High Performance Computing Core

The High Performance Computing Core (HPCC) at Weill Cornell Medicine gives researchers the computing power, storage, and expertise to do science at scale. Whether you’re processing terabytes of sequencing data, analyzing high-resolution imaging, or training machine learning models, we help you go beyond what a laptop or lab workstation can handle.

Our Mission

To accelerate discovery by giving WCM investigators secure, high-performance computing resources and the technical support to use them well.

What We Provide

We maintain a robust computing ecosystem designed for security, reliability, and speed. Our services bridge the gap between local data analysis and large-scale computation through:

• HPC Cluster Access: More than 15,000 CPU cores and 251 NVIDIA GPUs (H200, H100, A100, L40S, A40), with standard nodes of 256+ GB RAM and high-memory nodes for the largest jobs, all scheduled through SLURM over an InfiniBand network.

• Large-Scale Storage: Over 26 PB of high-performance Lustre, GPFS, and NFS storage with up to 200 GB/s throughput, built for genomics, imaging, simulation, and other large datasets.
  
  
• Scientific Software: 500+ curated modules through Lmod, including Python, R, MATLAB, TensorFlow, PyTorch, and hundreds of domain-specific tools.
  
  
• Open OnDemand (OOD) Access: A web portal for launching RStudio, Jupyter, MATLAB, and remote desktop sessions, managing files, and submitting and monitoring jobs, all from your browser.
  
  
• Bioinformatics Support: Consultation and pipeline development for genomics, transcriptomics, proteomics, and other omics work, with ready-to-run Nextflow, nf-core, and Snakemake pipelines and GPU nodes for tools like AlphaFold.
  
  
• Data & Workflow Services: Guidance on building reproducible workflows, containerizing pipelines with Singularity/Apptainer, and organizing your data, plus tools for fast, secure transfers between the cluster, cloud platforms, and collaborator sites through Internet2 and Globus.
  
  
• Cloud Connectivity: Use AWS and Google Cloud alongside the on-premises cluster. The awscli and Google Cloud SDK modules are installed on the cluster, so you can move data between the cluster and cloud storage.
  
  
• Training & Workshops: Regular sessions on HPC fundamentals, job scheduling, parallel programming, and best practices, open to all WCM researchers. https://its.weill.cornell.edu/scientific-computing-training-series