System Analyzes Atomic Collisions Before You Can Blink

Scientists at Oak Ridge and Berkeley national laboratories have developed a revolutionary data streaming system that could radically transform how nuclear physics experiments are conducted. The software, known as DELERIA, creates a high-speed data pipeline capable of analyzing massive amounts of experimental data almost instantaneously, allowing researchers to make real-time adjustments to their experiments rather than waiting hours or days for results.

This innovative approach connects advanced scientific instruments directly to the nation’s most powerful supercomputers via ultra-fast networks, potentially eliminating the need for expensive local computing infrastructure at research facilities. The system is currently being tested with GRETA, an advanced nuclear physics detector under construction at Berkeley Lab that will eventually be installed at Michigan State University’s Facility for Rare Isotope Beams (FRIB).

But what happens when you need to analyze hundreds of thousands of particle collisions per second while the data must travel more than 2,000 miles? That’s the challenge researchers had to overcome with some creative computing solutions.

Chasing the Speed of Light Across the Country

DELERIA – short for Distributed Event-Level Experiment Readout and Integrated Analysis – enables GRETA to stream data directly to computing centers across the country for near-instant processing. The system transmits information about photon collisions from Berkeley Lab in California to Oak Ridge National Laboratory in Tennessee via the Energy Sciences Network (ESnet).

“Our primary goal is to establish a data pipeline for future experimental needs without having to build a local computing infrastructure. The pipeline will also allow us to increase the amount of analysis that is performed online,” said Gustav Jansen, an ORNL computational nuclear physicist who has been developing DELERIA for the last two years.

In their current testbed, researchers have achieved data transmission rates of 35 gigabits per second—easily surpassing GRETA’s requirements and well within the capabilities of ESnet6, which offers multiple 400 Gbps paths between the labs.

How the System Works

GRETA itself is a marvel of nuclear physics technology—a spherical array of 120 hyper-pure germanium crystals that record photon interactions when experimental samples are bombarded with charged particles. These interactions generate enormous amounts of data that must be processed quickly to be useful to scientists.

“The crystals inside the detector have lots of electrical contacts connected to them. When a photon hits one of the crystals, an electrical signal goes out on each contact point,” Jansen explained. “We need to figure out where inside the crystal the photons hit—the X, Y, Z coordinates. And we need to do that within 10 seconds of the collision. That’s what we designed the pipeline to do.”

The processing happens at ORNL’s Defiant, a 36-node GPU-accelerated computing cluster. Beyond just speed, DELERIA also provides an impressive efficiency boost by reducing data storage requirements by 97.5%, or a factor of 40×.

Key Advantages of the DELERIA Pipeline

• Processes approximately 480,000 photon collision events every second
• Reduces data storage requirements by 97.5%
• Enables real-time experiment adjustments during data collection
• Eliminates need for expensive local computing infrastructure
• Can be adapted for other scientific facilities beyond GRETA
• Transmits data at 35 gigabits per second across 2,000+ miles

Overcoming the Physics of Latency

The biggest challenge for the system is not raw bandwidth but latency—the unavoidable delay caused by the physical distance data must travel. Even at the speed of light, it takes about 120 milliseconds (0.12 seconds) for data to make the round trip from Berkeley to Oak Ridge and back.

This latency is 24 times longer than the 5 milliseconds it takes to process a single photon collision event. To overcome this limitation, the team developed clever parallel processing techniques.

“Of course, only in science fiction can you bypass the speed of light,” Jansen said. “What we’re working on now is to get a 10× speedup by tricking latency, so to speak. Instead of running one event at a time, the solution is to run events in parallel so that we can process one event, while others are being transferred. By finding the right balance, we can make sure that the computing cluster is always busy. Otherwise, our analysis will take 10 times longer than it should.”

When it comes to moving faster than the speed of light, “Only Gustav can do that,” joked Tom Beck, section head for science engagement in ORNL’s National Center for Computation Sciences.

A Model for Future Scientific Computing

DELERIA represents a pioneering approach within the Department of Energy’s Integrated Research Infrastructure (IRI) initiative, which aims to connect major research facilities with leadership-class computing resources.

“This is really pioneering work in the IRI theme that’s setting the standards for how we network the nation’s leading centers of research. And the more use cases we have, the better we’re going to get,” Beck said. “One case at a time. And so far, this is really the first case that’s up and working at this level of maturity that I’m aware of, which means we’re off to a really great start.”

The implications extend far beyond nuclear physics. Jansen emphasized the broader potential of the approach: “We also intend to show that DELERIA can be scaled to work with other research facilities and a broader range of scientific applications.”

As scientific instruments become more sophisticated and generate increasingly massive datasets, this model of distributed computing could become essential for fields ranging from climate science to materials research. With GRETA scheduled for installation in 2026, researchers have time to refine the system further, potentially pushing data processing speeds even higher and establishing a new paradigm for how big science connects with big computing.