Newer NVIDA GPUs like the GeForce RTX 3090
require CUDA 11
.
TensorFlow provides pre-built Docker images on dockerhub - tensorflow/tensorflow. The underlying Docker files are availble in the TensorFlow GitHub repository.
Currently latest TF version is 2.7
. In table gives an overview of CUDA version used within the pre-built Docker images for the specified TF versions. Versions were retrieved from the TensorFlow GitHub repository (date 2022-01-10).
CUDA version | Oldest supported TF version | Latest supported TF version |
---|---|---|
11 |
2.4.x |
2.7.x |
10 |
1.14.x |
2.3.x |
Overview of tested build configurations.
NVIDIA also provides pre-built Docker images for TensorFlow. They also provide Docker images for older TF versions supporting CUDA 11.
The deep learning framework containers are stored in the
nvcr.io/nvidia
repository.
An overview can be found here: