ASU Research Computing supports research by offering supercomputers and other advanced computational resources to faculty, staff, and students. These supercomputers are a special configuration of multiple machines, usually containing hundreds of compute nodes, thousands of central processing unit (CPU) cores, hundreds of discrete graphic processing units (GPUs), and other specialized hardware, such as field programmable gate arrays (FPGAs). These supercomputers advance research and acceleration to discovery by offering:

• Increased Compute Capacity

• More Processing Speed

• Modeling and Simulation

• Data and Artificial Intelligence (AI)

• Visualization

Open

A visualization of a protein found in chlorophyll. This visualization was produced by Yuval Mazor, School of Molecular Sciences, on the Agave supercomputer.

Supercomputers Available at ASU:

There are three supercomputers available at ASU:

1. The Sol supercomputer is the main supercomputer open to the ASU community. This is a multi-million dollar supercomputer personally funded by President Crow.

2. The Phoenix supercomputer, formerly known as the Agave supercomputer.

3. The Aloe supercomputer is a HIPAA-compliant supercomputer that is available for research dealing with restricted data.

How Much Does it Cost?

The Sol and Phoenix supercomputers are free to ASU faculty, staff, students, and affiliates. Utilization and job scheduling rely on “Fairshare”, an algorithm to help equitably distribute compute resources to all users based on CPU-Hours.

The Aloe supercomputer is a paid service under the KE Secure Cloud.

How Is CPU-Hour Time Calculated?

A program that runs on a single CPU core for one hour uses one (1) CPU hour. If your program uses multiple CPUs, then it will use multiple CPU hours for each hour of execution time. For example, a program that runs for one hour and uses eight CPU cores would have used eight CPU hours.

CPU-hour time=Number of CPU cores×execution time

What Is the Maximum Amount of CPU Hours I Can Use?

There is no maximum or cap on how many CPU hours users can utilize. Users are welcome and encouraged to utilize the supercomputer as much as they need. Keep in mind that because of Fairshare, the more resources you use, the longer wait times you may experience.

When Do I Know That I’m Ready to Use a Supercomputer?

• You are unable to run your model or computations on your local machine.

• You are working with large amounts of data.

• You need to calculate multiple or complex computations quickly.

• You need to implement machine learning or similar methodologies for your research.