Serving technology enthusiasts for more than 25 years. TechSpot is the place to go for tech advice and analysis you can trust.
Exactly what just happened? Plans for a new supercomputer to accelerate research in a variety of scientific fields have been announced by the Department of Energy. The initiative highlights the convergence between commercial AI and the computational requirements of cutting-edge science discovery.
This advanced system will be housed at Lawrence Berkeley National Laboratory and is scheduled to become operational by 2026. It will be named “Doudna” after Nobel laureate Jennifer Doudna whose groundbreaking research on CRISPR gene-editing has revolutionized molecular biological.
Dell Technologies was selected to deliver the Doudna Supercomputer, marking a major shift in the landscape for government-funded high performance computing.
While Hewlett Packard Enterprise has traditionally dominated the high-performance computing market, Dell’s bid signals a new chapter. In an interview with The New York Times “A big win for Dell,” Addison Snell of Intersect360 Research noted the company’s limited presence in the field.
Dell executives explained the Doudna Project enabled them to move past the long-standing practice of building customized systems for individual labs. Instead, they focused their efforts on developing a platform that could serve a wide range of users. “This market had shifted into some form of autopilot. What we did was disengage the autopilot,” Paul Perez, senior Vice President and Technology Fellow at Dell.
Doudna’s use of Nvidia Vera Rubin platform will be a defining feature. It was designed to combine the strength of traditional scientific simulations and the power of modern AI. Doudna, unlike previous Department of Energy supercomputers that relied on processors made by AMD or Intel, will use a general-purpose Arm CPU from Nvidia paired with Rubin AI chips specifically designed for artificial intelligence and simulation workloads.
This architecture is designed to meet the demands of the 11,000 users at the laboratory, who are increasingly dependent on high-precision data modeling and rapid AI analysis. Jensen Huang is the founder and CEO of Nvidia. He described the new system enthusiastically. He added that the system will allow “scientists delve deeper and think bigger to seek the fundamental truths of the universe.”
Doudna to be over ten times faster than the lab’s current flagship. Jonathan Carter, Berkeley Lab’s associate lab director for computing science, said that the system’s design was shaped by researchers who are using AI to enhance simulations in areas such as geothermal energy or quantum computing.
Doudna’s design reflects an overall shift in supercomputing. The traditional systems prioritize 64-bit calculations to achieve maximum numerical accuracy. However, modern AI workloads benefit from lower precision operations (such 16-bit or eight-bit) which enable faster processing speeds. Dion Harris, Nvidia’s head of data centre product marketing, said that the ability to combine different levels precision opens up new frontiers in scientific research.
This supercomputer is also tightly integrated with Energy Sciences Network. Researchers from around the world can stream data directly into Doudna to perform real-time analyses. Sudip Dosanjh of the National Energy Research Scientific Computing Center described the new system “designed to accelerate a broad set of scientific workflows.”
