NVIDIA today announced the NVIDIA Vera CPU rack-scale platform for high-performance computing — combining NVIDIA’s first fully custom data center CPU with the gold standard for HPC networking to drive extreme efficiency and throughput for state-of-the-art scientific modeling, simulation and agentic science.
The rack-scale platform brings together 192 Vera CPUs interconnected by NVIDIA Quantum-X800 InfiniBand networking to meet the massive performance demands of climate modeling, molecular dynamics and other unaccelerated, next-generation scientific workloads.
The liquid-cooled system delivers more than 200TB/s of aggregate peak memory bandwidth to support scientific simulation and AI workflows.
“Scientific simulations need CPUs that can keep large datasets moving and sustain performance as applications scale,” said Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA. “Vera brings custom CPU cores, high-bandwidth memory and InfiniBand networking together to address those bottlenecks. With compilers and libraries optimized for Vera, researchers can run demanding simulations more efficiently and explore more complex scientific problems.”
Engineered for Memory-Intensive HPC
Each Vera CPU features 88 custom NVIDIA Olympus cores, 176 threads and compact, serviceable SOCAMM2 LPDDR5X memory delivering up to 1.2TB/s of bandwidth to accelerate data-intensive scientific applications.
The rack combines 192 Vera CPUs with approximately 300TB of memory and 1 petabyte of storage in a fully liquid-cooled NVIDIA MGX™ design. Each 1U tray integrates eight Vera CPUs, and NVMe storage and management.
The platform reuses NVL72 rack components, including chassis, power distribution and cooling infrastructure, to simplify deployment and service. NVIDIA Quantum-X800 InfiniBand networking connects the system, and NVIDIA SHARP™ In-Network Computing helps scientists finish complex simulations faster by reducing the time processors spend coordinating their work.
Source link







