Frameworks Support Matrix

Abstract

This support matrix is for NVIDIA® optimized frameworks. The matrix provides a single view into the supported software and specific versions that come packaged with the frameworks based on the container image.


Important:

Content that is included in <<>> brackets indicates new content from the previously published version.

Note:

The deep learning framework container packages follow a naming convention that is based on the year and month of the image release. For example, the 24.01 release of an image was released in January 2024.

24.xx container images

Table 1. Software stack packaged with the 24.xx container images
Container Image 24.02 24.01
DGX
DGX System
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
Operating System
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 91

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 92

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
System Requirements
NVIDIA Driver

Release 24.02 is based on CUDA 12.3.2, which requires NVIDIA Driver release 545 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 470.57 (or later R470), 525.85 (or later R525), 535.86 (or later R535), or 545.23 (or later R545).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, R450, R460, R510, and R520 drivers, which are not forward-compatible with CUDA 12.3. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.

Release 24.01 is based on CUDA 12.3.2, which requires NVIDIA Driver release 545 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 470.57 (or later R470), 525.85 (or later R525), 535.86 (or later R535), or 545.23 (or later R545).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, R450, R460, R510, and R520 drivers, which are not forward-compatible with CUDA 12.3. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.

GPU Model
Base Container Image (included in all containers)
Container OS Ubuntu 22.04 Ubuntu 22.04
CUDA NVIDIA CUDA 12.3.2 NVIDIA CUDA 12.3.2
cuBLAS NVIDIA cuBLAS 12.3.4.1 NVIDIA cuBLAS 12.3.4.1
cuDNN <<9.0.0.306>> 8.9.7.29
cuTENSOR 2.0 2.0
DALI <<1.34>> 1.33
NCCL 2.19.4 2.19.4
TensorRT <<TensorRT 8.6.3>> TensorRT 8.6.1.6
rdma-core 39.0 39.0
NVIDIA HPC-X 2.16rc4 with 2.16rc4 with
GDRcopy 2.3 2.3
Nsight Compute 2023.3.1.1 2023.3.1.1
Nsight Systems 2023.4.1.97 2023.4.1.97
NVIDIA Optimized Frameworks
DGL - 1.2 including:
- Multi arch support: x86, Arm SBSA
- Docker image size: 24.8 GB
JAX - -
  -
NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet 1.9.1 including: 1.9.1 including:
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 12.0 GB Docker image size: 12.1 GB
PaddlePaddle 2.5.2 including: 2.5.2 including:
Multi arch support: x86 only Multi arch support: x86 only
Docker image size: 8.94 GB Docker image size: 9.01 GB
PyTorch 2.3.0a0+ebedce2including 2.2.0a0+81ea7a4including
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 22.2 GB Docker image size: 22.0 GB
TensorFlow 2.15.0 including 2.14.0 including
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 14.4 GB Docker image size: 14.4 GB
TensorRT TensorRT 8.6.3 TensorRT 8.6.1.6
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 7.05 GB Docker image size: 7.46 GB
Triton Inference Server In addition to the hardware and software listed above, Triton Inference Server also supports:

2.43 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.41 including

Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 13.8 GB Docker image size: 14.7 GB
TensorFlow For Jetson 2.15.0  
PyTorch for Jetson 2.3.0a0+ebedce2  
Triton for Jetson 2.43.0  


Important:

Content that is included in <<>> brackets indicates new content from the previously published version.

Note:

The deep learning framework container packages follow a naming convention that is based on the year and month of the image release. For example, the 23.01 release of an image was released in January 2023.

23.xx container images

Table 2. Software stack packaged with the 23.xx container images
Container Image 23.12 23.11 23.10 23.09 23.08 23.07 23.06 23.05 23.04 23.03 23.02 23.01
DGX
DGX System
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
Operating System
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 93

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 94

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 91

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 91

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 91

  • EL91

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 91

  • EL91

Red Hat Enterprise Linux 8 / CentOS 81 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 9 / CentOS 91

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+5 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+
  • 6.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 71

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
System Requirements
NVIDIA Driver

Release 23.12 is based on CUDA 12.3.2, which requires NVIDIA Driver release 545 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), or 525.85 (or later R525), 535.86 (or later R535), or 545.23 (or later R545).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, R460, and R520 drivers, which are not forward-compatible with CUDA 12.3. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.

Release 23.11 is based on CUDA 12.3.0, which requires NVIDIA Driver release 545 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), or 525.85 (or later R525), 535.86 (or later R535), or 545.23 (or later R545).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, R460, and R520 drivers, which are not forward-compatible with CUDA 12.3. For a complete list of supported drivers, see the CUDA Application Compatibility topic. For more information, see CUDA Compatibility and Upgrades.

