Shamu Helps UTSA Faculty, Staff and Students Accelerate Research Projects

November 17, 2017

Written by Danicia Steele, Communications Specialist


Since its latest update in summer 2017, UTSA’s centrally shared high-performance computing (HPC) cluster Shamu, has helped a significant amount of faculty, staff, and students accelerate the completion of their research projects. The update resulted in an enhancement that has benefitted users with better data transfer rates, and increased disk performance and capacity.

Throughout fall 2017, Shamu has improved the performance level of the both the Bexar County Story Map and Password Cracking projects.


Bexar County Story Map

In October 2016, UTSA’s Center for Archaeological Research (CAR) and Department of History received a grant for the Bexar County Story Map Project. The project is designed to produce a series of web maps and applications that present the history of Bexar County from prehistory until 1821. To help make the history lesson more fun, CAR created 3D historical objects to allow viewers to see how the item actually looked as opposed to a photograph.

 3d arrow head model

(3D arrow head model)

The project lead is John Reynolds, Ph.D., a professor in the Department of History, who partnered with Research Scientist Associate I Clinton McKenzie and Research Data Analyst Jessica Nowlin, Ph.D., within the UTSA Center of Archaeological Research team. Throughout the project, the team selected five different prehistoric sites to tell the history of Bexar County from when the Europeans arrived. There are many different concepts in this project including prehistoric people, how land was allotted over time, when Europeans first settled in San Antonio, and much more. 

“This project is a much more accessible way for the general public to get into the history. It's literally right at their doorstep” said Nowlin. “I think it’s really beneficial for UTSA to be involved because we are the biggest university in the county, so we should be included in trying to help the residents understand the history.”


(Display set up area)

During the project, the team used Agisoft PhotoScan to help create the 3D images. The software takes 150 photos of an object and analyzes the common points between all of the photos to construct a 3D model of the artifact. After the pictures are taken, there is a minimum of 40,000 points per model the computer calculates. Before using Shamu, this process would take up to three days to complete. However, Shamu was capable of achieving the process within hours, helping the team reach their deadline of October 2017.

“Working with the Research Computing Support Group was fantastic,” said Nowlin. “I really appreciate that Zhiwei Wang came over to provide a walkthrough of the basic steps for using Shamu during this project.”


Password Cracking Project

In fall 2017, UTSA graduate student Gary Muller completed a research project using both Linux and Windows systems to develop a better understanding of how hackers crack passwords.  The professor provided the class with a combination of eight hashes, a function that allows users to map arbitrarily, for both Windows and Linux systems. 

During the project, Muller decided to use Hashcat, one of the fastest and advanced password recovery tools. While using his personal device, the tool provided a rate of 17,108 hashes per second which is equivalent to about 17,000 password applicants a second. Despite the speed, Muller found himself having to wait hours for the results of cracked passwords on his desktop.

screenshot of Hashcat running on Shamu

(Screenshot of Hashcat running on Shamu)

After hearing of Shamu and its Graphical Processing Units (GPU), Muller decided to take advantage of the system and run the Linux on Shamu. It was then that he discovered his previous results of 17,000 hashes per second increased to 17 million password candidates per second. Seeing that Shamu generated results 1,000 times faster, he decided to use Shamu as his primary source for the remaining research.  

“Not all applications are able to take advantage of GPU processing power, but more and more applications and code are being developed that can take advantage of it,” said Brent League, director of the Research Computing Support Group. “GPUs come in handy when you have large batches of jobs that need to do the same task over and over,  which is exactly what Gary’s experiment was doing and why Shamu’s GPU nodes helped him succeed much quicker than he could have on a typical computer.”

At the end of the project, Muller managed to crack all passwords and stated that Shamu was a big help and provided him with extra time to try different types of attacks.

 “Some of my classmates were struggling to find the time to let their desktop run. They had to run it for three hours to crack only a few passwords,” said Muller. “I completed all of mine in 45 minutes with Shamu and was given the time to experiment using different types of attempts.”


For more information on the Center for Archaeological Research, please click here.