AI-HPC - The Future of AI-ML Innovation is Row-Scale Disaggregation.pdf

《AI-HPC - The Future of AI-ML Innovation is Row-Scale Disaggregation.pdf》由会员分享，可在线阅读，更多相关《AI-HPC - The Future of AI-ML Innovation is Row-Scale Disaggregation.pdf（10页珍藏版）》请在三个皮匠报告上搜索。

1、OCP Global Summit October 18,2023|San Jose,CAPresented by Matthew Williams,CTO,Rockport Networks Now CerioThe Future of AI/ML Innovation Is Row-Scale DisaggregationAcceleration and memory are key for AI/ML innovation,growth and profitabilityDecades-old monolithic system model traps GPUs and other re

2、sources inside serversThe Problem:Closed ecosystems,low GPU utilization,and operational complexity at scaleTraditional Systems GPU“Capacity Trap”Open Systems GPU DisaggregationUnderlay FabricServer Pool(100s)Device Pool(100s)GPUGPUNVMeGPUNVMeTPUPCIePCIeDevice Enclosures(100s of devices)Servers(100s)

3、FabricNodeFabricNodeFabricNodeFabricNodeFabricNodeFabricNodeFabricNodeSHFLsGPUGPUNVMeGPUNVMeTPUGPUGPUNVMeGPUNVMeTPUGPUGPUNVMeGPUNVMeTPULogicalPhysicalFabric ManagerDiscovery,PolicyIT Service Management&OrchestrationAny device from any vendor best fit Commodity componentsLinear scale,highly resilient

4、Software-based repairOpen Systems GPU DisaggregationUnderlay FabricServer Pool(100s)Device Pool(100s)GPUGPUNVMeGPUNVMeTPUPCIePCIeDevice Enclosures(100s of devices)Servers(100s)FabricNodeFabricNodeFabricNodeFabricNodeFabricNodeFabricNodeFabricNodeSHFLsGPUGPUNVMeGPUNVMeTPUGPUGPUNVMeGPUNVMeTPUGPUGPUNVM

5、eGPUNVMeTPULogicalPhysicalFabric ManagerDiscovery,PolicyIT Service Management&OrchestrationAny device from any vendor best fit Commodity componentsLinear scale,highly resilientSoftware-based repairPCIe ServicesDevice CompositionEthernet ServicesLayer 2 SwitchingCXL ServicesAdvanced MemoryOverlayServ

6、ices(Logical)Adaptation ServicesResources.Optical Interconnect(Physical)ComposableServices(Logical)Capacity Calibration Deadlock-Free RoutingLink ReliabilityFLIT SwitchingTopology DiscoveryUnderlayFabric(Physical)Adaptive MultipathUltra High PrioritySoftware-based RepairApplicati