Lincoln Lab unveils essentially the most powerful AI supercomputer at any US university

The brand new TX-Generative AI Next (TX-GAIN) computing system on the Lincoln Laboratory Supercomputing Center  (LLSC) is essentially the most powerful AI supercomputer at any U.S. university. With its recent rating from  TOP500, which biannually publishes an inventory of the highest supercomputers in various categories, TX-GAIN joins the ranks of other powerful systems on the LLSC, all supporting research and development at Lincoln Laboratory and across the MIT campus. 

“TX-GAIN will enable our researchers to attain scientific and engineering breakthroughs. The system will play a big role in supporting generative AI, physical simulation, and data evaluation across all research areas,” says Lincoln Laboratory Fellow Jeremy Kepner, who heads the LLSC. 

The LLSC is a key resource for accelerating innovation at Lincoln Laboratory. Hundreds of researchers tap into the LLSC to research data, train models, and run simulations for federally funded research projects. The supercomputers have been used, for instance, to simulate billions of aircraft encounters to develop collision-avoidance systems for the Federal Aviation Administration, and to coach models within the complex tasks of autonomous navigation for the Department of Defense. Through the years, LLSC capabilities have been essential to quite a few award-winning technologies, including those who have improved  airline safety,  prevented the spread of recent diseases, and  aided in hurricane responses. 

As its name suggests, TX-GAIN is very equipped for developing and applying generative AI. Whereas traditional AI focuses on categorization tasks, like identifying whether a photograph depicts a dog or cat, generative AI produces entirely recent outputs. Kepner describes it as a mathematical combination of interpolation (filling within the gaps between known data points) and extrapolation (extending data beyond known points). Today, generative AI is widely known for its use of huge language models to create human-like responses to user prompts. 

At Lincoln Laboratory, teams are applying generative AI to varied domains beyond large language models. They’re using the technology, as an illustration, to judge radar signatures, complement weather data where coverage is missing, root out anomalies in network traffic, and explore chemical interactions to design recent medicines and materials.

To enable such intense computations, TX-GAIN is powered by greater than 600 NVIDIA graphics processing unit accelerators specially designed for AI operations, along with traditional high-performance computing hardware. With a peak performance of two AI exaflops (two quintillion floating-point operations per second), TX-GAIN is the highest AI system at a university, and within the Northeast. Since TX-GAIN got here online this summer, researchers have taken notice. 

“TX-GAIN is allowing us to model not only significantly more protein interactions than ever before, but additionally much larger proteins with more atoms. This recent computational capability is a game-changer for protein characterization efforts in biological defense,” says Rafael Jaimes, a researcher in Lincoln Laboratory’s Counter–Weapons of Mass Destruction Systems Group. 

The LLSC’s deal with interactive supercomputing makes it especially useful to researchers. For years, the LLSC has pioneered software that lets users access its powerful systems with no need to be experts in configuring algorithms for parallel processing.  

“The LLSC has at all times tried to make supercomputing feel like working in your laptop,” Kepner says. “The quantity of information and the sophistication of research methods needed to be competitive today are well beyond what could be done on a laptop. But with our user-friendly approach, people can run their model and get answers quickly from their workspace.”

Beyond supporting programs solely at Lincoln Laboratory, TX-GAIN is enhancing research collaborations with MIT’s campus. Such collaborations include the Haystack Observatory, Center for Quantum Engineering, Beaver Works, and Department of Air Force–MIT AI Accelerator. The latter initiative is rapidly prototyping, scaling, and applying AI technologies for the U.S. Air Force and Space Force, optimizing flight scheduling for global operations as one fielded example.

The LLSC systems are housed in an energy-efficient data center and facility in Holyoke, Massachusetts. Research staff within the LLSC are also tackling the immense energy needs of AI and leading research into various power-reduction methods. One software tool they developed can reduce the energy of coaching an AI model by as much as 80 percent.

“The LLSC provides the capabilities needed to do leading-edge research, while in a cheap and energy-efficient manner,” Kepner says.

The entire supercomputers on the LLSC use the “TX” nomenclature in homage to Lincoln Laboratory’s Transistorized Experimental Computer Zero (TX-0) of 1956. TX-0 was one in all the world’s first transistor-based machines, and its 1958 successor, TX-2, is storied for its role in pioneering human-computer interaction and AI. With TX-GAIN, the LLSC continues this legacy.