CN109416754B - 用于深度神经网络的加速器 - Google Patents
用于深度神经网络的加速器 Download PDFInfo
- Publication number
- CN109416754B CN109416754B CN201780039356.8A CN201780039356A CN109416754B CN 109416754 B CN109416754 B CN 109416754B CN 201780039356 A CN201780039356 A CN 201780039356A CN 109416754 B CN109416754 B CN 109416754B
- Authority
- CN
- China
- Prior art keywords
- bit
- serial
- neurons
- neuron
- tiles
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/063—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using electronic means
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Neurology (AREA)
- Image Processing (AREA)
- Memory System (AREA)
- Complex Calculations (AREA)
Abstract
Description
Claims (46)
Applications Claiming Priority (15)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201662341814P | 2016-05-26 | 2016-05-26 | |
US62/341,814 | 2016-05-26 | ||
US201662381202P | 2016-08-30 | 2016-08-30 | |
US62/381,202 | 2016-08-30 | ||
US201662395027P | 2016-09-15 | 2016-09-15 | |
US62/395,027 | 2016-09-15 | ||
US201662416782P | 2016-11-03 | 2016-11-03 | |
US62/416,782 | 2016-11-03 | ||
US201762448454P | 2017-01-20 | 2017-01-20 | |
US62/448,454 | 2017-01-20 | ||
US201762454268P | 2017-02-03 | 2017-02-03 | |
US62/454,268 | 2017-02-03 | ||
US201762490659P | 2017-04-27 | 2017-04-27 | |
US62/490,659 | 2017-04-27 | ||
PCT/CA2017/050638 WO2017201627A1 (en) | 2016-05-26 | 2017-05-26 | Accelerator for deep neural networks |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109416754A CN109416754A (zh) | 2019-03-01 |
CN109416754B true CN109416754B (zh) | 2020-06-23 |
Family
ID=60412016
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201780039356.8A Active CN109416754B (zh) | 2016-05-26 | 2017-05-26 | 用于深度神经网络的加速器 |
Country Status (7)
Country | Link |
---|---|
US (3) | US10387771B2 (zh) |
EP (1) | EP3465550B1 (zh) |
JP (2) | JP6890615B2 (zh) |
KR (2) | KR102459854B1 (zh) |
CN (1) | CN109416754B (zh) |
CA (1) | CA2990709C (zh) |
WO (1) | WO2017201627A1 (zh) |
Families Citing this family (62)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9875104B2 (en) | 2016-02-03 | 2018-01-23 | Google Llc | Accessing data in multi-dimensional tensors |
US10949736B2 (en) * | 2016-11-03 | 2021-03-16 | Intel Corporation | Flexible neural network accelerator and methods therefor |
CN106557332A (zh) * | 2016-11-30 | 2017-04-05 | 上海寒武纪信息科技有限公司 | 一种指令生成过程的复用方法及装置 |
US20180232627A1 (en) * | 2017-02-16 | 2018-08-16 | Intel IP Corporation | Variable word length neural network accelerator circuit |
EP3633526A1 (en) * | 2017-04-06 | 2020-04-08 | Shanghai Cambricon Information Technology Co., Ltd | Computation device and method |
US11531540B2 (en) * | 2017-04-19 | 2022-12-20 | Cambricon (Xi'an) Semiconductor Co., Ltd. | Processing apparatus and processing method with dynamically configurable operation bit width |
US10019668B1 (en) * | 2017-05-19 | 2018-07-10 | Google Llc | Scheduling neural network processing |
GB2568776B (en) * | 2017-08-11 | 2020-10-28 | Google Llc | Neural network accelerator with parameters resident on chip |
EP3654208A1 (en) * | 2017-08-31 | 2020-05-20 | Cambricon Technologies Corporation Limited | Chip device and related products |