Release 23.10 is based on CUDA 12.2.2, which requires NVIDIA Driver release 535 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), or 525.85 (or later R525), or 535.86 (or later R535).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.2. 6

Release 23.09 is based on CUDA 12.2.1, which requires NVIDIA Driver release 535 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), or 525.85 (or later R525), or 535.86 (or later R535).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.2. 3

Release 23.08 is based on CUDA 12.2.1, which requires NVIDIA Driver release 535 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), or 525.85 (or later R525), or 535.86 (or later R535).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.2. 3

Release 23.07 is based on CUDA 12.1.1, which requires NVIDIA Driver release 530 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), 525.85 (or later R525), or 530.30 (or later R530).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.1. 3

Release 23.06 is based on CUDA 12.1.1, which requires NVIDIA Driver release 530 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), 525.85 (or later R525), or 530.30 (or later R530).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.1. 3

Release 23.05 is based on CUDA 12.1.1, which requires NVIDIA Driver release 530 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), 525.85 (or later R525), or 530.30 (or later R530).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.1. 3

Release 23.04 is based on CUDA 12.1.0, which requires NVIDIA Driver release 530 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), 525.85 (or later R525), or 530.30 (or later R530).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.0. 3

Release 23.03 is based on CUDA 12.1.0, which requires NVIDIA Driver release 530 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), 525.85 (or later R525), or 530.30 (or later R530).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.0. 3

Release 23.02 is based on CUDA 12.0.1, which requires NVIDIA Driver release 525 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), or 525.85 (or later R525).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.0. 3

Release 23.01 is based on CUDA 12.0.1, which requires NVIDIA Driver release 525 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), 515.65 (or later R515), or 525.85 (or later R525).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 12.0. 3

GPU Model
Base Container Image (included in all containers)
Container OS Ubuntu 22.04 Ubuntu 22.04 Ubuntu 22.04 Ubuntu 22.04 Ubuntu 22.04 Ubuntu 22.04 Ubuntu 22.04 Ubuntu 22.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04
CUDA NVIDIA CUDA 12.3.2 NVIDIA CUDA 12.3.0 NVIDIA CUDA 12.2.2 NVIDIA CUDA 12.2.1 NVIDIA CUDA 12.2.1 NVIDIA CUDA 12.1.1 NVIDIA CUDA 12.1.1 NVIDIA CUDA 12.1.1 NVIDIA CUDA 12.1.0 NVIDIA CUDA 12.1.0 NVIDIA CUDA 12.0.1 NVIDIA CUDA 12.0.1
cuBLAS NVIDIA cuBLAS 12.3.4.1 NVIDIA cuBLAS 12.3.2.1 NVIDIA cuBLAS 12.2.5.6 NVIDIA cuBLAS 12.2.5.6 NVIDIA cuBLAS 12.2.5.1 NVIDIA cuBLAS 12.1.3.1 NVIDIA cuBLAS 12.1.3.1 NVIDIA cuBLAS 12.1.3.1 NVIDIA cuBLAS 12.1.3 cuBLAS from CUDA 12.1.0 12.0.2 from CUDA 12.0.2 from CUDA
cuDNN 8.9.7.29 8.9.6.50 8.9.5 8.9.5 8.9.4 8.9.3 8.9.2 8.9.1.23 8.9.0 8.8.1.3 8.7.0 8.7.0
cuTENSOR 1.7.0.1 1.7.0.1 1.7.0.1 1.7.0.1 1.7.0.1 1.7.0.1 1.7.0.1 1.7.0.1 1.7.0 1.6.2.3 1.6.2.3 1.6.2.3
DALI 1.32.0 1.31.0 1.30.0 1.29.0 1.28.0 1.27.0 1.26.0 1.25.0 1.24.0 1.23.0 1.22.0 1.21.0
NCCL 2.19.3 2.19.3 2.19.3 2.18.5 2.18.3 2.18.3 2.18.1 2.18.1 2.17.1 2.17.1 2.16.5 2.16.5
TensorRT TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.2 TensorRT 8.6.1 TensorRT 8.5.3 TensorRT 8.5.3 TensorRT 8.5.2.2
rdma-core 39.0 39.0 39.0 39.0 39.0 39.0 39.0 36.0 36.0 36.0 36.0 36.0
NVIDIA HPC-X 2.16 with 2.16 with 2.16 with 2.16 with 2.15 with 2.15 with 2.15 with 2.14 with 2.13 with 2.13 with 2.13 with 2.13 with
GDRcopy 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3
Nsight Compute 2023.3.1.1 2023.3.0.12 2023.2.1.3 2023.2.1.3 2023.2.1.3 2023.1.1.4 2023.1.1.4 2023.1.1.4 2023.1.0.15 2023.1.0.15 2022.4.1.6 2022.4.1.6
Nsight Systems 2023.4.1 2023.3.1.92 2023.3.1.92 2023.3.1.92 2023.2.3.1001 2023.2.3.1001 2023.2.3.1001 2023.2 2023.1.1.127 2023.1.1.127 2022.5.1 2022.5.1
NVIDIA Optimized Frameworks
DGL - 1.1.1 including: - 1.1.1 including: - 1.1.1 including: - - - - - -
      Multi arch support: x86, Arm SBSA   Multi arch support: x86, Arm SBSA            
      Docker image size: 23.4 GB   Docker image size: 20.8 GB            
JAX - - -              
    Multi arch support: x86 only   Multi arch support: x86 only              
Kaldi -
  Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only
  Docker image size: 9.13 GB Docker image size: 9.1 GB Docker image size: 9.16 GB Docker image size: 9.36 GB Docker image size: 9.14 GB Docker image size: 9.29 GB Docker image size: 9.19 GB Docker image size: 10.9 GB Docker image size: 11.1 GB Docker image size: 11.8 GB Docker image size: 11.1 GB
NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including:
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 12.1 GB Docker image size: 12.1 GB Docker image size: 12 GB Docker image size: 12.1 GB Docker image size: 12.1 GB Docker image size: 12.1 GB Docker image size: 12.0 GB Docker image size: 12.1 GB Docker image size: 13.1 GB Docker image size: 13.2 GB Docker image size: 13.9 GB Docker image size: 13.1 GB
PaddlePaddle 2.5.2 including: 2.5.2 including: 2.5.1 including: 2.5.0 including: 2.5.0 including: 2.4.1 including: 2.4.1 including: No 23.05 release. 2.4.1 including: 2.4.1 including: 2.4.0 including: 2.3.2 including:
Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only - Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only
Docker image size: 8.98 GB Docker image size: 8.98 GB Docker image size: 8.94 GB Docker image size: 8.99 GB Docker image size: 9.02 GB Docker image size: 8.58 GB Docker image size: 8.59 GB - Docker image size: 9.47 GB Docker image size: 9.74 GB Docker image size: 10.5 GB Docker image size: 9.41 GB
PyTorch 2.2.0a0+81ea7a48including 2.2.0a0+6a974be including 2.1.0a0+32f93b1 including 2.1.0a0+32f93b1 including 2.1.0a0+29c30b1 including 2.1.0a0+b5021ba including 2.1.0a0+4136153 including 2.0.0 including 2.1.0a0+fe05266f including 2.0.0a0+1767026 including 1.14.0a0+410ce96 including 1.14.0a0+410ce96 including
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 21.9 GB Docker image size: 21.9 GB Docker image size: 22.1 GB Docker image size: 22.0 GB Docker image size: 20.6 GB Docker image size: 19.8 GB Docker image size: 19.7 GB Docker image size: 22 GB Docker image size: 20.4 GB Docker image size: 20.4 GB Docker image size: 20.5 GB Docker image size: 19.7 GB
TensorFlow 2.14.0 including 2.13.0 including 2.13.0 including 2.13.0 including 2.13.0 including 2.12.0 including 2.12.0 including 2.12.0 including 2.12.0 including 2.11.0 including 1.15.5 including 2.11.0 including 1.15.5 including 2.11.0 including 1.15.5 including
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 14.3 GB Docker image size: 14.1 GB Docker image size: 14.2 GB Docker image size: 14.2 GB Docker image size: 14.2 GB Docker image size: 13.9 GB Docker image size: 14.3 GB Docker image size: 14.2 GB Docker image size: 15.4 GB Docker image size: 15.9 GB Docker image size: 16.3 GB Docker image size: 16.6 GB Docker image size: 17.0 GB Docker image size: 15.9 GB Docker image size: 16.2 GB
TensorRT TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.6 TensorRT 8.6.1.2 TensorRT 8.6.1 TensorRT 8.5.3 TensorRT 8.5.3 TensorRT 8.5.2.2
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 7.45 GB Docker image size: 7.45 GB Docker image size: 7.41 GB Docker image size: 7.47 GB Docker image size: 7.5 GB Docker image size: 7.45 GB Docker image size: 7.45 GB Docker image size: 7.5 GB Docker image size: 8.05 GB Docker image size: 8.32 GB Docker image size: 9.03 GB Docker image size: 8.3 GB
Triton Inference Server In addition to the hardware and software listed above, Triton Inference Server also supports:

