FAIRMONT, W.Va. — Scientists and engineers at TMC Technologies of West Virginia have developed a new mini-supercomputer called “Gargantua” and software that will advance the rate at which spaceflight data and software are analyzed.
The Smart Fault Management system is completely unique to TMC and is able to perform hyper-fast “what-if” simulations to assure space flight software operates as safely as efficiently as possible.
“The Smart Fault Management is a new technology capable of visualizing, exploring, and statistically quantifying fault management diagnostic data,” said TMC’s team Chief Scientist and Principal Investigator Dr. Max Spolaor. “Both are a universal tool for big data-driven fault management exploration.”
The project was made possible by a NASA Small Business Innovation Research (SBIR) Phase I grant. TMC is now in the process of applying for a SBIR Phase II proposal to expand upon the results of the developments in Phase I.
TMC Commercial Systems Division Manager Scott Zemerick says Phase II will likely begin in May.
“Phase II is where we take it and we apply it to a real-life scenario that is a real spacecraft that is getting ready to launch. So Phase I was developing it, coming up with the technologies and using technologies,” Zemerick said. “Phase II is actually using it and showing results. We’ve done a little bit of that in Phase I, but Phase II will be a lot more.”
This opportunity arose for TMC’s Independent Test Capability Team because of their success doing simulation work for other NASA missions, Zemerick said.
“We support a lot of NASA missions, and we’ve built simulators for all the NASA spacecraft,” he said. “So this work is kind of an offshoot to that success that we’ve already been doing here for several years now.”
Zemerick said this unique project doesn’t benefit only TMC but West Virginia as a whole.
“We’re very excited, one, to be doing this work in West Virginia, for NASA at the NASA at the I & IV facility,” he said. “This type of research is not normally done in this area or in West Virginia, so we’re really proud that TMC can do this work, to do this research and development, have a chief scientist lead this effort. We’re very excited about doing this for NASA and the area.”
It’s brought jobs to North Central West Virginia, as well.
“This has helped improve our job numbers,” Zemerick said. “We’ve been able to use this to hire some positions. In following work, we hope to hire even more positions as we mature the technology and move it forward.”
WVU Senior M.J. Durst, of Blacksville, is one of those hires. Durst is currently an intern with TMC and will be offered a position with the company upon graduating in May.
Durst had a hand in creating the mini-supercomputer, quite an experience for the Monongalia County native.
“It’s really cool, especially coming from school where everything that we’ve done is well learned and well researched,” he said. “To come out to something like this and be on the front end, you’re just kind of making decisions as you go based on things you know. I’m learning a lot of great information, trying to revolutionize the process.”
Durst explained that part of what makes the Smart Fault Management system so unique is the speed at which it can run.
“For instance, your home computer will run anything between probably 60 and 180 gigaFLOPS (floating-point operations per second),” he said. “This computer, however, is capable of running things that are GPU-optimized at a theoretical maximum of 16 teraFLOPS.”
So far in Phase I, Durst said he’s gotten approximately 10 teraFLOPS out of Gargantua.
“The main thing is that it’s GPU accelerated, not CPU accelerated, which is actually more energy efficient for they’re capable of doing,” he said. “This hardware, it runs four NVIDIA Tesla M60s, which are meant to run virtual desktops for an entire company. But we can use it in a different manner here, through software elements.”
By running Monte Carlo simulation-generated scenarios against the software, NASA missions will be made safer by modeling the probability of different outcomes.
“We generate a lot of data that then gets fed into the machine to train off of, essentially, and through that it gets a lot of very, very high volume, millions of points,” Durst said. “And from that it’s capable of doing statistical analysis and pretty much quantifying the air that machine made up.”
With just months to go until Phase II of the project is underway, TMC’s staff is looking forward to see Gargantua in action.
“I’m very excited to see the project move forward,” Zemerick said. “Max has done a great job developing it and leading it. TMC is very excited to move this forward and move it out of the R&D phase into a more operational and commercial phase.”