HTCondor

HTCondor: Open-Source High-Throughput Computing Framework

Discover HTCondor, an open-source software framework harnessing unused CPU and GPU cycles for scientific research and high-throughput computing.

What is HTCondor?

HTCondor is an open-source high-throughput computing software framework developed by the University of Wisconsin-Madison. It facilitates coarse-grained distributed parallelization of computationally intensive tasks across large collections of distributively owned computing resources.

HTCondor provides a job scheduling system that can manage a cluster of computers to harness otherwise unused CPU and GPU cycles for high-throughput scientific computing. It implements a classified advertisement matchmaking framework to facilitate resource allocation among users and resource owners. HTCondor can be used to manage clusters of dedicated compute nodes, or harvest cycles from idle desktop machines, grid resources, and cloud computing services.

Key capabilities of HTCondor include job queuing, resource monitoring, resource management, priority calculation, resource allocation, job migration, and fault tolerance. It provides a resilient computing environment that can automatically provide alternative computing resources when some become unavailable. This allows users to focus on application development rather than cluster administration.

HTCondor is well-suited for a variety of high throughput computing applications, including Monte Carlo simulations, parameter sweeps, combinatorial optimization, and bioinformatics. It is in widespread use by the physics, computer science, engineering, biology, economics, and other sciences communities across many universities, national laboratories, and supercomputing centers.

HTCondor Features

Official Links