Services
The Resnick High Performance Computing Center operates Caltech’s central research computing infrastructure and supports the researchers who rely on it every day. Beyond maintaining the systems themselves, we work directly with research groups to make workloads run faster, scale further, and operate more reliably — from a first batch job to large-scale parallel and GPU-accelerated computing.
Getting started
Whether you are new to the cluster or bringing an entire research group onboard, we help users get productive quickly:
Accounts and access — provisioning individual and group access with the appropriate permissions, storage, and computational resources.
Allocations and scheduling guidance — matching workloads to the right partitions, node types, and accelerators so jobs run efficiently.
Onboarding and training — helping new users submit their first jobs, understand cluster workflows, and build long-term self-sufficiency through documentation and hands-on support.
Performance & optimization
We work alongside research groups to improve performance and make better use of computational resources:
Profiling — identifying where runtime, memory, and I/O bottlenecks occur so optimization effort goes where it matters most.
Parallelization — scaling workloads with MPI, OpenMP, multithreading, and GPU acceleration.
GPU and accelerator support — assisting with CUDA environments, GPU-aware applications, and performance tuning for accelerated workloads.
Debugging — diagnosing crashes, hangs, scheduler failures, environment conflicts, and numerical instability in distributed computing environments.
Performance and scaling assessments — evaluating how applications behave as workloads, node counts, and problem sizes grow.
Scheduler and job efficiency recommendations — improving resource requests, walltime estimates, queue efficiency, and overall cluster utilization.
Software & environment support
Getting the right software running reliably is often the first challenge. We assist with:
Installing and building software — from centrally maintained applications and modules to third-party and source-built research software.
Python and conda environments — creating reproducible software environments that behave consistently across runs and systems.
Module management — locating available software, selecting compatible versions, and maintaining clean runtime environments.
Containerized workflows — supporting Apptainer/Singularity containers for portable and reproducible research environments.
Workflow automation and pipelines — assisting with batch workflows, job dependencies, and scalable computational pipelines.
If something you need is not currently available, contact us at help-hpc@caltech.edu — in many cases we can deploy it for the broader research community.
Data transfer & storage
Data transfer workflows — guidance on moving data to and from the cluster, including Globus for large or scheduled transfers alongside home, group, and scratch storage tiers.
Quota management and storage planning — monitoring usage, identifying space constraints, and helping research groups manage long-term data growth efficiently.
Interactive & visualization computing
Open OnDemand and interactive computing — browser-based access to the cluster including shells, file management, Jupyter, RStudio, MATLAB, and full Linux desktop environments for GUI-based applications, all without requiring local software installation.
Get in touch
Email help-hpc@caltech.edu or open a ticket through the Caltech Help System to discuss any of the above.