深入了解MicrosoftAI硬件创新

AI Hardware & Systems@aiandsystemsInside Microsoft AI hardware innovationMark RussinovichCTO, Deputy CISO and Technical Fellow, Microsoft Azure@markrussinovichMicrosoft AI supercomputer10,000 V100 GPUs#5 supercomputerMay 202014,400 H100 GPUs #3 in TOP500Nov 202330x supercomputersMay 2024AcceleratorsCoolingNetworkingPowerConfidential AICoolingNetworkingAcceleratorsPowerConfidential AIDiverse accelerators on AzureA100, H100(available today)H200, GB200 (coming soon)MI300x(available today)Azure Maia 100(Internal)Maia 100 SpecsChip Size~820mm2@N5Package/Interposer TechnologyTSMC COWOS-SHBM BW/Cap1.8TB/s @64GB HBM2E​Peak Dense Tensor POPS​6bit: 3​9bit: 1.5​BF16: 0.8​L1/L2 ​~500MB​Backend Network BW​600GB/s (12x400gbe)​Host BW (PCIe)​32GB/s PCIe Gen5x8​Design to TDP​700W​Provision TDP ​500W​Maia 100 SpecsInside Maia 100Single tileTile control processorTDMATile data movement engineL1 SRAMTensor unitVector engineInside Maia 100SoCMesh-like NOC topology, with features optimized for MLGreat perf/