Endpoints 2025 - Matt Selensky/Olivia Hull: Using MLPerf for HPC Procurement at NREL скачать в хорошем качестве

Endpoints 2025 - Matt Selensky/Olivia Hull: Using MLPerf for HPC Procurement at NREL

Title: Buying a Supercomputer: Using MLPerf for HPC Procurement at NREL Description: Computational scientists from the National Renewable Energy Laboratory (NREL) reveal how they use MLPerf to ensure their next-generation supercomputer meets the lab's AI needs. Matt Selensky and Olivia Hall discuss why distributed training benchmarks like DeepCAM are critical for choosing hardware that can handle climate simulation and physics-informed AI. Key Topics: The HPC life cycle: From benchmarking procurement suites to system acceptance. Why NREL chose the (retired) DeepCAM benchmark for its next-gen machine. Balancing compute, storage, and network "pillars" in a multi-petaflop system. The value of community-driven, vendor-trusted benchmarks like MLPerf. Speakers: Matt Selensky (Computational Scientist, NREL) Olivia Hall (Computational Scientist, NREL) #Supercomputing #NREL #HPC #MLPerf #ClimateAI #HighPerformanceComputing