WO2019114842A1 (zh) | 2017-12-14 | 2019-06-20 | 北京中科寒武纪科技有限公司 | 一种集成电路芯片装置 |
CN109961137B (zh) * | 2017-12-14 | 2020-10-09 | 中科寒武纪科技股份有限公司 | 集成电路芯片装置及相关产品 |
US10872291B2 (en) * | 2017-12-22 | 2020-12-22 | Alibaba Group Holding Limited | On-chip communication system for neural network processors |
CN109978131B (zh) * | 2017-12-28 | 2020-05-22 | 中科寒武纪科技股份有限公司 | 集成电路芯片装置、方法及相关产品 |
CN108256644B (zh) * | 2018-01-05 | 2021-06-22 | 上海兆芯集成电路有限公司 | 微处理器电路以及执行神经网络运算的方法 |
KR102463532B1 (ko) * | 2018-01-16 | 2022-11-08 | 한국전자통신연구원 | 신경망 컴퓨팅 장치 및 그것의 동작 방법 |
US11630997B2 (en) * | 2018-01-23 | 2023-04-18 | Samsung Electronics Co., Ltd. | Method and apparatus with bit-serial data processing of a neural network |
GB201801639D0 (en) * | 2018-02-01 | 2018-03-21 | Ruff Brendan Patrick | Low precision efficient multiplication free convolutional filter bank device |
CN111742331A (zh) * | 2018-02-16 | 2020-10-02 | 多伦多大学管理委员会 | 神经网络加速器 |
US11256841B1 (en) * | 2018-03-06 | 2022-02-22 | Ansys Lumerical Ip, Llc | Method and system to implement a composite, multi-domain model for electro-optical modeling and simulation |
US10621489B2 (en) * | 2018-03-30 | 2020-04-14 | International Business Machines Corporation | Massively parallel neural inference computing elements |
WO2019194466A1 (ko) * | 2018-04-03 | 2019-10-10 | 주식회사 퓨리오사에이아이 | 뉴럴 네트워크 프로세서 |
US10698730B2 (en) * | 2018-04-03 | 2020-06-30 | FuriosaAI Co. | Neural network processor |
KR102162749B1 (ko) * | 2018-04-03 | 2020-10-07 | 주식회사 퓨리오사에이아이 | 뉴럴 네트워크 프로세서 |
US10657442B2 (en) * | 2018-04-19 | 2020-05-19 | International Business Machines Corporation | Deep learning accelerator architecture with chunking GEMM |
US10522226B2 (en) * | 2018-05-01 | 2019-12-31 | Silicon Storage Technology, Inc. | Method and apparatus for high voltage generation for analog neural memory in deep learning artificial neural network |
EP3769266A4 (en) * | 2018-05-08 | 2021-12-15 | Tartan Ai Ltd. | PROCESSING ELEMENT FOR NEURAL NETWORK |
US20190378013A1 (en) * | 2018-06-06 | 2019-12-12 | Kneron Inc. | Self-tuning model compression methodology for reconfiguring deep neural network and electronic device |
US11275713B2 (en) * | 2018-06-09 | 2022-03-15 | International Business Machines Corporation | Bit-serial linear algebra processor |
US20190392287A1 (en) | 2018-06-22 | 2019-12-26 | Samsung Electronics Co., Ltd. | Neural processor |
KR102650660B1 (ko) * | 2018-09-03 | 2024-03-25 | 삼성전자주식회사 | 뉴로모픽 장치 및 뉴로모픽 장치에서 멀티-비트 뉴로모픽 연산을 처리하는 방법 |
WO2020072619A1 (en) * | 2018-10-02 | 2020-04-09 | The Regents Of The University Of Michigan | Addressing bottlenecks for deep neural network execution of a graphics processor unit |
US11354888B2 (en) | 2018-11-16 | 2022-06-07 | GM Global Technology Operations LLC | Method and apparatus for a neural network |
KR102451519B1 (ko) * | 2018-11-20 | 2022-10-07 | 삼성전자주식회사 | 룩업 테이블 기반의 비트-시리얼 처리 요소를 포함하는 심층 신경망 가속기 |
US10657617B1 (en) | 2018-11-26 | 2020-05-19 | GM Global Technology Operations LLC | Method and apparatus for memory access management for data processing |
KR102562320B1 (ko) | 2018-12-24 | 2023-08-01 | 삼성전자주식회사 | 비트 연산 기반의 뉴럴 네트워크 처리 방법 및 장치 |
CN109711542B (zh) * | 2018-12-29 | 2020-08-18 | 西安交通大学 | 一种支持动态精度的dnn加速器及其实现方法 |
