91原创

A 91原创 team led by Sunita Chandrasekaran, the David L. and Beverly J.C. Mills Career Development Chair in Computer and Information Sciences and director of the First State AI Institute, is one of through a highly competitive process to stress-test the system and prepare it for scientific use.

鈥淥ur goal is to run thousands of experiments across many different physics scenarios聽 and see what breaks 鈥� whether it鈥檚 software bugs, performance bottlenecks or system issues under stress,鈥� Chandrasekaran said. 鈥淭hat鈥檚 where we learn, and where our team can help fix problems in real time, working directly with the system鈥檚 builders.鈥�

The work is part of ORNL鈥檚 Discovery Center for Accelerated Application Readiness (CAAR), which is preparing cutting-edge scientific applications under the U.S. Department of Energy鈥檚 . The initiative aims to build the world鈥檚 most powerful scientific platform by combining AI, high-performance computing and emerging quantum technologies to tackle complex problems at unprecedented speed.聽

Pressing peak performance聽

Discovery goes beyond today鈥檚 exascale machines, which exceed a quintillion calculations per second. But raw speed isn鈥檛 enough. The system must sustain that performance across complex, AI-driven workloads.

Testing will push every component 鈥� processors, memory, networking and data movement 鈥� to uncover weaknesses and ensure reliability.

鈥淵ou鈥檙e not just asking, 鈥楧oes it run?鈥欌�� Chandrasekaran said. 鈥淵ou鈥檙e asking, 鈥楥an it handle the most demanding science continuously, at scale, without failing?鈥欌��

Once the initial computing cluster is installed later this year, her team will run thousands of simulations using , a software package that models how charged particles, such as electrons and ions, interact in electromagnetic fields.

By using specialized chips called graphics processing units (GPUs) to simulate billions of particles simultaneously, the system will allow researchers to study plasma behavior, with applications in fusion energy, spacecraft propulsion, astrophysics and laser-based medical technologies.

A key challenge in fusion energy 鈥� despite its promise as a clean, nearly limitless energy source 鈥� is efficiently transferring laser energy into the tiny fuel pellets that drive reactions.

To address this, the team will leverage Discovery鈥檚 enhanced performance and memory to pair high-fidelity PIConGPU simulations with AI, identifying optimal fusion target materials and designs.

A rare opportunity for students

The project gives students hands-on experience with next-generation computing.

Nikhil Rao, a doctoral student in computer science at 91原创, is developing workflows that connect plasma simulations with machine learning models, without storing the massive data they generate.

鈥淲e鈥檙e working with simulations that produce data at incredible rates, on the order of petabytes per second,鈥� Rao said. A petabyte 鈥� about 1 million gigabytes 鈥� is enough to store 250 million photos. 鈥淚nstead of storing it, we send the data directly to machine learning models while it鈥檚 still in memory.鈥�

Rao said the opportunity to work on a system still in development is rare.

鈥淰ery few people get to work directly with the hardware vendors and scientists building a system like this,鈥� he said. 鈥淚t鈥檚 quite amazing.鈥�

Rao will collaborate with engineers at AMD and Hewlett Packard Enterprise, along with researchers at Oak Ridge National Laboratory and Helmholtz-Zentrum Dresden-Rossendorf (HZDR) in Germany, a key collaborator on Chandrasekaran鈥檚 large-scale computing efforts.

HZDR previously partnered with her to stress-test Frontier, Oak Ridge鈥檚 current exascale system, and now brings expertise in plasma physics and high-performance computing to help design demanding, scientifically meaningful test scenarios for Discovery, which is expected to surpass Frontier in both speed and efficiency.

鈥淭his work builds on a longstanding collaboration with Sunita and her team,鈥� said Michael Bussmann, plasma physicist and founding manager of the Center for Advanced Systems Understanding at HZDR. 鈥淭heir expertise in high-performance computing and scalable AI has been critical to our participation in this prestigious program. Our success relies on continued international collaboration like this.鈥�

For the United States, Discovery represents a major step forward in high-performance computing and AI. For 91原创 researchers and students, it offers a front-row seat to the future.

鈥淲e鈥檙e preparing for a machine that doesn鈥檛 fully exist yet,鈥� Chandrasekaran said. 鈥淭hat鈥檚 what makes this work so fascinating and rewarding. We鈥檙e not just preparing for what鈥檚 next 鈥� we鈥檙e helping make it possible.鈥