본문

서브메뉴

Energy-Performance Tradeoffs in Data Centers and Machine Learning
Energy-Performance Tradeoffs in Data Centers and Machine Learning
Energy-Performance Tradeoffs in Data Centers and Machine Learning

상세정보

자료유형  
 학위논문 서양
최종처리일시  
20260209102915
ISBN  
9798265427021
DDC  
001
저자명  
Mann, Ariana Joy.
서명/저자  
Energy-Performance Tradeoffs in Data Centers and Machine Learning
발행사항  
[Sl] : Stanford University, 2023
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2023
형태사항  
218 p
주기사항  
Source: Dissertations Abstracts International, Volume: 87-05, Section: A.
주기사항  
Advisor: Bambos, Nicholas.
학위논문주기  
Thesis (Ph.D.)--Stanford University, 2023.
초록/해제  
요약The ability to scale computing over the coming decade is limited by power and energy: whether that be the power limits of individual processor chips due to heating effects, the ability of the power grid to supply a certain amount of power relative to the massive demand of hyperscale data centers, or the growing emphasis on reducing the carbon impact of computing. All of these problems benefit not only from revolutionary technological improvements, but also from fine-grained power consumption control.In this thesis, a general three tiered approach to control the performance-energy tradeoff in computing is presented, which combines tools from stochastic modeling, dynamic programming, and operating systems. A concrete instance of this approach is developed for the processor speed control setting, where a higher speed improves a program's slowdown performance at the convex cost of unit energy or power. The first tier is the low level control algorithm. For the modern slowdown performance metric, I show that the processor-queue's state must be modeled with its underlying multilevel structure otherwise the control policy will be sub-optimal even in expectation. To avoid the complexity of the complete multi-level policy solution, I develop an approximate control policy that accounts for the multi-level state and functions under any scheduling policy. The second tier is a meta-algorithm that leverages the first tier's control policy to achieve a particular target. It functions for a wide range of particular performance or energy targets, which itself enables increased robustness to parameter misestimation. Of particular interest are the 5-minute and 1-hour average power targets of data centers' power purchase agreements. The final tier adjusts the reconfiguration frequency of the computations required for the lower tiers and the estimation of system parameters, in order to tradeoff between the algorithms' computational overhead and the control accuracy.While in some settings like processor speed control a clear performance-energy tradeoff is possible, in modern machine learning (and deep neural networks in particular) large gains can be achieved by actually redesigning the training algorithms themselves for energy efficiency. Distributed Distillation (D-Dist) is one example of this in the small, power-limited devices setting. Presented in this thesis, D-Dist achieves a 10,000x reduction in the power-hungry communication required for distributed on-device training compared to the vanilla distributed stochastic gradient decent algorithm. Both of these approaches (improved algorithms and tradeoff control policies) can be combined to produce even further improvement. Today the current use of large-language-models is limited by inference time and expense; to begin to address this and demonstrate how these two approaches can be combined, in the final chapter, the application of the performance-energy tradeoff management stack is outlined for deep neural network inference-stage batch-size control.
일반주제명  
Software
일반주제명  
Dynamic programming
일반주제명  
Cooling
일반주제명  
Neural networks
일반주제명  
Energy management
일반주제명  
Hard disks
일반주제명  
Energy efficiency
일반주제명  
Alternative energy sources
일반주제명  
Energy resources
일반주제명  
Transistors
일반주제명  
Cloud computing
일반주제명  
Energy consumption
일반주제명  
Alternative energy
일반주제명  
Electrical engineering
일반주제명  
Sustainability
기타저자  
Stanford University.
기본자료저록  
Dissertations Abstracts International. 87-05A.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008260203s2023        us                              c    eng  d
■001000017366021
■00520260209102915
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798265427021
■035    ▼a(MiAaPQ)AAI32316407
■035    ▼a(MiAaPQ)Stanfordqz590zt0478
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a001
■1001  ▼aMann,  Ariana  Joy.
■24510▼aEnergy-Performance  Tradeoffs  in  Data  Centers  and  Machine  Learning
■260    ▼a[Sl]▼bStanford  University▼c2023
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2023