2.41 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.40 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.39 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.38 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.37 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.36 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.35 including

In addition to the hardware and software listed above, Triton Inference Server also supports:

2.34 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.33 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.32 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.31 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.30 including

Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA
Docker image size: 14.7 GB Docker image size: 14.3 GB Docker image size: 12.6 GB Docker image size: 12.6 GB Docker image size: 12.4 GB Docker image size: 12.3 GB Docker image size: 12.3 GB Docker image size: 12.5 GB Docker image size: 13 GB Docker image size: 14.7 GB Docker image size: 15.3 GB Docker image size: 15.3 GB
TensorFlow For Jetson TensorFlow 2.14.0 for Jetson           TensorFlow 2.12.0 for Jetson TensorFlow 2.12.0 for Jetson TensorFlow 2.12.0 for Jetson TensorFlow 1.15.5 and 2.10.1 for Jetson TensorFlow 1.15.5 and 2.10.1 for Jetson TensorFlow 1.15.5 and 2.10.1 for Jetson
PyTorch for Jetson             2.1.0a0+4136153 for Jetson 2.0.0 for Jetson 2.1.0a0+fe05266f for Jetson PyTorch 2.0.0a0+1767026 for Jetson PyTorch 1.14.0a0+44dac51 for Jetson PyTorch 1.14.0a0+44dac51 for Jetson
TensorFlow Wheel for x86           - - - - TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86
Triton for Jetson           Triton Inference Server 2.36.0 for Jetson Triton Inference Server 2.35.0 for Jetson Triton Inference Server 2.34.0 for Jetson Triton Inference Server 2.33.0 for Jetson Triton Inference Server 2.32.0 for Jetson Triton Inference Server 2.31.0 for Jetson Triton Inference Server 2.30.0 for Jetson


Important:

Content that is included in <<>> brackets indicates new content from the previously published version.

Note:

The deep learning framework container packages follow a naming convention that is based on the year and month of the image release. For example, the 22.03 release of an image was released in March 2022.

22.xx container images

Table 3. Software stack packaged with the 22.xx container images
Container Image 22.12 22.11 22.10 22.09 22.08 22.07 22.06 22.05 22.04 22.03 22.02 22.01
DGX
DGX System
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX H100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • NVIDIA DGX-1™
  • DGX-2™
  • NVIDIA DGX™ A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • NVIDIA DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
Operating System
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0+

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0
  • 5.1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
System Requirements
NVIDIA Driver

Release 22.12 is based on CUDA 11.8.0, which requires NVIDIA Driver release 520 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), or 515.65 (or later R515).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.8. 3

Release 22.11 is based on CUDA 11.8.0, which requires NVIDIA Driver release 520 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), or 515.65 (or later R515).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.8. 3

Release 22.10 is based on CUDA 11.8.0, which requires NVIDIA Driver release 520 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), or 515.65 (or later R515).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.8. 3

Release 22.09 is based on CUDA 11.8.0, which requires NVIDIA Driver release 520 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), 510.47 (or later R510), or 515.65 (or later R515).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.8. 3

Release 22.08 is based on CUDA 11.7.1, which requires NVIDIA Driver release 515 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), or 510.47 (or later R510).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.8. 3

Release 22.07 is based on CUDA 11.7 Update 1 Preview, which requires NVIDIA Driver release 515 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), or 510.47 (or later R510).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.7. 3

Release 22.06 is based on CUDA 11.7 Update 1 Preview, which requires NVIDIA Driver release 515 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), or 510.47 (or later R510).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.7. 3

Release 22.05 is based on CUDA 11.7, which requires NVIDIA Driver release 515 or later. However, if you are running on a data center GPU (for example, T4 or any other data center GPU), you can use NVIDIA driver release 450.51 (or later R450), 470.57 (or later R470), or 510.47 (or later R510).

The CUDA driver's compatibility package only supports particular drivers. Thus, users should upgrade from all R418, R440, and R460 drivers, which are not forward-compatible with CUDA 11.7. 3

