CUDAToolkit 11.1 and CUDAToolkit 11.3

CUDAToolkit is a free and powerful suite of GPU-accelerated tools and libraries for developing applications that leverage NVIDIA GPUs for high-performance computing. In this article, we will discuss the differences between two major releases of CUDAToolkit: version 11.1 and version 11.3.

CUDAToolkit 11.1

CUDAToolkit 11.1 was released on April 28, 2020. This release includes several new features and improvements, such as:

• Improved performance for FP16 and INT8 convolutions in cuDNN
• Support for CUDA 11 features such as cooperative groups, warp-level primitives, and async-copy to shared
• cuBLAS_GEMMEx support for INT8 and FP16 datatypes
• Support for NVIDIA Ampere architecture

This version also includes several bug fixes and updated libraries such as cuTENSOR, cuSPARSE, and nvJPEG. It is compatible with NVIDIA driver version 440 or later.

To install CUDAToolkit 11.1 on Linux, you can use the following commands:

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/cuda-ubuntu1804.pin
sudo mv cuda-ubuntu1804.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/11.1.0/local_installers/cuda-repo-ubuntu1804-11-1-local_11.1.0-455.23.05-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1804-11-1-local_11.1.0-455.23.05-1_amd64.deb
sudo apt-key add /var/cuda-repo-ubuntu1804-11-1-local/7fa2af80.pub
sudo apt-get update
sudo apt-get -y install cuda

CUDAToolkit 11.3

CUDAToolkit 11.3 was released on April 7, 2021. This version includes several new features and improvements:

• Performance improvements in cuDNN for training and inference workloads
• Support for CUDA 11.2 features such as cross-process device access and bank conflict-free shared memory
• cuBLAS GEMMEx enhancements for INT8 and FP16 datatypes
• Support for multi-threaded RTP and CUDA Graphs APIs
• Compatibility with NVIDIA Ampere architecture

This version also includes bug fixes and updated libraries such as nvJPEG, nvCRIO, and nvCOMP.

To install CUDAToolkit 11.3 on Linux, you can use the following commands:

wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2004/x86_64/cuda-ubuntu2004.pin
sudo mv cuda-ubuntu2004.pin /etc/apt/preferences.d/cuda-repository-pin-600
wget https://developer.download.nvidia.com/compute/cuda/11.3.0/local_installers/cuda-repo-ubuntu2004-11-3-local_11.3.0-465.19.01-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu2004-11-3-local_11.3.0-465.19.01-1_amd64.deb
sudo apt-key add /var/cuda-repo-ubuntu2004-11-3-local/7fa2af80.pub
sudo apt-get update
sudo apt-get -y install cuda

Conclusion

CUDAToolkit 11.1 and 11.3 are two major releases that bring new features and improvements to GPU programming. Developers can choose the version that best suits their needs and leverage the power of NVIDIA GPUs to accelerate their computations.

Note: The installation steps may vary depending on the Linux distribution and version you are using. Please refer to the official NVIDIA documentation for more information.