E4S: Extreme-scale Scientific Software Stack
summary
E4S assembles a coherent HPC stack of about 70 products and a v1.1 release with over 45 full packages, providing Spack source builds plus reproducible containers and binary build caches with more than 8,000 ppc64le binaries. It ships GPU-enabled images (CUDA 10.1/10.2, TensorFlow 2.2, PyTorch, Keras preinstalled), CI build pipelines, a validation test suite, a VirtualBox dev image and deployments at DOE centers using Singularity/Shifter/Charliecloud and supports MPI ABI swap for native interconnects.
AI-written summary · about
chapters (15)
- 0:00Intro
- 0:16Challenges
- 0:48ECP's three technical areas have the necessary components to meet national goals
- 1:27We work on products applications need now and into the future
- 2:19Software Technology Ecosystem
- 2:48Motivation: Improving Productivity
- 5:31E4S: Extreme-scale Scientific Software Stack
- 9:45Download E4S v1.1 GPU image
- 11:10Scientific software is becoming extremely complex
- 12:15The Exascale Computing Project is building an entire ecosystem
- 13:07How to install software on a supercomputer
- 14:11Spack is a flexible package manager for HPC
- 14:41Spack provides the spec syntax to describe custom configurations
- 19:48E4S Validation Test Suite
- 23:07E4S Summary What E4S is not
related talks
23mNeSI's Flexible HPC-A programmable infrastructure for science
25mE4S: Enabling Interoperable Edge Architectures for Secondary Sub...
26mWhy the Next Generation of HPC systems Needs Open Source Driven Standardization
11mKeynote: Open Source Stack Empowers the AI Era
33mFighting against the instability : Debian Science at the synchrotron SOLEIL
32mEPI and beyond: perspectives of a European supercomputing center on opensource hardware and software
34mSCS: Buildig Open Source Cloud and Container Infrastructure
28mBuilding a Sustainable Future through Open Source