Release 22.04 is based on NVIDIA CUDA® 11.6.2, which requires NVIDIA Driver release 510 or later. However, if you are running on a Data Center GPU (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), 460.27 (or later R460), or 470.57 (or later R470). The CUDA driver's compatibility package only supports particular drivers. 3 Release 22.03 is based on NVIDIA CUDA® 11.6.1, which requires NVIDIA Driver release 510 or later. However, if you are running on a Data Center GPU (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), 460.27 (or later R460), or 470.57 (or later R470). The CUDA driver's compatibility package only supports particular drivers. 3 Release 22.02 is based on NVIDIA CUDA 11.6.0, which requires NVIDIA Driver release 510 or later. However, if you are running on a Data Center GPU (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), 460.27 (or later R460), or 470.57 (or later R470). The CUDA driver's compatibility package only supports particular drivers. 3 Release 22.01 is based on NVIDIA CUDA 11.6.0, which requires NVIDIA Driver release 510 or later. However, if you are running on a Data Center GPU (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), 460.27 (or later R460), or 470.57 (or later R470). The CUDA driver's compatibility package only supports particular drivers. 3
GPU Model
Base Container Image (included in all containers)
Container OS Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04
CUDA NVIDIA CUDA 11.8.0 NVIDIA CUDA 11.8.0 NVIDIA CUDA 11.8.0 NVIDIA CUDA 11.8.0 NVIDIA CUDA 11.7 Update 1 NVIDIA CUDA 11.7 Update 1 Preview NVIDIA CUDA 11.7 Update 1 Preview NVIDIA CUDA 11.7.0 NVIDIA CUDA 11.6.2 NVIDIA CUDA 11.6.1 NVIDIA CUDA 11.6.0 NVIDIA CUDA 11.6.0
cuBLAS 11.11.3.6 11.11.3.6 11.11.3.6 11.11.3.6 11.10.3.66 11.10.3.66 11.10.3.66 11.10.1.25 11.9.3.115 11.8.1.74 11.8.1.74 11.8.1.74
cuDNN 8.7.0 GA 8.7.0.80 8.6.0.163 8.6.0.163 8.5.0.96 8.4.1 8.4.1 8.4.0.27 8.4.0.27 8.3.3.40 8.3.2.44 8.3.2.44
cuTENSOR 1.6.1.5 1.6.1.5 1.6.1.5 1.6.1.5 1.6.0.2 1.5.0.3 1.5.0.3 1.5.0.3 1.5.0.3 1.5.0.1 1.4 1.4
DALI 1.20.0 1.18.0 1.18.0 1.17.0 1.16.0 1.15.0 1.14.0 1.13.0 1.12.0 1.11.1 1.10.0 1.9.0
NCCL 2.15.5 2.15.5 2.15.5 2.15.1 2.12.12 2.12.12 2.12.12 2.12.10 2.12.10 2.12.9 2.11.4 2.11.4
TensorRT TensorRT 8.5.1 TensorRT 8.5.1 TensorRT 8.5.0.12 TensorRT 8.5.0.12 TensorRT 8.4.2.4 TensorRT 8.4.1 TensorRT 8.2.5 TensorRT 8.2.5 TensorRT 8.2.4.2 TensorRT 8.2.3 TensorRT 8.2.3 TensorRT 8.2.2
rdma-core 36.0 36.0 36.0 36.0 36.0 36.0 36.0 36.0 36.0 36.0 36.0 36.0
NVIDIA HPC-X 2.13 with 2.12.2tp1 with 2.12.2tp1 with 2.12.1a0 with 2.10 with 2.10 with 2.10 with 2.10 with 2.10 with 2.10 with 2.10 with 2.10 with
GDRcopy 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3 2.3
Nsight Systems 2022.4.2.1 2022.4.2.1 2022.4.2.1 2022.4.1 2022.1.3.18 2022.1.3.3 2022.1.3.3 2022.1.3.3 2022.2.1.31-5fe97ab 2021.5.2.53 2021.5.2.53 2021.5.2.53
NVIDIA Optimized Frameworks
Kaldi
Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only
Docker image size: 10.4 GB Docker image size: 10.3 GB Docker image size: 10.3 GB Docker image size: 10.3 GB Docker image size: 8.89 GB Docker image size: 9.07 GB Docker image size: 9 GB Docker image size: 9.11 GB Docker image size: 9.01 GB Docker image size: 9 GB Docker image size: 9 GB Docker image size: 8.96 GB
NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.1 including: 1.9.0.rc6 including: 1.9.0.rc6 including: 1.9.0.rc6 including: Release paused Release paused
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA    
Docker image size: 12 GB Docker image size: 11.7 GB Docker image size: 11.7 GB Docker image size: 11.7 GB Docker image size: 9.97 GB Docker image size: 10.2 GB Docker image size: 10.1 GB Docker image size: 10.7 GB Docker image size: 10.6 GB Docker image size: 11.0 GB    
PyTorch 1.14.0a0+410ce96 including 1.13.0a0+936e930 including 1.13.0a0+d0d6b1f including 1.13.0a0+d0d6b1f including 1.13.0a0+d321be6 including 1.13.0a0+08820cb including 1.13.0a0+340c412 including 1.12.0a0+8a1a93a including 1.12.0a0+bd13bc6 including 1.12.0a0+2c916ef including 1.11.0a0+17540c5c including 1.11.0a0+bfe5ad28 including
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)
Docker image size: 18.3 GB Docker image size: 17.3 GB Docker image size: 16.9 GB Docker image size: 16.8 GB Docker image size: 14.6 GB Docker image size: 14.8 GB Docker image size: 14.6 GB Docker image size: 14.6 GB Docker image size: 14.1 GB Docker image size: 14.6 GB Docker image size: 14.4 GB Docker image size: 14.8 GB
TensorFlow 2.10.1 including 1.15.5 including 2.10.0 including 1.15.5 including 2.10.0 including 1.15.5 including 2.9.1 including 1.15.5 including 2.9.1 including 1.15.5 including 2.9.1 including 1.15.5 including 2.9.1 including 1.15.5 including 2.8.0 including 1.15.5 including 2.8.0 including 1.15.5 including 2.8.0 including 1.15.5 including 2.7.0 including 1.15.5 including 2.7.0 including 1.15.5 including
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)
Docker image size: 14.3 GB Docker image size: 14.8 GB Docker image size: 14.4 GB Docker image size: 15.0 GB Docker image size: 14.4 GB Docker image size: 14.9 GB Docker image size: 14.1 GB Docker image size: 14.9 GB Docker image size: 12 GB Docker image size: 12.8 GB Docker image size: 12.2 GB Docker image size: 13.0 GB Docker image size: 12.2 GB Docker image size: 14.4 GB Docker image size: 12.2 GB Docker image size: 14.4 GB Docker image size: 13.1 GB Docker image size: 14.4 GB Docker image size: 13.6 GB Docker image size: 14.9 GB Docker image size: 13.1 GB Docker image size: 14.5 GB Docker image size: 13.1 GB Docker image size: 15.1 GB
TensorRT TensorRT 8.5.1 TensorRT 8.5.1 TensorRT 8.5.0.12 TensorRT 8.5.0.12 including: TensorRT 8.4.2.4 including: TensorRT 8.4.1 including: TensorRT 8.2.5 including: TensorRT 8.2.5 including: TensorRT 8.2.4.2 including: TensorRT 8.2.3 including: TensorRT 8.2.2 including: TensorRT 8.2.2 including:
Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)
Docker image size: 7.61 GB Docker image size: 7.25 GB Docker image size: 7.51 GB Docker image size: 7.49 GB Docker image size: 6.09 GB Docker image size: 6.27 GB Docker image size: 6.21 GB Docker image size: 6.33 GB Docker image size: 6.21 GB Docker image size: 6.21 GB Docker image size: 6.21 GB Docker image size: 6.17 GB
Triton Inference Server In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.29.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.28.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.27.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.26.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.25.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.24.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.23.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.22.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.21.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.20.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.19.0 including

In addition to the hardware and software listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.18.0 including

Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)
Docker image size: 14 GB Docker image size: 13.8 GB Docker image size: 13.4 GB Docker image size: 13.7 GB Docker image size: 11.7 GB Docker image size: 11.9 GB Docker image size: 11 GB Docker image size: 11 GB Docker image size: 11.4 GB Docker image size: 12.1 GB Docker image size: 12.3 GB Docker image size: 12.4 GB

