서브메뉴
검색
Algorithm-Hardware Co-Design for Deep Learning and Probabilistic Computing with Compute-In-Memory Accelerators
Algorithm-Hardware Co-Design for Deep Learning and Probabilistic Computing with Compute-In-Memory Accelerators
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105601
- ISBN
- 9798263394080
- DDC
- 620
- 저자명
- Lu, Anni.
- 서명/저자
- Algorithm-Hardware Co-Design for Deep Learning and Probabilistic Computing with Compute-In-Memory Accelerators
- 발행사항
- [Sl] : Georgia Institute of Technology, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 117 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: A.
- 주기사항
- Advisor: Yu, Shimeng.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2024.
- 초록/해제
- 요약Compute-in-memory (CIM) is an attractive solution to accelerate the multiplyaccumulate (MAC) operations of deep learning and beyond due to its high processing parallelism and energy efficiency. In this thesis, algorithm-hardware co-optimizations of CIM are explored across diverse models and applications. First, the integrated benchmark framework "NeuroSim" for CIM accelerators is validated with actual silicon data and calibrated with adjustment factors. Based on this simulator, the CIM systems for deep neural network (DNN) with limited on-chip resources are explored, addressing the challenge of accommodating large-scale models on area-constrained CIM chips, and the reconfiguration of deploying different models on prefabricated CIM chip with fixed hardware resources.The CIM scheme is further extended to probabilistic computing, where the memory and circuit intrinsic stochasticity are no longer harmful to model accuracy but provides energy-efficient random number generation. A novel CIM accelerator for Bayesian neural network is proposed to generate Gaussian distributed weights using the probabilistic switching of spin-orbit torque magnetic random-access memory (SOT-MRAM) in weak programming. The inherent memory noise is also utilized for scalable and clustered inmemory Ising annealers to solve NP-hard combinatorial optimization problems. An analog CIM annealer using temporal variation of charge-trap transistor and a digital CIM annealer using process variation of static random-access memory (SRAM) are proposed. Design space explorations and system-level performance evaluations are also performed by modifying the validated NeuroSim simulator.
- 일반주제명
- Random access memory
- 일반주제명
- Deep learning
- 일반주제명
- Semiconductors
- 일반주제명
- Recommender systems
- 일반주제명
- Normal distribution
- 일반주제명
- Circuits
- 일반주제명
- CMOS
- 일반주제명
- Co-design
- 일반주제명
- Energy consumption
- 일반주제명
- Annealing
- 일반주제명
- Space exploration
- 일반주제명
- Neural networks
- 일반주제명
- Energy efficiency
- 일반주제명
- Aerospace engineering
- 일반주제명
- Computer science
- 일반주제명
- Electrical engineering
- 일반주제명
- Industrial engineering
- 일반주제명
- Information science
- 일반주제명
- Sustainability
- 기본자료저록
- Dissertations Abstracts International. 87-05A.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2024 us c eng d■001000017360649
■00520260202105601
■006m o d
■007cr#unu||||||||
■020 ▼a9798263394080
■035 ▼a(MiAaPQ)AAI32315975
■035 ▼a(MiAaPQ)GeorgiaTech75293
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a620
■1001 ▼aLu, Anni.
■24510▼aAlgorithm-Hardware Co-Design for Deep Learning and Probabilistic Computing with Compute-In-Memory Accelerators
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a117 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: A.
■500 ▼aAdvisor: Yu, Shimeng.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2024.
■520 ▼aCompute-in-memory (CIM) is an attractive solution to accelerate the multiplyaccumulate (MAC) operations of deep learning and beyond due to its high processing parallelism and energy efficiency. In this thesis, algorithm-hardware co-optimizations of CIM are explored across diverse models and applications. First, the integrated benchmark framework "NeuroSim" for CIM accelerators is validated with actual silicon data and calibrated with adjustment factors. Based on this simulator, the CIM systems for deep neural network (DNN) with limited on-chip resources are explored, addressing the challenge of accommodating large-scale models on area-constrained CIM chips, and the reconfiguration of deploying different models on prefabricated CIM chip with fixed hardware resources.The CIM scheme is further extended to probabilistic computing, where the memory and circuit intrinsic stochasticity are no longer harmful to model accuracy but provides energy-efficient random number generation. A novel CIM accelerator for Bayesian neural network is proposed to generate Gaussian distributed weights using the probabilistic switching of spin-orbit torque magnetic random-access memory (SOT-MRAM) in weak programming. The inherent memory noise is also utilized for scalable and clustered inmemory Ising annealers to solve NP-hard combinatorial optimization problems. An analog CIM annealer using temporal variation of charge-trap transistor and a digital CIM annealer using process variation of static random-access memory (SRAM) are proposed. Design space explorations and system-level performance evaluations are also performed by modifying the validated NeuroSim simulator.
■590 ▼aSchool code: 0078.
■650 4▼aRandom access memory
■650 4▼aDeep learning
■650 4▼aSemiconductors
■650 4▼aRecommender systems
■650 4▼aNormal distribution
■650 4▼aCircuits
■650 4▼aCMOS
■650 4▼aCo-design
■650 4▼aEnergy consumption
■650 4▼aField programmable gate arrays
■650 4▼aAnnealing
■650 4▼aSpace exploration
■650 4▼aNeural networks
■650 4▼aEnergy efficiency
■650 4▼aAerospace engineering
■650 4▼aComputer science
■650 4▼aElectrical engineering
■650 4▼aIndustrial engineering
■650 4▼aInformation science
■650 4▼aSustainability
■690 ▼a0538
■690 ▼a0800
■690 ▼a0984
■690 ▼a0544
■690 ▼a0546
■690 ▼a0723
■690 ▼a0640
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05A.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360649▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


