NERSC

NERSC’s upcoming Doudna supercomputer is poised to make big impacts in scientific computing. To ensure a smooth deployment of this complex system, an Early Access System, to be called Cech, is now installed at NERSC. This smaller system offers NERSC and its partner vendors Dell Technologies, NVIDIA, and VAST Data the opportunity to refine assembly, delivery, installation, and integration processes, paving the way for seamless deployment of the full-scale Doudna system in late 2026.

Learn more: https://bit.ly/4bNMivd

U.S. Department of Energy Berkeley Lab Computing Sciences Berkeley Lab

Berkeley Lab Takes Major Step Toward Doudna with Delivery of Early Access System, Cech The road to NERSC’s Doudna supercomputer reached a key milestone with the delivery of the Early Access System (EAS), to be known as Cech.

NERSC Director Sudip Dosanjh recently sat down with our vendor partner, Dell Technologies, to discuss the promise of the upcoming Doudna supercomputer — not a one-of-a-kind system, but a first-of-a-kind system, setting the stage for breakthroughs across scientific computing.

Learn more: https://ow.ly/Ayk550YtxeF

Berkeley Lab Berkeley Lab Computing Sciences

Thermodynamic noise hinders the efficiency of computing. But what if that noise could be used as a power source instead? Computers powered by thermal noise could dramatically reduce the energy demands of AI and other computing processes.

Using thermodynamic noise to power computers, rather than inhibit them, is the goal of the field of thermodynamic computing, and a collaboration between researchers at the Molecular Foundry and NERSC is bringing it closer to reality for both classical and quantum computers. They’ve proposed a design and training framework for a type of thermodynamic computer that mimics a neural network, which could drastically reduce the energy requirements of machine learning.

Learn more: https://bit.ly/40cPGcA

U.S. Department of Energy Berkeley Lab Computing Sciences Berkeley Lab