PaddlePaddle

2.3.2 including: 2.3.2 including: 2.3.2 including: 2.3.0 including: 2.3.0 including: 2.3.0 including: 2.2.2 including: 2.2.2 including:        
Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only        
Docker image size: 8.72 GB Docker image size: 8.48 GB Docker image size: 8.46 GB Docker image size: 8.44 GB Docker image size: 8.43 GB Docker image size: 8.43 GB Docker image size: 7.98 GB Docker image size: 8.09 GB        
TensorFlow For Jetson TensorFlow 1.15.5 and 2.10.1 for Jetson TensorFlow 1.15.5 and 2.10.0 for Jetson TensorFlow 1.15.5 and 2.10.0 for Jetson TensorFlow 1.15.5 and 2.9.1 for Jetson This release was skipped. TensorFlow 1.15.5 and 2.9.1 for Jetson TensorFlow 1.15.5 and 2.9.1 for Jetson TensorFlow 1.15.5 and 2.8.0 for Jetson TensorFlow 1.15.5 and 2.8.0 for Jetson TensorFlow 1.15.5 and 2.8.0 for Jetson TensorFlow 1.15.5 and 2.7.0 for Jetson TensorFlow 1.15.5 and 2.7.0 for Jetson
PyTorch for Jetson PyTorch 1.14.0a0+410ce96 for Jetson PyTorch 1.13.0a0+936e930 for Jetson PyTorch 1.13.0a0+d0d6b1f for Jetson PyTorch 1.13.0a0+d0d6b1f for Jetson This release was skipped. PyTorch 1.13.0a0+08820cb for Jetson PyTorch 1.13.0a0+340c412 for Jetson PyTorch 1.12.0a0+8a1a93a for Jetson PyTorch 1.12.0a0+84d1cb9 for Jetson PyTorch 1.12.0a0+2c916ef for Jetson This release was skipped. PyTorch 1.11.0a0+bfe5ad28 for Jetson
TensorFlow Wheel for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86 TensorFlow 1.15.5 for x86
Triton for Jetson Triton Inference Server 2.27.0 for Jetson Triton Inference Server 2.27.0 for Jetson Triton Inference Server 2.27.0 for Jetson Triton Inference Server 2.26.0 for Jetson Triton Inference Server 2.24.0 for Jetson Triton Inference Server 2.24.0 for Jetson Triton Inference Server 2.23.0 for Jetson Triton Inference Server 2.22.0 for Jetson Triton Inference Server 2.21.0 for Jetson Triton Inference Server 2.20.0 for Jetson Triton Inference Server 2.19.0 for Jetson Triton Inference Server 2.18.0 for Jetson


Important:

Content that is included in <<>> brackets indicates new content from the previously published version.

Note:

The deep learning framework container packages follow a naming convention that is based on the year and month of the image release. For example, the 21.02 release of an image was released in February 2021.

21.xx container images

Table 4. Software stack packaged with the 21.xx container images
Container Image 21.12 21.11 21.10 21.09 21.08 21.07 21.06 21.05 21.04 21.03 21.02
DGX
DGX System
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX Station A100
Operating System DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0
  • 5.1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0
  • 5.1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0
  • 5.1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0
  • 5.1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1(4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1

Red Hat Enterprise Linux 8 / CentOS 82 (All DGX systems except DGX Station A100)

  • EL8-20.11+1
NVIDIA Driver Release 21.12 is based on NVIDIA CUDA 11.5.0, which requires NVIDIA Driver release 495 or later. However, if you are running on a Data Center GPU (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), 460.27 (or later R460), or 470.57 (or later R470). The CUDA driver's compatibility package only supports particular drivers. 3 Release 21.11 is based on NVIDIA CUDA 11.5.0, which requires NVIDIA Driver release 495 or later. However, if you are running on a Data Center GPU (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), 460.27 (or later R460), or 470.57 (or later R470). The CUDA driver's compatibility package only supports particular drivers. 3 Release 21.10 is based on NVIDIA CUDA 11.4.2 with cuBLAS 11.6.5.2, which requires NVIDIA Driver release 470 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.09 is based on NVIDIA CUDA 11.4.2, which requires NVIDIA Driver release 470 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.08 is based on NVIDIA CUDA 11.4.1, which requires NVIDIA Driver release 470 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.07 is based on NVIDIA CUDA 11.4.0, which requires NVIDIA Driver release 470 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.06 is based on NVIDIA CUDA 11.3.1, which requires NVIDIA Driver release 465.19.01 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.05 is based on NVIDIA CUDA 11.3.0, which requires NVIDIA Driver release 465.19.01 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers.3

Release 21.04 is based on NVIDIA CUDA 11.3.0, which requires NVIDIA Driver release 465.19.01 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51 (or later R450), or 460.27 (or later R460).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.03 is based on NVIDIA CUDA 11.2.1, which requires NVIDIA Driver release 460.32.03 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51(or later R450).

The CUDA driver's compatibility package only supports particular drivers. 3

Release 21.02 is based on NVIDIA CUDA 11.2.0, which requires NVIDIA Driver release 460.27.04 or later. However, if you are running on Data Center GPUs (formerly Tesla), for example, T4, you may use NVIDIA driver release 418.40 (or later R418), 440.33 (or later R440), 450.51(or later R450).

