Tools, Data Sets & Source Code

This page has links to the software tools, open datasets, and source code repositories released by the Tachyon Resilient Modeling project. Most of our tools are open source and released under permissive licenses. Datasets fall into several categories and are released either as currated collections (with permanent DOIs) or as non-currated but explorable datasets which can be subsetted to allow for specific training or analysis tasks.

Man of the resources on this page are still under development and will be released in the near future, or after initial publications and documentation on them are available. If there are specific resources you are interested in, please contact us and we will try to provide access to them.

Software Tools

tool

CODES

The Co-Design of Exascale Storage Architectures (CODES) simulation framework builds upon the ROSS parallel discrete-event simulation engine to provide high-performance simulation utilities and models for building scalable distributed systems simulations.

Repository · Language: C

tool

CQSim

An open-source, discrete-event driven cluster scheduling simulator used to emulate job submission, allocation, and execution against workload traces under configurable scheduling policies.

Repository · License: MIT

tool

tachyon-core

Core library for resilient model composition, uncertainty propagation, and fault-tolerant execution. Python and C++ APIs.

Repository · Docs · License: Apache-2.0

tool

tachyon-bench

Command-line benchmarking harness for evaluating resilience and reproducibility of modeling pipelines.

Repository · Docs · License: MIT

Data Sets

data

Data Explorer - Perlmutter

A fully anonymized dataset of node failure, batch queue behavior, and workflow behavior on the Perlmutter Supercomputer.

Explore · DOI · License: CC-BY-4.0

data

Data throughput time series - HEP Experiments Coming Soon

Datasets on the data throughput characteristics of experiments in high energy physics. Includes real time series sets from operations of detectors.

data

Curated Datasets - HEP Workflows Coming Soon

Curated datasets for batch system behavior, node failure, data throughput, workflow characterization and other purposes.

Source Code Repositories

Repository	Description	Language	License	Link
`tachyon-core`	Core resilient modeling library	C++ / Python	Apache-2.0	GitHub
`tachyon-bench`	Benchmarking harness	Python	MIT	GitHub
`tachyon-datasets`	Dataset generation and packaging scripts	Python	MIT	GitHub
`tachyon-org-website`	This website	HTML / CSS / JS	MIT	GitHub

Reproducibility

Each release is versioned and archived with a permanent DOI. Datasets include checksums and provenance metadata, and every tool ships with a CITATION.cff file and example workflows so results can be reproduced end to end.