欢迎参加未来技术研讨会.pdf

1、Dr Jon Summers,Chair of Future Technologies SymposiumScientific Lead in Data Centers at Research Institutes of SwedenAdjunct Professor in Fluid Mechanics at Lule Technical University,SEVisitingProfessor in Thermofluidsat University of Leeds,UKWelcome to the Future Technologies Symposium08:00 08:50 F

2、TS:Keynotes Chair,Invited,and Sponsors08:55 09:30 FTS:AI/HPC 2 presentations09:30 09:45 FTS:POSTER SESSION BREAK09:45 10:10 FTS:AI/HPC 5 Lightning Talks10:15 10:50 FTS:Data Center Sustainability 2 presentations10:55 11:20 FTS:Data Center Sustainability 5 Lightning Talks11:25 11:30 FTS:Awards Ceremon

3、y12:30 17:00 FTS:QuantumFuture Technologies Symposium scheduleMAINAdditionally,FTS giveawaysThanks to the FTS Global 2025 sponsorsDr Jon SummersFTS Chair,Scientific Lead in Data Centers,Research Institutes of SwedenSome thoughts on energy efficiency of the computing processSketches and thoughts on t

4、he computing processRichard Feynman had this to say in the 1980sFrom the data center perspective HPC has used floating point operations per jouleMeasure of compute throughput and efficiency/=UltimatelyThe watt includes the power of the infrastructurethat wraps around the compute.For irreversible log

5、icMinimal thermal energy dissipation2 configurationstates1 configurationstateEntropy=kBx log(2)=0.69 x kBEntropy=kBx log(1)=0HEATNo change in internal energy=work done in the computing process of one bit generates heat.Second law of thermodynamics=thermal energy is(change in Entropy)x process temper

6、ature.Change in Entropy=(0.69 0)x kB,so thermal energy per bit=0.69 x kBx T=0.000000000000000000003 joules per bit=3 zJ(zepto joules)at around 40C(313.15 K)Boltzmann constant,kB=1.3806491023JK1Energy efficiency of irreversible computeThomas N.Theis,H.-S.Philip