The CUDA driver's compatibility package only supports particular drivers.3

GPU Model
Base Container Image
Container OS Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04 Ubuntu 20.04
CUDA NVIDIA CUDA 11.5.0 NVIDIA CUDA 11.5.0 NVIDIA CUDA 11.4.2 with cuBLAS 11.6.5.2 11.4.2 11.4.1 11.4.0 11.3.1 11.3.0 11.3.0 11.2.1 11.2.0
cuBLAS 11.7.3.1 11.7.3.1 11.6.1.51 11.6.1.51 11.5.4 11.5.2.43 11.5.1.109 11.5.1.101 11.5.1.101 11.4.1.1026 11.3.1.68
cuDNN 8.3.1.22 8.3.0.96 8.2.4.15 8.2.4.15 8.2.2.26 8.2.2.26 8.2.1 8.2.0.51 8.2.0.41 8.1.1 8.1.0.77
NCCL 2.11.4 2.11.4 2.11.4 2.11.4 2.10.3 2.10.3 2.9.9 2.9.8 2.9.6 2.8.4 2.8.4
TensorRT TensorRT 8.2.1.8 TensorRT 8.0.3.4

TensorRT 8.0.3.4

               
NVIDIA Optimized Frameworks
Kaldi
Multi arch support: x86 only Multi arch support: x86 only Multi arch support: x86 only                
Docker image size: 8.78 GB Docker image size: 8.69 GB Docker image size: 9.16 GB Docker image size: 9.12 GB Docker image size: 8.86 GB Docker image size: 8.77 GB Docker image size: 8.62 GB Docker image size: 8.43 GB Docker image size: 8.3 GB Docker image size: 8.62 GB Docker image size: 8.73 GB
DIGITS Release paused Release paused Release paused 6.1.1 including 6.1.1 including 6.1.1 including 6.1.1 including 6.1.1 including 6.1.1 including 6.1.1 including 6.1.1 including
- - - Docker image size: 14.6 GB Docker image size: 14.9 GB Docker image size: 15 GB Docker image size: 14.7 GB Docker image size: 15.1 GB Docker image size: 15.1 GB Docker image size: 15.4 GB Docker image size: 15.5 GB
NVIDIA Optimized Deep Learning Framework, powered by Apache MXNet Release paused Release paused Release paused 1.9.0.rc6 including 1.9.0.rc6 including 1.9.0.rc3 including 1.9.0.rc2 including 1.8.0 including 1.8.0 including 1.8.0 including 1.8.0.rc2 including
- - - Docker image size: 11.2 GB Docker image size: 10.9 GB Docker image size: 10.6 GB Docker image size: 10.4 GB Docker image size: 10.8 GB Docker image size: 10.7 GB Docker image size: 11.1 GB Docker image size: 10.8 GB
PyTorch 1.11.0a0+b6df043 including 1.11.0a0+b6df043 including 1.10.0a0+0aef44c including 1.10.0a0+3fd9dcf including 1.10.0a0+3fd9dcf including 1.10.0a0+ecc3718 including 1.9.0a0+c3d40fd including 1.9.0a0+2ecb2c7 including 1.9.0a0+2ecb2c7 including 1.9.0a0+df837d0 including 1.8.0a0+52ea372 including
Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)                
Docker image size: 14.7 GB Docker image size: 14.5 GB Docker image size: 13.2 GB Docker image size: 13.1 GB Docker image size: 12.7 GB Docker image size: 15 GB Docker image size: 14.5 GB Docker image size: 14.5 GB Docker image size: 14.3 GB Docker image size: 14.4 GB Docker image size: 12.9 GB
TensorFlow 2.6.2 including 1.15.5 including 2.6.0 including 1.15.5 including 2.6.0 including 1.15.5 including 2.6.0 including 1.15.5 including 2.5.0 including 1.15.5 including 2.5.0 including 1.15.5 including 2.5.0 including 1.15.5 including 2.4.0 including 1.15.5 including 2.4.0 including 1.15.5 including 2.4.0 including 1.15.5 including 2.4.0 including 1.15.5 including
Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)                                
Docker image size: 12.8 GB Docker image size: 16.9 GB Docker image size: 12.5 GB Docker image size: 16.5 GB Docker image size: 10.6 GB Docker image size: 14.5 GB Docker image size: 11.5 GB Docker image size: 13.6 GB Docker image size: 11.5 GB Docker image size: 13.9 GB Docker image size: 11.1 GB Docker image size: 14 GB Docker image size: 10.8 GB Docker image size: 13.7 GB Docker image size: 10.8 GB Docker image size: 14.1 GB Docker image size: 10.6 GB Docker image size: 14.1 GB Docker image size: 10.9 GB Docker image size: 14.4 GB Docker image size: 11.1 GB Docker image size: 14.5 GB
TensorRT TensorRT 8.2.1.8 including: TensorRT 8.0.3.4 including: TensorRT 8.0.3.4 including: TensorRT 8.0.3 including: TensorRT 8.0.1.6 including: TensorRT 8.0.1.6 including: TensorRT 7.2.3.4 including: TensorRT 7.2.3.4 including: TensorRT 7.2.3.4 including: TensorRT 7.2.2.3 including: TensorRT 7.2.2.3+cuda11.1.0.024 including:
Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)                
Docker image size: 5.98 GB Docker image size: 5.88 GB Docker image size: 6.37 GB Docker image size: 6.3 GB Docker image size: 6.04 GB Docker image size: 5.95 GB Docker image size: 5.8 GB Docker image size: 5.76 GB Docker image size: 5.63 GB Docker image size: 5.94 GB Docker image size: 7.09 GB
Triton Inference Server In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.17.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.16.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.15.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.14.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.13.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.12.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.11.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.10.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.9.0 including

