Doudna System

Doudna System poster art 1025 x 685

NERSC’s next flagship supercomputer is scheduled to be delivered in late 2026.

This system will be designed for a new era that requires the integration of advanced capabilities — the convergence of simulations and modeling, advances in AI, and an explosion of experimental and observational data — to enable new modes of scientific discovery.

The new system will be named in honor of Jennifer Doudna, the Berkeley Lab-based biochemist who was awarded the 2020 Nobel Prize for Chemistry for her work on the gene-editing technology CRISPR.

Doudna will be a key tool for research across the DOE Office of Science mission space, powering scientific discovery in areas like fusion energy, advanced materials design, fundamental physics, biomolecular design, quantum computing, and more.

This state-of-the-art platform is expected to provide at least 10x the performance of Perlmutter to accelerate time to solution for large-scale simulations and training scientific AI models.

Doudna will also support quantum simulation tools, including NVIDIA’s CUDA-Q platform, to allow scalable quantum algorithm development, modeling, and verification of quantum computers at scale, and codesign of future integrated quantum-HPC systems.

Next-generation technology

Doudna will be built using cutting-edge technologies:

  • Dell Integrated Rack Scalable Systems and PowerEdge servers
  • Dell ORv3 direct liquid-cooled server technology
  • NVIDIA Vera-Rubin CPU-GPU platform
  • High-speed Quantum-X800 InfiniBand networking platform

The dense packaging enabled by direct-cooling technology, combined with state-of-the-art processors, will deliver next-generation AI performance in the smallest possible physical footprint, accelerating time-to-solution for large-scale simulations and training scientific AI models.

Pushing the boundaries of scientific computing

The Doudna supercomputing system is expected to push the boundaries of scientific computing:

  • Users will be able to configure and program its capabilities in ways that work best for their research projects.
  • Doudna will support urgent and interactive workflows from DOE SC’s experimental and observational facilities.
  • The system will provide seamless performance across converged networks, accelerators, and storage components.
  • Doudna will leverage container technologies to create a highly customizable environment for a wide variety of science use cases.
  • Integrating container technologies will, in turn, enable new compute, storage, and software to be easily integrated into a comprehensive, extensible HPC system for science.