The DGX-2 Enterprise AI Research System by NVIDIA provides breakthrough computing capacity for machine-learning (ML) and artificial intelligence (AI) algorithms. The DGX-2 system is located within the CNMS building, to enable in-situ data analysis with direct feedback to instruments. Connections to several CNMS instruments enables autonomous steering of scanning probe and electron microscopy experiments, by running resource-intensive algorithms such as Bayesian optimization, Reinforcement Learning and image segmentation and reconstruction algorithms.
Specifications/Capabilities/Extras
• 2 petaFLOPS of computational power
• 16 Nvidia Tesla V100 GPUs with 32GB of memory each (512GB total)
• Dual Intel Xeon Platinum 8168, 2.7 GHz, 24-cores
• 1.5TB system memory
• NVSwitch delivering 2.4TB/s of bisection bandwidth