In addition to the hardware listed above, Triton Inference Server also supports:
  • AMD x86 CPU
  • Intel x86 CPU

2.8.0 including

2.7.0 including
Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta) Multi arch support: x86, Arm SBSA (beta)                
Docker image size: 12.1 GB Docker image size: 12.2 GB Docker image size: 13.7 GB Docker image size: 13.6 GB Docker image size: 13.1 GB Docker image size: 14.6 GB Docker image size: 13.4 GB Docker image size: 10.6 GB Docker image size: 11.1 GB Docker image size: 11.3 GB Docker image size: 15.6 GB
TensorFlow For Jetson TensorFlow 1.15.5 and 2.6.2 for Jetson TensorFlow 1.15.5 and 2.6.0 for Jetson TensorFlow 1.15.5 and 2.6.0 for Jetson TensorFlow 1.15.5 and 2.6.0 for Jetson TensorFlow 1.15.5 and 2.5.0 for Jetson TensorFlow 1.15.5 and 2.5.0 for Jetson TensorFlow 1.15.5 and 2.5.0 for Jetson TensorFlow 1.15.5 and 2.4.0 for Jetson TensorFlow 1.15.5 and 2.4.0 for Jetson TensorFlow 1.15.5 and 2.4.0 for Jetson TensorFlow 1.15.5 and 2.4.0 for Jetson
Triton for Jetson Triton Inference Server 2.17.0 for Jetson Triton Inference Server 2.16.0 for Jetson Triton Inference Server 2.15.0 for Jetson Triton Inference Server 2.14.0 for Jetson Triton Inference Server 2.13.0 for Jetson Triton Inference Server 2.12.0 for Jetson Triton Inference Server 2.11.0 for Jetson Triton Inference Server 2.10.0 for Jetson Triton Inference Server 2.9.0 for Jetson Triton Inference Server 2.8.0 for Jetson Triton Inference Server 2.7.0 for Jetson


Important:

Content that is included in <<>> brackets indicates new content from the previously published version.

Note:

The deep learning framework container packages follow a naming convention that is based on the year and month of the image release. For example, the 20.06 release of an image was released in June, 2020.

20.xx container images

Table 5. Software stack packaged with the 20.xx container images
Container Image 20.12 20.11 20.10 20.09 20.08 20.07 20.06 20.03 20.02 20.01
DGX
DGX System
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX A100
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX Station
  • DGX-1
  • DGX-2
  • DGX Station
  • DGX-1
  • DGX-2
Operating System DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1
DGX OS
  • 4.1+1 (4.6+ multi-node NCCL)
  • 4.99.x (DGX A100)
  • 5.0

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+1
DGX OS
  • 4.1+
  • 4.99.x (DGX A100)

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+7
DGX OS
  • 4.1+
  • 4.99.x (DGX A100)

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+
DGX OS
  • 4.1+
  • 4.99.x (DGX A100)

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+
DGX OS
  • 4.1+1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+
DGX OS
  • 4.1+1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+
DGX OS
  • 4.1+1

Red Hat Enterprise Linux 7 / CentOS 72

  • EL7-20.02+
DGX OS
  • 4.1+1
NVIDIA Certified Systems
NVIDIA Driver

Release 20.12 is based on CUDA 11.1.1, which requires NVIDIA driver release 455.23.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx, 440.30, or 450.51.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.11 is based on CUDA 11.1.0, which requires NVIDIA driver release 455.23.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx, 440.30, or 450.51.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.10 is based on CUDA 11.1.0, which requires NVIDIA driver release 455.23.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx, 440.30, or 450.51.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.09 is based on CUDA 11.0.3, which requires NVIDIA driver release 450.51.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.08 is based on CUDA 11.0.3, which requires NVIDIA driver release 450.51.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.07 is based on CUDA 11.0.194, which requires NVIDIA driver release 450.51.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.06 is based on CUDA 11.0.167, which requires NVIDIA driver release 450.36.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.03 is based on CUDA 10.2.89, which requires NVIDIA driver release 440.33.01.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 396, 384.111+, 410, 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.02 is based on CUDA 10.2.89, which requires NVIDIA driver release 440.33.01.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 396, 384.111+, 410, 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

Release 20.01 is based on CUDA 10.2.89, which requires NVIDIA driver release 440.33.01.

However, if you are running on Tesla (for example, T4 or any other Tesla board), you may use NVIDIA driver release 396, 384.111+, 410, 418.xx or 440.30.

The CUDA driver's compatibility package only supports particular drivers. 3

GPU Model
Base Container Image
Container OS Ubuntu 20.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04 Ubuntu 18.04
CUDA 11.1.1 11.1.0 11.1.0