The Compute Unified Device Architecture (CUDA) software stack consists of the driver layer, runtime layer, and function library layer. APIs used in the CUDA software stack include the driver API and runtime API. This topic describes how to obtain the versions of the driver API and runtime API. This topic also describes the compatibility between the CUDA Toolkit versions and NVIDIA driver versions.
Introduction to CUDA
CUDA is a parallel computing platform and programming model released by NVIDIA in 2007. CUDA uses the GPU to significantly improve the computing performance.
The following figure shows the architecture of CUDA. The driver API and runtime API in the CUDA software stack have the following differences:
Driver API: provides a variety of complex features.
Runtime API: encapsulates partial drivers and provides implicit driver initialization.
The CUDA driver API is included in the NVIDIA driver package and the CUDA library and CUDA runtime are included in the CUDA Toolkit package.
Obtain the version of the driver API
If a node is already installed with the NVIDIA driver package, run the nvidia-smi command to view the versions of the NVIDIA driver and driver API. The following figure shows that the version of the NVIDIA driver installed on the node is 515.105.01 and the version of the driver API is 11.7. This means that the driver supports CUDA runtime API 11.7 or earlier versions.
Obtain the version of the runtime API
GPU-accelerated applications deployed in containers and GPU-accelerated applications deployed on nodes have the following differences.
The NVIDIA driver is installed on nodes.
The CUDA base images provided by NVIDIA are used to build container images. The CUDA Toolkit is already installed in these base images. You can build container images based on the base images. You can also select a CUDA base image based on the version of the CUDA Toolkit.
If a GPU-accelerated application is deployed in containers, the version of the CUDA runtime API used by the application is the same as the version of the CUDA base image that is used to build the Docker image. For example, if your Docker image is built based on CUDA base image NVIDIA/CUDA:12.2.0-base-Ubuntu20.04, the version of the CUDA runtime API used by your application is 12.2.0.
Compatibility between the CUDA Toolkit versions and NVIDIA driver versions
The following table describes the compatibility between the CUDA Toolkit versions and NVIDIA driver versions. Some NVIDIA driver versions listed in this table may not be supported by ACK. For more information about the NVIDIA driver versions supported by ACK, see NVIDIA driver versions supported by ACK. This table helps you choose an NVIDIA driver version based on the CUDA Toolkit version used by your application. For more information, see cuda-toolkit-release-notes.
CUDA Toolkit version | Linux x86_64 driver version |
CUDA 12.1 Update 1 | >=530.30.02 |
CUDA 12.1 GA | >=530.30.02 |
CUDA 12.0 Update 1 | >=525.85.12 |
CUDA 12.0 GA | >=525.60.13 |
CUDA 11.8 GA | >=520.61.05 |
CUDA 11.7 Update 1 | >=515.48.07 |
CUDA 11.7 GA | >=515.43.04 |
CUDA 11.6 Update 2 | >=510.47.03 |
CUDA 11.6 Update 1 | >=510.47.03 |
CUDA 11.6 GA | >=510.39.01 |
CUDA 11.5 Update 2 | >=495.29.05 |
CUDA 11.5 Update 1 | >=495.29.05 |
CUDA 11.5 GA | >=495.29.05 |
CUDA 11.4 Update 4 | >=470.82.01 |
CUDA 11.4 Update 3 | >=470.82.01 |
CUDA 11.4 Update 2 | >=470.57.02 |
CUDA 11.4 Update 1 | >=470.57.02 |
CUDA 11.4.0 GA | >=470.42.01 |
CUDA 11.3.1 Update 1 | >=465.19.01 |
CUDA 11.3.0 GA | >=465.19.01 |
CUDA 11.2.2 Update 2 | >=460.32.03 |
CUDA 11.2.1 Update 1 | >=460.32.03 |
CUDA 11.2.0 GA | >=460.27.03 |
CUDA 11.1.1 Update 1 | >=455.32 |
CUDA 11.1 GA | >=455.23 |
CUDA 11.0.3 Update 1 | >= 450.51.06 |
CUDA 11.0.2 GA | >= 450.51.05 |
CUDA 11.0.1 RC | >= 450.36.06 |
CUDA 10.2.89 | >= 440.33 |
CUDA 10.1 (10.1.105 general release, and updates) | >= 418.39 |
CUDA 10.0.130 | >= 410.48 |
CUDA 9.2 (9.2.148 Update 1) | >= 396.37 |
CUDA 9.2 (9.2.88) | >= 396.26 |
CUDA 9.1 (9.1.85) | >= 390.46 |
CUDA 9.0 (9.0.76) | >= 384.81 |
CUDA 8.0 (8.0.61 GA2) | >= 375.26 |
CUDA 8.0 (8.0.44) | >= 367.48 |
CUDA 7.5 (7.5.16) | >= 352.31 |
CUDA 7.0 (7.0.28) | >= 346.46 |
References
When you install an NVIDIA driver on a node in an ACK cluster, you must make sure that the version of the NVIDIA driver that you install is supported by ACK. For more information, see NVIDIA driver versions supported by ACK.