■300    ▼a218  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  87-05,  Section:  A.
■500    ▼aAdvisor:  Bambos,  Nicholas.
■5021  ▼aThesis  (Ph.D.)--Stanford  University,  2023.
■520    ▼aThe  ability  to  scale  computing  over  the  coming  decade  is  limited  by  power  and  energy:  whether  that  be  the  power  limits  of  individual  processor  chips  due  to  heating  effects,  the  ability  of  the  power  grid  to  supply  a  certain  amount  of  power  relative  to  the  massive  demand  of  hyperscale  data  centers,  or  the  growing  emphasis  on  reducing  the  carbon  impact  of  computing.  All  of  these  problems  benefit  not  only  from  revolutionary  technological  improvements,  but  also  from  fine-grained  power  consumption  control.In  this  thesis,  a  general  three  tiered  approach  to  control  the  performance-energy  tradeoff  in  computing  is  presented,  which  combines  tools  from  stochastic  modeling,  dynamic  programming,  and  operating  systems.  A  concrete  instance  of  this  approach  is  developed  for  the  processor  speed  control  setting,  where  a  higher  speed  improves  a  program's  slowdown  performance  at  the  convex  cost  of  unit  energy  or  power.  The  first  tier  is  the  low  level  control  algorithm.  For  the  modern  slowdown  performance  metric,  I  show  that  the  processor-queue's  state  must  be  modeled  with  its  underlying  multilevel  structure  otherwise  the  control  policy  will  be  sub-optimal  even  in  expectation.  To  avoid  the  complexity  of  the  complete  multi-level  policy  solution,  I  develop  an  approximate  control  policy  that  accounts  for  the  multi-level  state  and  functions  under  any  scheduling  policy.  The  second  tier  is  a  meta-algorithm  that  leverages  the  first  tier's  control  policy  to  achieve  a  particular  target.  It  functions  for  a  wide  range  of  particular  performance  or  energy  targets,  which  itself  enables  increased  robustness  to  parameter  misestimation.  Of  particular  interest  are  the  5-minute  and  1-hour  average  power  targets  of  data  centers'  power  purchase  agreements.  The  final  tier  adjusts  the  reconfiguration  frequency  of  the  computations  required  for  the  lower  tiers  and  the  estimation  of  system  parameters,  in  order  to  tradeoff  between  the  algorithms'  computational  overhead  and  the  control  accuracy.While  in  some  settings  like  processor  speed  control  a  clear  performance-energy  tradeoff  is  possible,  in  modern  machine  learning  (and  deep  neural  networks  in  particular)  large  gains  can  be  achieved  by  actually  redesigning  the  training  algorithms  themselves  for  energy  efficiency.  Distributed  Distillation  (D-Dist)  is  one  example  of  this  in  the  small,  power-limited  devices  setting.  Presented  in  this  thesis,  D-Dist  achieves  a  10,000x  reduction  in  the  power-hungry  communication  required  for  distributed  on-device  training  compared  to  the  vanilla  distributed  stochastic  gradient  decent  algorithm.  Both  of  these  approaches  (improved  algorithms  and  tradeoff  control  policies)  can  be  combined  to  produce  even  further  improvement.  Today  the  current  use  of  large-language-models  is  limited  by  inference  time  and  expense;  to  begin  to  address  this  and  demonstrate  how  these  two  approaches  can  be  combined,  in  the  final  chapter,  the  application  of  the  performance-energy  tradeoff  management  stack  is  outlined  for  deep  neural  network  inference-stage  batch-size  control.
■590    ▼aSchool  code:  0212.
■650  4▼aSoftware
■650  4▼aDynamic  programming
■650  4▼aCooling
■650  4▼aNeural  networks
■650  4▼aEnergy  management
■650  4▼aHard  disks
■650  4▼aEnergy  efficiency
■650  4▼aAlternative  energy  sources
■650  4▼aEnergy  resources
■650  4▼aTransistors
■650  4▼aCloud  computing
■650  4▼aEnergy  consumption
■650  4▼aAlternative  energy
■650  4▼aElectrical  engineering
■650  4▼aSustainability
■690    ▼a0363
■690    ▼a0800
■690    ▼a0544
■690    ▼a0640
■71020▼aStanford  University.
■7730  ▼tDissertations  Abstracts  International▼g87-05A.
■790    ▼a0212
■791    ▼aPh.D.
■792    ▼a2023
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17366021▼nKERIS▼z이  자료의  원문은  한국교육학술정보원에서  제공합니다.

미리보기

내보내기

chatGPT토론

Ai 추천 관련 도서


    신착도서 더보기
    최근 3년간 통계입니다.

    소장정보

    • 예약
    • 소재불명신고
    • 나의폴더
    • 우선정리요청
    • 비도서대출신청
    • 야간 도서대출신청
    소장자료
    등록번호 청구기호 소장처 대출가능여부 대출정보
    TF16343 전자도서 대출가능 마이폴더 부재도서신고 비도서대출신청 야간 도서대출신청

    * 대출중인 자료에 한하여 예약이 가능합니다. 예약을 원하시면 예약버튼을 클릭하십시오.

    해당 도서를 다른 이용자가 함께 대출한 도서

    관련 인기도서

    로그인 후 이용 가능합니다.