Student Experiences with TACC HPC

Describe your experience using TACC resources as a student?

TACC was a great resource for me as a graduate trainee. I had a little experience from my undergraduate research using a computational cluster, but that cluster lacked information for anyone new to HPC. My experience with TACC was completely different. TACC's website includes an excellent documentation page on how to use the cluster and what is available on each node. Furthermore, the team at TACC was a great resource, helping me get familiar with the environment and scheduler, and making sure the code I wanted to run in the cluster ran properly. This gave me the confidence I needed to take full-advantage of the TACC resources.

What were you able to accomplish that would otherwise not have been possible without HPC?

At the time of my PhD training, GPU computing was rising in popularity.

Very quickly we were able to convert our software which was running on CPUs to run more efficiently on GPUs. But unfortunately, our department had a limited number of GPUs available at the time. We were excited to learn that TACC had launched the Maverick cluster, which at one point had as many as 132 Nvidia Tesla K40s.

Being able to run our software on Maverick literally saved me months of computing time by parallelizing the training of the models I was developing. I'm confident the initial data from this study (which was later published in the International Journal of Radiation Oncology Biology Physics) helped convince my PhD advisory committee that I wasn't wasting my time investigating the use of deep neural networks to learn patterns from our clinical practice.

What was the most valuable aspect of your experience using HPC for computational research?

I'm a big fan of HPC systems. Once I became more comfortable using TACC for research projects, I became interested in learning what happens under the hood of a research cluster.

After finishing up my PhD and joining the faculty at MD Anderson, I helped build the cluster we used to clinically-deploy the image processing tools we developed through our research. Furthermore, I joined our institution's Research HPC Consortium as the HPC team began expanding GPU computational resources within Seadragon (the MD Anderson research cluster).

I truly believe that trainees are in a prime position to leverage [HPC] resources to explore new ideas relevant to their discipline and research.

Did you take advantage of any of the other services that TACC offers?

I took a few courses and had extensive conversations with Joe Allen (TACC Life Sciences research and UT System liaison). Joe was a tremendous resource and I'm truly grateful that he was available to walk me through the beginning of my journey using TACC.

Personally, I think the main reason researchers shy away from using HPC resources is because they are not comfortable with the new environment. Education and training can really boost one's productivity and efficiency when using HPC resources. TACC does a great job making this process of getting familiar with HPC very smooth and has excellent customer support to make anyone willing to try HPC as comfortable as possible through their learning journey.

How did access to HPC resources, and experience using cutting-edge HPC systems, impact your career trajectory/job search?

I think having access to HPC resources definitely had an impact on my career path. I honestly believe the computational skills I developed while using HPC resources for medical physics research were one of the main reasons I was able to go from finishing my PhD training in medical physics directly to a medical physics faculty position at The University of Texas MD Anderson Cancer Center. Using HPC resources also allowed our research group to work more efficiently and rapidly scale our research to other cancer sites as additional trainees joined our group.

More recently, I joined The University of Alabama at Birmingham (UAB) as a clinical medical physicists in the Department of Radiation Oncology. One of the first things I did before applying for this position was getting in contact with the HPC group at UAB. Having HPC experience allowed me to better access the computational resources that UAB had available to faculty, which further increased my interest in the position.

I am very fortunate to have had HPC resources available through my training, as I am confident they have had a positive impact on my development towards becoming an independent scientist.

Why are centers like TACC, and investments in HPC, important for student development?

State and Federal investments in HPC centers is paramount in the advancement of knowledge for many research topics. I recently participated in a workshop that was organized by the National Cancer Institute on how to better connect computational experts with clinical experts to promote the integration of computational resources for basic and translational sciences in cancer research. Great academic research is made possible through HPC investments and I truly believe that trainees are in a prime position to leverage these resources to explore new ideas relevant to their discipline and research.

As a medical physicist whose research is focused on improving cancer care and patient outcomes, I am very fortunate to have had HPC resources available through my training, as I am confident they have had a positive impact on my development towards becoming an independent scientist.