Determine whether high performance computing can help your research

Those new to high performance computing (HPC) often have a difficult time understanding if or when HPC can help their research. The information here offers a starting point, though it does not answer the question comprehensively. The staff of Indiana University's Research Technologies division of University Information Technology Services are happy to answer any additional questions. See Research and high performance computing for more information.

HPC systems provided by Research Technologies offer significant advantages over personal computers for unique or demanding computational tasks. However, for many day-to-day tasks, running programs on local personal computers may be more convenient and/or faster than using an HPC system. For example, depending on the specific situation, an HPC system may take longer to load a given application than a personal computer, but provides an advantage in terms of huge throughput for processing a large job.

You may benefit from HPC systems if you are working with one of the following:

• You have computationally intensive programs that can be divided into multiple tasks, taking advantage of parallel processing with multiple processors. For example, if your program takes only an hour to run but you need to run it a thousand times, or your program can be scaled up to run across multiple processors, HPC systems may be beneficial.
• Your programs require more CPUs, GPUs, or memory than your personal computer can handle.
• Your projects involve working with massive data sets that exceed the storage capacity of your personal computer.
• You estimate that you cannot complete your research tasks in a reasonable amount of time with your personal computing resources.

However, HPC systems may not provide an advantage if:

• Your tasks are simple or small in scale; for smaller tasks that cannot be scaled up, personal computing devices will likely perform better (for example, office-type tasks, spreadsheets, word processing, etc.).
• You are working with smaller data sets that can be processed effectively on your personal computer.

The increased parallel processing, expanded memory capacity, ample disk space, and efficient file system access offered by HPC systems make them indispensable for computationally demanding projects. On the other hand, do not use HPC systems for simple CPU jobs and smaller data sets, as these systems might not perform any better than your personal computing devices.

If you need help evaluating whether HPC is right for you, contact the UITS Research Applications and Deep Learning team.