Presentation Full Schedule · Contributors · Organizations · Search Program · My Schedule · Happening NowMore…Search ProgramMy ScheduleHappening NowReducing I/O Bottleneck for Pretraining AI Foundation Models for ClimateSessionPDSW24: The 9th International Parallel Data Systems WorkshopDescriptionThe High-Velocity AI Cache (HVAC) mechanism promises to improve I/O performance by 12% while pertaining large foundation models for climate.Author/PresentersGabriele PadovaniUniversity of Trento, ItalyAwais KhanOak Ridge National Laboratory (ORNL)Sandro FioreUniversity of Trento, ItalyValentine AnantharajOak Ridge National Laboratory (ORNL)Event TypeWorkshopTimeSunday, 17 November 202412:05pm - 12:10pm ESTLocationB309TagsData Movement and MemoryI/O, Storage, ArchiveRegistration Categories W Archive view Next PresentationNext PresentationBULKI: Binary Unified Layout for Key-Value InterchangeSimilar PresentationsTraining Large-Scale Vision Transformer Foundation Models for Science and Engineering ApplicationsScalable Multi-Facility Workflows for Artificial Intelligence Applications in Climate ResearchAgents for Climate Change Mitigation and Adaptation in Cities