Welcome to E4S, the HPC-AI Software Ecosystem for Science — an open-source, community-driven collection of high-quality HPC and AI libraries and tools. E4S is an Extreme-scale Scientific Software Stack that enables scientists, developers, and institutions to develop, deploy, and run performance-portable applications across CPUs and GPUs from NVIDIA, AMD, Intel, and Arm. Ready to adapt to new architectures as they emerge.
Supported by the U.S. Department of Energy (DOE) and its partners, E4S accelerates scientific innovation on systems ranging from laptops to exascale supercomputers.
Performance Portability
Run anywhere—from laptops to exascale supercomputers using modern GPU and CPU architectures.
Flip for details
More Details
E4S supports programming models like Kokkos and MPI to enable single-source performance across CPUS and GPUs, and scalable clusters to leadership systems.
Curated Ecosystem
Over 120 interoperable libraries and tools for HPC and AI workflows, validated for research and production use.
Flip for details
More Details
E4S organizes products into product families such as math libraries, I/O, AI/ML, and performance tools—curated for compatibility and tested integration.
Reproducible Environments
Deploy via Spack, pre-built containers for Docker, Singularity, and commercial cloud environments.
Flip for details
More Details
Binary caches, container images, and continuous builds ensure consistent environments across facilities, enabling reproducible experiments and fair benchmarking.
Continuous Validation
E4S packages are tested across DOE leadership and cloud platforms for quality and reproducibility.
Flip for details
More Details
Automated pipelines validate builds nightly across dozens of devices on the Frank testbed, tracking API/ABI changes and ensuring cross-platform reproducibility.
Chat with the E4S Bot
Ask questions, explore tools, or get recommendations from the E4S Guide Bot.
Visit the E4S Product Catalog
Browse member products, product families, with extensive links, via the E4S Product Catalog
Trusted by the HPC-AI Community
E4S is developed and supported by contributors from U.S. national laboratories, universities, and industry partners. It is available as both open-source and commercial distributions such as ParaTools Pro for E4S™ for commercial cloud platforms such as AWS, Azure, GCP, OCI, and IBM Cloud.