Lawrence Livermore National Laboratory (LLNL) and SambaNova Systems will equip LLNL’s Livermore Computing Center with SambaNova’s Spatial Data Flow Accelerator as part of an effort to upgrade the Institute’s Cognitive Simulation (CogSim) program. announced that it was added
LLNL is integrating new hardware to develop a CogSim approach that combines artificial intelligence (AI) and high performance computing (HPC), as well as a deep neural network hardware architecture as part of the National Nuclear Security Administration (NNSA). We further investigate how traditional physics-based simulations can be accelerated. Advanced simulation and computing programs. The lab will use the SambaNova AI system to improve model fidelity and manage growing data volumes to improve overall speed and performance of stockpile management applications, fusion energy research, and other basic science work. , is expected to improve productivity.
“Multiphysics simulations are complex. Our inertial confinement fusion (ICF) experiments generate vast amounts of data,” said Brian Van Essen, LLNL informatics group leader. “Connecting physics to experimental data is a very difficult scientific challenge.” ”
In 2020, LLNL integrated SambaNova Systems’ DataScale AI accelerators into NNSA’s Corona supercomputing cluster of over 11 petaFLOP machines. This has enabled scientists to use his AI computations to more quickly and efficiently conduct research on stockpile management and inertial confinement fusion applications. Also about research related to new coronavirus infections. The system is funded by his ASC program and is part of an agreement between the Department of Energy (DOE) and SambaNova Systems to accelerate AI within the DOE National Laboratory.
The LLNL researchers believe that the next phase of their collaboration will allow heterogeneous systems to be more loosely coupled with the lab’s supercomputing clusters, supporting a wider range of workloads and allowing LLNL to use more traditional resources. It states that it will be widely used, resulting in a more generalized solution. Expand the possible use cases.
“We are looking at using AI to improve speed, energy usage and data movement,” said Bronis R. de Supinski, chief technology officer at Livermore Computing. We have a different architecture and we are leveraging that.” Create a CogSim enhanced approach that leverages heterogeneous systems combining SambaNova DataScale with our supercomputing clusters. ”
SambaNova Systems CEO Rodrigo Liang added: “Scientific discovery relies on speed, accuracy and collaboration. We have built great partnerships to solve complex scientific problems. — together. We are excited to leverage the agility, power, and flexibility of SambaNova to help Lawrence Livermore National Laboratory achieve its scientific mission in the national interest. ”
