager for K8s |
NVIDIA DCGM |
|---|---|---|---|---|---|---|---|---|---|
1.10.1 |
0.10.1 |
||||||||
1.10 |
0.10.1 |
||||||||
1.9.1 |
0.8.2 |
||||||||
1.9.0 |
0.8.2 |
||||||||
1.8.2 |
0.8.2 |
||||||||
1.8.1 |
0.8.2 |
||||||||
1.8.0 |
0.8.2 |
||||||||
1.7.1 |
N/A |
0.8.2 |
N/A |
||||||
1.7.0 |
N/A |
0.6.0 |
N/A |
||||||
1.6.2 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.6.1 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.6.0 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.5.2 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.5.1 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.5.0 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.4.0 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.3.0 |
N/A |
0.6.0 |
N/A |
N/A |
|||||
1.2.0 |
N/A |
0.6.0 |
N/A |
N/A |
N/A |
||||
1.1.0 |
N/A |
0.5.0 |
N/A |
N/A |
N/A |
Note
Driver version could be different with NVIDIA vGPU, as it depends on the driver version downloaded from the NVIDIA vGPU Software Portal.
The GPU Operator is supported on all the R450, R470 and R510 NVIDIA datacenter production drivers. For a list of supported datacenter drivers versions, visit this link.
GPUDirect RDMA
For more information on GPUDirect RDMA refer to this document.
The following Linux distributions are supported:
Ubuntu 20.04 LTS
RedHat OpenShift 4.10 using RHCOS
The following NVIDIA drivers are supported:
R470 datacenter drivers (470.57.02+)
R510 datacenter drivers (510.47.03+)
Note
For Red Hat OpenShift GPUDirect RDMA is only supported from 470.103.01+ when using R470 datacenter drivers.