Welcome to the Illinois Campus Cluster’s High-Throughput Computing Service (HTC)
The Illinois Campus Cluster program is delighted to introduce HTC, a high-throughput computing service powered by the open-source workload manager, HTCondor. This service, partially funded by the Illinois Computes program, offers researchers free access to a pool of computational resources. These resources primarily consist of retired campus cluster nodes and donated hardware from projects like Compute for Humanity. HTC is currently in open-beta.
Getting Started with HTC
To apply for access, please fill out this form.
Understanding HTCondor
HTCondor is an open-source workload management system designed for high-throughput computing (HTC) environments. It specializes in managing and scheduling many compute-intensive jobs across diverse and distributed resources.
Technical Capabilities of HTCondor
- Resource Pooling: HTCondor has a much looser structure for its execute nodes (compute nodes) allowing for contributing servers to be less tightly coupled than a traditional HPC cluster
- Job Queuing and Scheduling: Users submit jobs with specific requirements (CPU, memory, etc.). HTCondor matches jobs to available resources based on priorities and policies.
Benefits of Using HTC
- It is free!
- Increased efficiency by utilizing idle computing resources.
- Cost-effectiveness by leveraging shared resources.
Is HTC Right for You?
- SLURM: Primarily designed for High Performance Computing (HPC). It’s optimized for tightly-coupled parallel jobs with high resource requirements (cores, memory, I/O) running on dedicated clusters, like the Campus Cluster.
- HTCondor: Focuses on High Throughput Computing (HTC). It manages a wider variety of workloads, including independent tasks and loosely coupled jobs. It works well with heterogeneous servers and can scavenge resources from underutilized servers.