The High Performance Software Foundation welcomed Ramble, a Python framework for capturing provenance and improving the reproducibility of computational experiments across HPC, AI and machine learning environments, as a new project. Ramble can create, execute, analyze and manage large sets of experiments and automate stages such as downloading inputs, building software stacks through package managers such as Spack, running experiments, extracting figures of merit and archiving results, with integrations for Slurm, Spack, environment modules, EESSI and pip. Its creator Doug Jacobsen, a Google software engineer, said joining HPSF would let the project collaborate more closely with the broader open source community and keep improving accessibility for HPC, AI and ML users.
HPSF welcomes Ramble, a reproducible HPC experimentation framework
The High Performance Software Foundation welcomed Ramble, a Python framework for capturing provenance and improving the reproducibility of computational experiments across HPC, AI and machine learning environments, as a new project. Ramble can create, execute, analyze and manage large sets of experiments and automate stages such as downloading inputs, building software stacks through package managers such as Spack, running experiments, extracting figures of merit and archiving results, with integrations for Slurm, Spack, environment modules, EESSI and pip. Its creator Doug Jacobsen, a Google software engineer, said joining HPSF would let the project collaborate more closely with the broader open source community and keep improving accessibility for HPC, AI and ML users.
Source: Hpsf