NextSilicon has announced that the Spectra supercomputer at Sandia National Laboratories has achieved full system acceptance under Sandia’s Vanguard program, marking a significant milestone in the evaluation of next-generation high-performance computing architectures for national security applications.
Spectra is the second system deployed under the Vanguard program. The platform is designed to evaluate emerging computing technologies against mission-critical workloads before broader deployment decisions are made.
The system comprises 64 compute nodes equipped with 128 Maverick-2 dual-die accelerators developed by NextSilicon. Spectra was built through a collaboration involving Sandia National Laboratories, NextSilicon, and Penguin Solutions.
The system supports the National Nuclear Security Administration’s Advanced Simulation and Computing (ASC) program. The Vanguard initiative also involves Lawrence Livermore National Laboratory and Los Alamos National Laboratory.
According to NextSilicon, the acceptance process required Spectra to demonstrate performance, system stability, and compatibility across multiple scientific and engineering workloads.
During the evaluation, the system executed workloads including HPCG, LAMMPS molecular dynamics simulations, and SPARTA Monte Carlo simulations. These applications are commonly used in high-performance computing environments for scientific modeling and large-scale simulations.
Unlike traditional GPU accelerators, Maverick-2 uses a runtime-reconfigurable architecture that adapts computational resources dynamically to different applications during operation.
Elad Raz, founder and CEO of NextSilicon, said the acceptance process demonstrated Maverick-2’s ability to execute mission-relevant workloads while meeting the operational requirements of a national security computing environment.
He said the company’s architecture was designed to deliver computing performance while reducing power consumption, an increasingly important factor in large-scale HPC infrastructure.
The Vanguard program serves as Sandia’s primary framework for evaluating emerging computing architectures under real operational conditions. Spectra is only the second platform accepted into the program.
The first Vanguard deployment was Astra, introduced in 2018, which evaluated Arm-based processors for HPC applications.
James H. Laros III, Senior Scientist and Vanguard Program Lead at Sandia National Laboratories, said the program is intended to rigorously test new architectures using workloads directly tied to Sandia’s mission requirements.
He stated that the partnership successfully completed all benchmarks and application tests required for this phase of the evaluation process.
Phil Pokorny, Chief Technology Officer at Penguin Solutions, said the project provided an opportunity to evaluate NextSilicon’s accelerator technology at scale within one of the most demanding HPC testing environments in the United States.
He added that the Spectra cluster could support further research into accelerating mixed physics and AI workloads.
High-performance computing systems are increasingly being evaluated for their ability to support AI workloads alongside traditional scientific simulations. Power efficiency, scalability, and workload adaptability have become key focus areas as organizations assess alternatives to conventional GPU-based architectures.
NextSilicon said Maverick-2 accelerators are currently deployed at dozens of customer sites globally. The successful acceptance of Spectra extends the company’s presence into one of the most closely evaluated government HPC programs in the United States.
The milestone also provides a new reference point for commercial and government organizations evaluating future HPC infrastructure for scientific computing, AI, and national security applications.
