Sign up for our newsletter and get the latest HPC news and analysis.
Send me information from insideHPC:


NVIDIA Makes GPU Computing Easier in the Cloud

Setting up an application deployment environment for high performance computing (HPC) can be daunting. There can be multiple dependencies, a number of supporting libraries required, and complex installation instructions that have to align with the available system software stack, all while also supporting hundreds of other applications with diverse requirements.

NVIDIA has made application deployment easier with HPC application containers on NVIDIA GPU Cloud.

HPC applications make scientific discoveries possible by harnessing the power of CPUs and GPUs working together. Over the past decade, GPUs have been rapidly adopted to speed up certain types of calculations. Now, simulations run much faster and the pace of scientific discoveries is accelerated. With the availability of containers for accelerated applications on NVIDIA GPU Cloud (NGC), it’s never been simpler to set up an entire deployment environment.

Development and deployment of complex codes are simplified by incorporating all of the libraries, drivers, compilers, and the application itself in a container. Developers and users can easily set up an environment to deploy applications in minutes—enabling productivity and delivering faster results.

“LIBCCHEM is our commercial-free GPU-accelerated quantum chemistry library that uses many other open-source libraries, and using LIBCCHEM requires a very complex build process for the end user,” noted Sarom Leang, GAMESS developer. “Containerizing the complexity of our software stack and making it available on a registry like NGC allows us to minimize end users’ effort to get started with our tools. The GAMESS container also ensures numerical- and performance-reproducibility for our computational methods.”

As new application features or improved libraries become available, a container can be created and easily released, giving end users access to all of the new features and improved performance. And the container can be released without having to be tied to the underlying system.

All the third-party-managed HPC applications available through the NGC registry are designed to run on GPU-accelerated systems. Application users can log in to NGC and pull the containers to run on a local workstation, GPU-accelerated HPC clusters, NVIDIA® DGX™ Systems, and in the cloud. This kind of flexibility enables more widespread use of applications and changes the way HPC drives scientific breakthroughs.

Learn more about and get access to HPC application containers with NVIDIA GPU Cloud.

Leave a Comment

*

Resource Links: