Skip to content

HPC Overview

Picture of the cluster

This page describes the High Performance Compute Cluster at FSU.

Overview#

The High Performance Computing cluster (HPC) at FSU is the core of our computational infrastructure. With more than 16,000 cores and over 600 compute nodes, the HPC provides a powerful and scalable computing platform for large, multithreaded and distributed parallel computations.

The HPC is a tightly integrated system of uniform servers connected by a fast InfiniBand data network that is designed for long-running, compute-intensive programs. This uniformity and integration makes the system extremely well suited for processing workloads that would not scale on regular computers because of memory requirements or CPU limitations.

What is it used for?#

The HPC is used for long-running jobs that require a large amount of compute resources (CPUs and memory). To allow many users to run programs at the same time, HPC systems make use of batch, non-interactive jobs that are scheduled on a large but finite amount of resources.

In a batch system, users describe the workflow of their program and submit it to a queue (or Slurm Account, as it is implemented at FSU), it runs independently of any user input until it finishes. Most jobs can be monitored, but not interacted with.

Compute jobs on the HPC can operate in parallel using popular frameworks like MVAPICH2 or OpenMPI.

Many users write and/or compile their own software to run on the HPC, for which we provide a number of tools and libraries to support. Other users can run jobs using general-purpose applications, such as MATLAB or Python.

Who has access?#

Access to the HPC is available for all FSU faculty and students/staff with a faculty sponsor. To obtain priority access to our resources, many faculty members have made investments in the HPC by purchasing dedicated resources.

How can my research group purchase dedicated resources on the HPC?#

Information about how to purchase dedicated queues and priority access for your research group is available in the ITS Service Catalog.

The HPC can run many types of jobs, which makes it appropriate for all areas of scientific research. Some popular platforms and technologies include the following:

  • OpenMPI - OpenMPI is one of the several implementations of the Message Passing Interface (MPI) model of parallel computing for distributed systems. OpenMPI is an open-source implementation of this model with a wide range of powerful features
  • Python - The HPC provides a robust implementation of Python, which is increasingly used in computational science. We provide support for a large number of Python utilities for compiling Python code and working with Python visually
  • MATLAB - The HPC provides support for distributed MATLAB jobs. In addition, you can compile MATLAB to C and run that on the HPC for higher performance and fewer license restrictions
  • GPUs and CUDA - A portion of the HPC cluster provides GPU resources. CUDA is available for running GPU optimized code