본문

서브메뉴

Scheduling and Codesigning Machine Learning Models for GEMMs
Scheduling and Codesigning Machine Learning Models for GEMMs
Scheduling and Codesigning Machine Learning Models for GEMMs

상세정보

자료유형  
 학위논문 서양
최종처리일시  
20260202103135
ISBN  
9798311951364
DDC  
330
저자명  
Fu, Daniel Y.
서명/저자  
Scheduling and Codesigning Machine Learning Models for GEMMs
발행사항  
[Sl] : Stanford University, 2025
발행사항  
Ann Arbor : ProQuest Dissertations & Theses, 2025
형태사항  
194 p
주기사항  
Source: Dissertations Abstracts International, Volume: 86-12, Section: B.
주기사항  
Advisor: Fatahalian, Kayvon;Ré, Chris.
학위논문주기  
Thesis (Ph.D.)--Stanford University, 2025.
초록/해제  
요약Large-scale training of machine learning (ML) models on hardware accelerators like GPUs has led to substantial progress in recent years. Of critical importance is achieving high FLOP utilization during training. However, efficiently scheduling a diverse range of machine learning operators onto the specialized compute cores present in these hardware accelerators remains a challenge. For example, tensor core units, which are specialized to only compute matrix-matrix multiplication (GEMMs), compose up to 94% of the available FLOPs on Nvidia H100 GPUs. This specialization presents a challenge for modern ML models, which often interleave GEMMs with non-linearities (e.g., exponentiation), reduction operations (e.g., row-wise average), or non-GEMM operations (e.g., FFT).This thesis investigates how to schedule and codesign core machine learning operations onto specialized GEMM units, with a focus on Nvidia tensor cores (though the findings have been generalized to other hardware accelerators). We present two case studies on scheduling two core ML operations, attention and state space models (SSMs), followed by a study on codesigning a new architecture, Monarch Mixer, to utilize tensor cores from the ground-up. First, we focus on scheduling the attention operation, the core primitive at the heart of the Transformer architecture. We find that a simple algebraic reformulation of the softmax operation (combination of exponentiation and row-wise reduction) allows a tiling-based algorithm to achieve high FLOP utilization onto tensor cores. Second, we focus on SSMs, a promising next-generation architecture with better asymptotic complexity than Transformers. SSMs rely on a non-GEMM operation at their core, an FFT convolution. We show that an alternate algebraic formulation of the FFT, called a Monarch decomposition, allows the FFT convolution to be rewritten as a series of tensor core operations. Finally, we conduct a study on codesigning a new architecture, Monarch Mixer, to utilize tensor cores from the ground-up. We show that a generalized Monarch decomposition allows for a high-quality, asymptotically-efficient, and GEMM-friendly ML architecture. Overall, this thesis demonstrates that algebraic reformulation of ML operations is a powerful tool for scheduling diverse ML operations onto GEMMs.
일반주제명  
Sparsity
일반주제명  
Decomposition
일반주제명  
Benchmarks
일반주제명  
Computer science
기타저자  
Stanford University.
기본자료저록  
Dissertations Abstracts International. 86-12B.
전자적 위치 및 접속  
로그인 후 원문을 볼 수 있습니다.

MARC

 008260126s2025        us                              c    eng  d
■001000017357124
■00520260202103135
■006m          o    d                
■007cr#unu||||||||
■020    ▼a9798311951364
■035    ▼a(MiAaPQ)AAI31974593
■035    ▼a(MiAaPQ)Stanfordcs793yc4618
■040    ▼aMiAaPQ▼cMiAaPQ
■0820  ▼a330
■1001  ▼aFu,  Daniel  Y.
■24510▼aScheduling  and  Codesigning  Machine  Learning  Models  for  GEMMs
■260    ▼a[Sl]▼bStanford  University▼c2025
■260  1▼aAnn  Arbor▼bProQuest  Dissertations  &  Theses▼c2025
■300    ▼a194  p
■500    ▼aSource:  Dissertations  Abstracts  International,  Volume:  86-12,  Section:  B.
■500    ▼aAdvisor:  Fatahalian,  Kayvon;Ré,  Chris.
■5021  ▼aThesis  (Ph.D.)--Stanford  University,  2025.
■520    ▼aLarge-scale  training  of  machine  learning  (ML)  models  on  hardware  accelerators  like  GPUs  has  led  to  substantial  progress  in  recent  years.  Of  critical  importance  is  achieving  high  FLOP  utilization  during  training.  However,  efficiently  scheduling  a  diverse  range  of  machine  learning  operators  onto  the  specialized  compute  cores  present  in  these  hardware  accelerators  remains  a  challenge.  For  example,  tensor  core  units,  which  are  specialized  to  only  compute  matrix-matrix  multiplication  (GEMMs),  compose  up  to  94%  of  the  available  FLOPs  on  Nvidia  H100  GPUs.  This  specialization  presents  a  challenge  for  modern  ML  models,  which  often  interleave  GEMMs  with  non-linearities  (e.g.,  exponentiation),  reduction  operations  (e.g.,  row-wise  average),  or  non-GEMM  operations  (e.g.,  FFT).This  thesis  investigates  how  to  schedule  and  codesign  core  machine  learning  operations  onto  specialized  GEMM  units,  with  a  focus  on  Nvidia  tensor  cores  (though  the  findings  have  been  generalized  to  other  hardware  accelerators).  We  present  two  case  studies  on  scheduling  two  core  ML  operations,  attention  and  state  space  models  (SSMs),  followed  by  a  study  on  codesigning  a  new  architecture,  Monarch  Mixer,  to  utilize  tensor  cores  from  the  ground-up.  First,  we  focus  on  scheduling  the  attention  operation,  the  core  primitive  at  the  heart  of  the  Transformer  architecture.  We  find  that  a  simple  algebraic  reformulation  of  the  softmax  operation  (combination  of  exponentiation  and  row-wise  reduction)  allows  a  tiling-based  algorithm  to  achieve  high  FLOP  utilization  onto  tensor  cores.  Second,  we  focus  on  SSMs,  a  promising  next-generation  architecture  with  better  asymptotic  complexity  than  Transformers.  SSMs  rely  on  a  non-GEMM  operation  at  their  core,  an  FFT  convolution.  We  show  that  an  alternate  algebraic  formulation  of  the  FFT,  called  a  Monarch  decomposition,  allows  the  FFT  convolution  to  be  rewritten  as  a  series  of  tensor  core  operations.  Finally,  we  conduct  a  study  on  codesigning  a  new  architecture,  Monarch  Mixer,  to  utilize  tensor  cores  from  the  ground-up.  We  show  that  a  generalized  Monarch  decomposition  allows  for  a  high-quality,  asymptotically-efficient,  and  GEMM-friendly  ML  architecture.  Overall,  this  thesis  demonstrates  that  algebraic  reformulation  of  ML  operations  is  a  powerful  tool  for  scheduling  diverse  ML  operations  onto  GEMMs.
■590    ▼aSchool  code:  0212.
■650  4▼aSparsity
■650  4▼aDecomposition
■650  4▼aBenchmarks
■650  4▼aComputer  science
■690    ▼a0984
■690    ▼a0800
■71020▼aStanford  University.
■7730  ▼tDissertations  Abstracts  International▼g86-12B.
■790    ▼a0212
■791    ▼aPh.D.
■792    ▼a2025
■793    ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17357124▼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
    TF16801 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.