- Supercomputing involves connecting smaller computers to a larger processing network for more computing options and capabilities.
- The University of Miami partnered with IBM this year for a $3.7 million update to the supercomputer it's had in place since 2008.
- The new system is eight to 10 times more powerful than the previous one.
- Business Insider talked to Nicholas Tsinoremas, founder and director of the UM Center for Computational Science and UM's vice provost for research computing and data, and Dave Turek, VP of IBM's technical computing department, to better understand the effect of supercomputing on a university campus.
- Click here for more BI Prime stories.
Supercomputers are making the jump from government labs to higher education — and they may soon play as vital a role in college life as cramming for finals or ultimate Frisbee on the quad.
For a recent case study, look to the University of Miami in Florida.
UM partnered with tech giant IBM this year for a $3.7 million update to the number-crunching supercomputer it's had in place since 2008, turning it into one that can process data in real or near-real time. The new supercomputer, named TRITON, has since attracted an influx of requests across UM's three campuses and countless academic departments.
While supercomputing was once limited to laboratories with the funding to support it, the technology now encompasses industries like energy, transportation, and finance — with academia being the latest to join the fray.
Supercomputing systems are appealing because of their processing and analytic power. They consist of collections of smaller computers that link together, and are controlled through specific software to work as a unified whole.
For example, if the supercomputer is given a complicated math problem, the software that controls the whole system will detect the nature of the problem, parcel out activity and jobs to different nodes (the smaller computers within the supercomputer), collect the output from each of those nodes, and then bring everything together and analyze it as part of a bigger picture.
To get a sense of how supercomputing has worked in higher education, and to see what capabilities it offers that other schools might soon adopt, Business Insider talked with Nicholas Tsinoremas, who founded and directs the UM Center for Computational Science and serves as UM's vice provost for research computing and data. He says that demand in higher education for supercomputing is set to rise.
Heightened demand for supercomputing
TRITON is already being used in several classes within UM's genetics program, and there's a new class set to launch in January for computer science students who are interested in utilizing the system's AI and machine-learning capabilities.
"There are plenty more," Tsinoremas told Business Insider "We have similar classes in the marine and atmospheric sciences."
But supercomputers are applicable in more than just the hard sciences.
"At this point the students are pushing, and they want it in all their classes, not only in the traditional scientific classes," Tsinoremas said.
For humanities students, for example, TRITON can "read" hundreds of books within a day and extract relevant information. This capability underscores Tsinoremas' assertion that supercomputing is "very pervasive" in terms of applicability for students of all disciplines.
Business school offerings at UM have also expanded to include supercomputing. The university's business analytics program, called digital marketing, and master's in data science both use supercomputing to prepare students to better process data in the digital age.
"One of the things that we want to do is not only to catalyze the research that our researchers are doing here at the University of Miami, but also, if you will, educate the next generation of students to be familiar with these kinds of technologies and approaches," Tsinoremas said.
TRITON's computing power sets it apart: The system is about eight to 10 times more powerful than UM's previous supercomputer.
Redefining supercomputing
Dave Turek, vice president of technical computing at IBM, says that supercomputers essentially solve problems that can be represented mathematically. Over the past 10 years, though, their capabilities have expanded tremendously, keeping pace with ever-increasing quantities of data. Now what's important is how much data they can process in as short a time as possible.
TRITON currently processes more than 10 petabytes (10 million gigabytes) of data, which over 1,500 individual users across UM's three campuses currently utilize for research and education.
Turek says an easy way to visualize the power of supercomputing is thinking of a honeycomb in a beehive.
"Each one of those cells in a beehive is an independent thing, but the hive itself is composed of all these honeycombs that are interconnected to the same kind of thing," Turek told Business Insider. "We take each of these small computers, we connect them to each other and we communicate with them via software to orchestrate their behavior in a coherent way to tackle the problem at hand."
Analyzing cost
In terms of cost to benefit, Turek says that $3.7 million for the update to what became TRITON was inexpensive for a scientific instrument of that processing power.
"If I put in a supercomputer into a university, everybody gets to use it," Turek said. "As a result, it really amplifies the leverage that the university gets from the investment they've made."
Under software control, multiple people can use different parts of the supercomputer at the same time. So one full rack, or column of computing nodes, could go to the English department, and a quarter of a rack could go to a chemistry project.
The system can be divided again under software control to process different kinds of jobs at different times of the day.
"In a university context you might at one time be running problems from the chemistry department, prompts from physics, problems from the economics department, all at the same time in different locations inside this physical computer," Turek said.
The ROI for supercomputers, then, is found in the accelerated way that its benefits are distributed throughout the university.
"Things that might have taken months now take place in a day," Turek said.
Join the conversation about this story »
Visit dearJulius.com to get free premium content for all of your lifestyle needs.
COMMENTS