The University of Lethbridge is one of the latest universities to reap the benefits of high performance computing, courtesy of a cluster from Dell Canada that has pumped up the computing power of the institution’s chemistry calculations.
Stacey Wetmore, a Canada Research Chair and associate professor with the Alberta-based University of Lethbridge, said that when she was looking for a cluster vendor, “the most important thing was the amount of computing power we could get — Dell provided the best solution with their quad core processors and their robustness far outbid everyone else.” (The University of Lethbridge already had allegiance to Dell over on the user side. The company provides the school with all-new student computers every year.)
The university went with what Dell Canada server brand manager Kevin Smith said is a growing trend – vendor-sponsored implementation. “It’s more and more common,” said Smith. “Researchers used to build the clusters themselves, but now companies are getting to act as that resource.” Said Wetmore, “You don’t see as many people standing up at conferences and saying how they got their cluster to work!”
The university now has a powerful solution that has been in place since January. The cluster, called Uracil (after Upscale and Robust Abacus for Chemistry in Lethbridge), is composed of 85 computer nodes, each with two quad-core processors, a 146GB SCSI drive, and at least 4GB of RAM. The cluster will be used for computational chemistry, according to Wetmore, who said that she and her undergraduate and graduate students are looking at ways in which DNA is damaged, such as the effects of UV rays, and how the body tries to repair the damage so that they can aid in developing drugs that mimic this effect.
“The undergraduate and graduate students set up calculations on the Dell workstations within the lab. The advantage is that you can actually make it bigger, see more atoms, or be more accurate,” she said. Before acquiring Uracil, according to Wetmore, one calculation (done on one workstation) could take a week, and what previously took one graduate two months to compute took a week-and-a-half.
Smith said that cluster computing is becoming increasingly common at the university level.
“It’s getting really important,” Wetmore said, citing the rise in high performance computing university networks. These types of networks and clusters at the academic level will also help foster Canadian innovation, as it will help get students acclimatized to the high performance computing environment for their own future research, she said. Wetmore has also found that this is being helped along by the increasingly user-friendly interfaces, which her students found easy to maneuver.