US11551093B2 (en) * | 2019-01-22 | 2023-01-10 | Adobe Inc. | Resource-aware training for neural networks |
WO2020172494A1 (en) | 2019-02-22 | 2020-08-27 | Neureality Ltd. | Directed and interconnected grid dataflow architecture |
US11211944B2 (en) | 2019-04-17 | 2021-12-28 | Samsung Electronics Co., Ltd. | Mixed-precision compression with random access |
US11671111B2 (en) | 2019-04-17 | 2023-06-06 | Samsung Electronics Co., Ltd. | Hardware channel-parallel data compression/decompression |
US11880760B2 (en) | 2019-05-01 | 2024-01-23 | Samsung Electronics Co., Ltd. | Mixed-precision NPU tile with depth-wise convolution |
US11488650B2 (en) | 2020-04-06 | 2022-11-01 | Memryx Incorporated | Memory processing unit architecture |
US10998037B2 (en) | 2019-05-07 | 2021-05-04 | Memryx Incorporated | Memory processing units and methods of computing dot products |
US10853066B1 (en) | 2019-05-07 | 2020-12-01 | Memryx Incorporated | Memory processing units and methods of computing DOT products including zero bit skipping |
US11507349B2 (en) * | 2019-06-26 | 2022-11-22 | Microsoft Technology Licensing, Llc | Neural processing element with single instruction multiple data (SIMD) compute lanes |
US11604972B2 (en) | 2019-06-28 | 2023-03-14 | Microsoft Technology Licensing, Llc | Increased precision neural processing element |
KR102435447B1 (ko) * | 2019-07-23 | 2022-08-25 | 포항공과대학교 산학협력단 | 뉴럴 네트워크 시스템 및 이의 동작 방법 |
US12039430B2 (en) * | 2019-11-15 | 2024-07-16 | Samsung Electronics Co., Ltd. | Electronic device and method for inference binary and ternary neural networks |
WO2021100900A1 (ko) * | 2019-11-20 | 2021-05-27 | 전자부품연구원 | 적응형 딥러닝 가속 장치 및 방법 |
CN112990449B (zh) * | 2019-12-12 | 2022-10-25 | 上海交通大学 | 一种面向神经网络的可变计算精度的simt*** |
US11520584B2 (en) * | 2019-12-13 | 2022-12-06 | Intel Corporation | FPGA specialist processing block for machine learning |
US11809798B2 (en) * | 2019-12-13 | 2023-11-07 | Intel Corporation | Implementing large multipliers in tensor arrays |
US11314483B2 (en) * | 2020-01-08 | 2022-04-26 | International Business Machines Corporation | Bit-serial computation with dynamic frequency modulation for error resiliency in neural network |
US12039433B2 (en) | 2020-02-20 | 2024-07-16 | Ohio University | Optical multiply and accumulate unit |
US11281554B2 (en) | 2020-03-17 | 2022-03-22 | Samsung Electronics Co., Ltd. | System and method for in-memory computation |
US12001929B2 (en) * | 2020-04-01 | 2024-06-04 | Samsung Electronics Co., Ltd. | Mixed-precision neural processing unit (NPU) using spatial fusion with load balancing |
US11514326B2 (en) | 2020-06-18 | 2022-11-29 | International Business Machines Corporation | Drift regularization to counteract variation in drift coefficients for analog accelerators |
KR102474053B1 (ko) * | 2020-06-22 | 2022-12-06 | 주식회사 퓨리오사에이아이 | 뉴럴네트워크 프로세서 |
KR102474054B1 (ko) * | 2020-06-22 | 2022-12-06 | 주식회사 퓨리오사에이아이 | 뉴럴네트워크 프로세서 |
WO2022027172A1 (zh) * | 2020-08-03 | 2022-02-10 | 深圳市大疆创新科技有限公司 | 数据处理装置、方法和***以及神经网络加速器 |
CN114004347A (zh) | 2021-08-30 | 2022-02-01 | 平头哥(上海)半导体技术有限公司 | 用于加速图神经网络属性访问的硬件加速器、***和方法 |
