NVIDIA Releases the Latest Version of the CUDA Toolkit – Version 11.8

20 October, 2022

1.167

Views0 Comments

NVIDIA has released the CUDA Toolkit 11.8, the most recent version of the general-purpose graphics processing unit software. The latest CUDA Toolkit brings a big usability improvement for Jetson owners. It is equipped with new hardware capabilities which aim at speeding up CUDA applications and improving the programming model.

The CUDA Toolkit 11.8 release guarantees simpler and faster upgrades for customers working in the company’s embedded hardware ecosystem. It comes with a compatibility package that makes it easier to upgrade CUDA versions without having to utterly refresh the entire operating system.

“Starting from CUDA 11.8, [NVIDIA] has introduced an upgrade path that provides Jetson developers with an option to update the CUDA driver and the CUDA Toolkit to the latest versions. These upgrades are made possible by the introduction of the CUDA driver upgrade (also referred to as the CUDA compatibility package). Without this package, you were previously limited to the functionality provided by the default CUDA driver that was packaged in the Jetson Linux BSP (Board Support Package). You had no mechanism to upgrade to the latest CUDA driver and toolkit,”

the company explains.

A staff at NVIDIA also hinted that simplifying CUDA upgrades for NVIDIA Jetson users is only the tip of the iceberg as the company is working towards a time when Jetson developers will be able to just migrate to newer versions of the compute libraries without necessarily upgrading Jetson Linux.

“a future where Jetson developers can migrate to newer versions of the compute libraries without upgrading Jetson Linux. This CUDA feature that enables upgrading CUDA is a step in that direction.”

The CUDA Toolkit 11.8 version also comes with other crucial enhancements such as:

Support for lazy loading of CPU-side modules. This means functions and libraries will load faster on the CPU side along with reasonable memory footprint reductions, which will thereafter reduce the overall latency.
Improved multi-process service (MPS) signal handling – allows you to halt and terminate programs running in MPS environments without affecting other running processes.
Enhanced versions of the CUDA developer tools (such as Nsight Compute and Nsight Systems profiling tools) which help to identify and correct performance issues,
Support for next-generational NVIDIA Hopper and NVIDIA Ada Lovelace architecture processor. And,
NVIDIA JetPack installation simplification