E4S: An HPC-AI Software Ecosystem for Science

E4S offers multiple methods for deploying its comprehensive collection of HPC and AI software packages. Below is an overview of these deployment options.

From-Source Builds

E4S utilizes Spack, a flexible package manager, to facilitate building software directly from source. This approach allows users to customize builds according to their specific system architectures and requirements. Detailed instructions for manual installation are available in the E4S Manual Installation Guide.

Spack Build Cache

To expedite the installation process, E4S provides pre-built binaries through Spack build caches. These caches contain binaries for major operating systems and architectures, enabling users to install software without the need for local compilation. More information can be found on the E4S About Page.

Containers

E4S offers containerized versions of its software stack, compatible with platforms such as Docker, Singularity, Shifter, and CharlieCloud. You can see the current list of E4S containers under the Downloads page. These containers provide a consistent and portable environment for running HPC and AI applications across diverse systems. E4S provides containers from DockerHub and on cloud platforms, such as AWS, Azure, and Google Cloud. ParaTools Pro for E4S™ is a commercial version of E4S that provides additional support and services for E4S users including availability on AWS.

Cloud Options

  • Amazon Web Services (AWS): E4S is available on AWS, allowing users to deploy the software stack on cloud-based infrastructure. This facilitates scalable and flexible computing resources for various workloads. Details about E4S on AWS can be found on the E4S Home Page.
  • Google Cloud Platform (GCP): Users can also deploy E4S on GCP, leveraging Google's cloud services to run HPC and AI applications. This option provides access to high-performance virtual machines and storage solutions. Information regarding E4S on GCP is available on the E4S Home Page.