For thousands of Minnesota Supercomputing Institute users, getting enough time with its five “official” supercomputers is a competitive process.

Researchers at the University of Minnesota and other institutions use the supercomputers to process large and complex programs, tests and data sets. Projects can range from visualizing biological processes like protein folding to bringing ancient Greece to life.

MSI staff approve requests for smaller projects, but the High Performance Computing Allocation Committee has the task of assigning supercomputer usage hours of more than 50,000 service units, which represent units of time, to competing research groups.

Jiali Gao, chair of the committee, said requests they receive are “significantly” more than the available time.

 “Most people don’t get all the computer time they requested,” he said.

Currently, more than 550 principal investigators’ research groups have access to MSI’s resources, which translates to more than 3,000 people.

Allocation decisions

When it comes to deciding which research group gets service units over another, Gao said the committee considers a number of factors.

The most important one, he said, is the nature of the research. If it is cutting-edge research that would especially benefit from more computing power, then it is more likely to have its request approved.

Next, Gao said they evaluate whether the investigator has research support, “especially from established professors.”

Finally, MSI requires that investigators submit reports when they have published scientific papers using supercomputing support. If they have not done their paperwork in a timely manner, Gao said, they are less likely to be approved than a competing investigator who has.

Computing resources

In 1981, the University became the first academic institution to buy a supercomputer and much has changed since then. Today, MSI’s supercomputers have a combined 128 million core computing hours.

Each supercomputer has a different number of core hours per service unit based on its computing power, processing speed and memory.

Itasca, MSI’s most powerful supercomputer, was rated the 59th most powerful supercomputer in the world when it debuted in 2009. As supercomputers have advanced over the years, Itasca dropped to the 346th most powerful in 2012.

There will soon be four supercomputers when Elmo is transitioned to run smaller jobs. Bill Hellriegel, assistant to the MSI director, said Elmo’s current functionality will be replaced with a new, permanent addition to Itasca.

But the “official” supercomputers are not all MSI has to offer. Josh Baller, a doctoral researcher in genetics, cell biology and development, said MSI’s smaller systems are all he needs.

“The available resources outside of the actual supercomputers, for most things, are more than enough,” Baller said.

These systems are multiple computers all networked together, he said, that researchers do not need special approval to use.

Users can access them either in the lab or from their own computers. They simply sign in, specify how long they need their program to run and what resources they need, and they automatically go into a queue.

When the program has finished running, the computer then emails the user to let him or her know. Baller said getting a time slot can take anywhere from no time at all to half an hour.

He also said some times of the day are busier than others, so if he needs to do a longer program, he lets it run overnight, and it’s done by the next morning.

But the programs Baller needs to run are not as complex as the ones that run on the “official” supercomputers.

A state resource

Since the MSI is not only a University of Minnesota institute — it is just housed on campus — every institution and university in Minnesota can gain access.

John Abraham, an associate engineering professor at the University of St. Thomas, said he has always received all the service units he has requested because he has never asked for more than several thousand.

Molly McCue, assistant professor in the University’s College of Veterinary Medicine, said she is grateful for MSI.

“Compared to what a lot of my colleagues at other institutions have, there’s a lot more resources here as far as computing,” she said.r