- 1、本文档被系统程序自动判定探测到侵权嫌疑,本站暂时做下架处理。
- 2、如果您确认为侵权,可联系本站左侧在线QQ客服请求删除。我们会保证在24小时内做出处理,应急电话:400-050-0827。
- 3、此文档由网友上传,因疑似侵权的原因,本站不提供该文档下载,只提供部分内容试读。如果您是出版社/作者,看到后可认领文档,您也可以联系本站进行批量认领。
TSMCxNvidia:BreakingtheThermalWall
HowAdvancedCoolingIsPoweringthe
FutureofComputing
台积电xNvidia:突破热壁:先进冷却技术如何驱
动未来计算
OriginalArticlebySemiVisionResearch(TSMC,NVIDIA,Coherent,Jentech,AVC,
Auras,CoolerMaster,MikrosTechnologies,Inspur,Invek,NingboJingda,
Fabric8Labs,Intel,xMEMS)
SEMIVISION
OCT05,2025·PAID
175Share
ThedevelopmentofAIchipsisfundamentallydrivenbythepursuitofhigher
performance,butthisalsobringsacriticalchallenge:thermalmanagement.
Effectivelydissipatingtheenormousheatgeneratedbyrisingpower
consumptionhasbecomeakeydesignconsiderationfornext-generationAI
chiparchitectures.
Onthesystemlevel,theintegrationofopticalenginesiswidelyviewedasthe
futuresolutiontoovercomehigh-bandwidthinterconnectbottlenecks.
However,opticalsignalsarehighlysensitivetotemperaturefluctuations—even
slightdeviationscanleadtotransmissionlossanddegradedperformance.This
iswhyTSMChasadoptedtheMicroringModulator(MRM)approach,butMRM
itselfhasstrictthermaloperatingrequirements.Asaresult,companieslike
TSMCandNVIDIAareactivelyexploringnext-generationcoolingtechnologies
tomeetthesethermaldemands.
Onthelogicprocessside,TSMC’sevolutionfromN3→N2→A16involves
morethangeometricscaling—itrepresentsatransistorarchitecturetransition
fromFinFET,toGAA(Gate-All-Around),andfinallytotheSuperPowerRail
architectureintheA16generation.Thisseriesofchangesisaimedatachieving
theoptimalbalanceofPPA(Power,Performance,andArea),layingthe
foundationforhigh-efficiencyAIcomputing.
However,fromapackagingandmaterialsperspective,newchallengesemerge.
TraditionalcoolingdependsonTIM(ThermalInterfaceMaterials),whose
materialselectionislimited,andthermalreliabilityisconstrainedwithinspecific
文档评论(0)