CN114254740B (zh) * | 2022-01-18 | 2022-09-30 | 长沙金维信息技术有限公司 | 卷积神经网络加速计算方法、计算***、芯片及接收机 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5751913A (en) * | 1996-07-29 | 1998-05-12 | Industrial Technology Research Institute | Reconfigurable neural network and difference-square neuron |
WO2014085975A1 (zh) * | 2012-12-04 | 2014-06-12 | 中国科学院半导体研究所 | 可动态重构的多级并行单指令多数据阵列处理*** |
CN105488563A (zh) * | 2015-12-16 | 2016-04-13 | 重庆大学 | 面向深度学习的稀疏自适应神经网络、算法及实现装置 |
WO2016073803A1 (en) * | 2014-11-06 | 2016-05-12 | Microsoft Technology Licensing, Llc | Context-based actions in voice user interface |
Family Cites Families (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3260357B2 (ja) * | 1990-01-24 | 2002-02-25 | 株式会社日立製作所 | 情報処理装置 |
WO1991018351A1 (en) * | 1990-05-22 | 1991-11-28 | International Business Machines Corporation | Pyramid learning architecture neurocomputer |
JP2785155B2 (ja) * | 1990-09-10 | 1998-08-13 | 富士通株式会社 | ニューロコンピュータの非同期制御方式 |
JPH0652132A (ja) * | 1992-07-28 | 1994-02-25 | Mitsubishi Electric Corp | 並列演算半導体集積回路装置およびそれを用いたシステム |
US5630024A (en) * | 1994-01-19 | 1997-05-13 | Nippon Telegraph And Telephone Corporation | Method and apparatus for processing using neural network with reduced calculation amount |
JPH07210533A (ja) * | 1994-01-19 | 1995-08-11 | Nippon Telegr & Teleph Corp <Ntt> | ニューラルネットワーク回路及びこれを用いた演算方法 |
US6199057B1 (en) * | 1996-10-23 | 2001-03-06 | California Institute Of Technology | Bit-serial neuroprocessor architecture |
RU2131145C1 (ru) * | 1998-06-16 | 1999-05-27 | Закрытое акционерное общество Научно-технический центр "Модуль" | Нейропроцессор, устройство для вычисления функций насыщения, вычислительное устройство и сумматор |
JP2010134389A (ja) | 2008-12-08 | 2010-06-17 | Oki Data Corp | 定着装置及び画像形成装置 |
GB2471067B (en) * | 2009-06-12 | 2011-11-30 | Graeme Roy Smith | Shared resource multi-thread array processor |
US9400955B2 (en) * | 2013-12-13 | 2016-07-26 | Amazon Technologies, Inc. | Reducing dynamic range of low-rank decomposition matrices |
US11551076B2 (en) | 2014-09-05 | 2023-01-10 | Qualcomm Incorporated | Event-driven temporal convolution for asynchronous pulse-modulated sampled signals |
-
2017
- 2017-05-26 CA CA2990709A patent/CA2990709C/en active Active
- 2017-05-26 KR KR1020207015810A patent/KR102459854B1/ko active IP Right Grant
- 2017-05-26 US US15/606,118 patent/US10387771B2/en active Active
- 2017-05-26 KR KR1020187037389A patent/KR102120396B1/ko active IP Right Grant
- 2017-05-26 JP JP2018562055A patent/JP6890615B2/ja active Active
- 2017-05-26 CN CN201780039356.8A patent/CN109416754B/zh active Active
- 2017-05-26 EP EP17801877.6A patent/EP3465550B1/en active Active
- 2017-05-26 WO PCT/CA2017/050638 patent/WO2017201627A1/en unknown
-
2019
- 2019-07-07 US US16/504,275 patent/US11610100B2/en active Active
-
2021
- 2021-05-25 JP JP2021087474A patent/JP7166389B2/ja active Active
-
2023
- 2023-02-10 US US18/108,292 patent/US20230186065A1/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5751913A (en) * | 1996-07-29 | 1998-05-12 | Industrial Technology Research Institute | Reconfigurable neural network and difference-square neuron |
WO2014085975A1 (zh) * | 2012-12-04 | 2014-06-12 | 中国科学院半导体研究所 | 可动态重构的多级并行单指令多数据阵列处理*** |
