On the cutting edge of supercomputing: How TACC leverages new technology to drive innovation

The rapid adoption of automation through artificial intelligence and other cutting-edge technologies is creating new opportunities. Organizations of all sizes around the world are embracing automation as a way to create significant value.

According to , the Texas Advanced Computing Center is building Stampede3, an Intel-based supercomputing system with 560 new nodes and liquid cooling. tommy minyard (Photo), Director of Advanced Computing Systems at TACC. TACC operates and maintains systems that support scientific research at the University of Texas and other universities, and has plans to incorporate new Intel Vecchio GPUs for experiments.

“It’s exciting. It’s multi-generational. We’re going to run this system for five years, so it’s going to have a pretty long lifespan,” Minyard said. “We are excited about this offer and will also be using some of the new Intel Vecchio GPUs as a kind of experiment in that system.”

Minyard spoke with theCUBE industry analysts John Farrier, lisa martin and Savannah Peterson in SC23, broadcast exclusively on theCUBE, SiliconANGLE Media’s live streaming studio. They discussed how TACC is at the forefront of research in supercomputing, AI, and machine learning, leveraging new technologies and open standards to drive innovation and address data challenges. . (*Disclosure below.)

Scaling your deployment

Minyard explained that AI and high-performance computing intersect, leveraging high-bandwidth interconnects to enable scaling, mainstream adoption in product development, and new access and enablement possibilities. AI and ML are moving towards using proxy models for rapid iteration and optimization. The challenge lies in obtaining clean data for training.

“What’s interesting from our perspective is that a lot of what we’ve supported so far is simulation, where we can model the physics of the world and everything else that we think about through equations,” Minyard said. Told. “What we see in AI is taking surrogate models, small-scale approaches that can be iterated and optimized very quickly towards specific answers and solutions.”

As Arm ecosystem software can now support scientific applications using Grace processors, Minyard explained that TACC is deploying large-scale Arm systems to provide options and diversity to users. did. Open standards are critical for innovation and building new software on top of existing platforms. TACC supports open software and providing source code to the community.

“We’re researching a lot of new technologies and trying to assess what’s going to happen, so we know our users very well,” Minyard said. “One type of system may not work well for them. We want to provide options and variety. AMD has some great platforms and is very good at certain features. It works well. Intel is working on some things. Nvidia GPUs are good for something. Vista carries over all of that technology. This is our first big Arm system. Become.”

The full video interview, part of SiliconANGLE and theCUBE’s coverage, is below. SC23:

(*Disclosure: TheCUBE is a paid media partner of SC23. Neither Dell Technologies Inc., the primary sponsor of theCUBE’s event coverage, nor any other sponsors have editorial control over theCUBE or SiliconANGLE content.)

Photo: SiliconANGLE