WO2016073803A1 (en) * | 2014-11-06 | 2016-05-12 | Microsoft Technology Licensing, Llc | Context-based actions in voice user interface |
CN105488563A (zh) * | 2015-12-16 | 2016-04-13 | 重庆大学 | 面向深度学习的稀疏自适应神经网络、算法及实现装置 |
Also Published As
Publication number | Publication date |
---|---|
CA2990709A1 (en) | 2017-11-30 |
CN109416754A (zh) | 2019-03-01 |
JP2021144727A (ja) | 2021-09-24 |
KR20200067915A (ko) | 2020-06-12 |
EP3465550A1 (en) | 2019-04-10 |
US11610100B2 (en) | 2023-03-21 |
WO2017201627A1 (en) | 2017-11-30 |
KR102120396B1 (ko) | 2020-06-08 |
EP3465550B1 (en) | 2023-09-27 |
JP6890615B2 (ja) | 2021-06-18 |
US20200125931A1 (en) | 2020-04-23 |
KR20190010642A (ko) | 2019-01-30 |
EP3465550A4 (en) | 2020-01-22 |
KR102459854B1 (ko) | 2022-10-27 |
US20230186065A1 (en) | 2023-06-15 |
US10387771B2 (en) | 2019-08-20 |
JP2019522271A (ja) | 2019-08-08 |
CA2990709C (en) | 2018-09-04 |
US20170357891A1 (en) | 2017-12-14 |
JP7166389B2 (ja) | 2022-11-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109416754B (zh) | 用于深度神经网络的加速器 | |
JP7266065B2 (ja) | ディープニューラルネットワーク用のシステム、コンピュータ実装方法及びコンピュータプログラム | |
Song et al. | 7.1 An 11.5 TOPS/W 1024-MAC butterfly structure dual-core sparsity-aware neural processing unit in 8nm flagship mobile SoC | |
CN111291859B (zh) | 通用矩阵-矩阵乘法数据流加速器半导体电路 | |
CN111742331A (zh) | 神经网络加速器 | |
US11462003B2 (en) | Flexible accelerator for sparse tensors in convolutional neural networks | |
US10915297B1 (en) | Hardware accelerator for systolic matrix multiplication | |
US11797830B2 (en) | Flexible accelerator for sparse tensors in convolutional neural networks | |
CN110851779B (zh) | 用于稀疏矩阵运算的脉动阵列架构 | |
US20220374691A1 (en) | Processing element and neural processing device including same | |
Chen et al. | An efficient accelerator for multiple convolutions from the sparsity perspective | |
WO2022016261A1 (en) | System and method for accelerating training of deep learning networks | |
Choi et al. | MLogNet: A logarithmic quantization-based accelerator for depthwise separable convolution | |
Dey et al. | An application specific processor architecture with 3D integration for recurrent neural networks | |
Judd | Exploiting Data Characteristics in The Design of Accelerators for Deep Learning |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20210207 Address after: Toronto, Ontario, Canada Patentee after: Datan artificial intelligence Co.,Ltd. Address before: Ontario, Canada Patentee before: THE GOVERNING COUNCIL OF THE University OF TORONTO |
|
TR01 | Transfer of patent right | ||
CB03 | Change of inventor or designer information |
Inventor after: P Judd Inventor after: J. Alberio Inventor after: A. Delmas Laskoz Inventor after: A. Mohovos Inventor after: S. Sharif mogadam Inventor before: P Judd Inventor before: J. Alberio Inventor before: A. Delmas Laskoz Inventor before: A. Mohovos Inventor before: S. Sally |
|
CB03 | Change of inventor or designer information | ||
TR01 | Transfer of patent right |
Effective date of registration: 20220330 Address after: Gyeonggi Do Korea Suwon Patentee after: SAMSUNG ELECTRONICS Co.,Ltd. Address before: Toronto, Ontario, Canada Patentee before: Datan artificial intelligence Co.,Ltd. |
|
TR01 | Transfer of patent right |