WO2022179075A1 - Procédé et appareil de traitement de données, dispositif informatique et support de stockage - Google Patents

Procédé et appareil de traitement de données, dispositif informatique et support de stockage Download PDF

Info

Publication number
WO2022179075A1
WO2022179075A1 PCT/CN2021/115789 CN2021115789W WO2022179075A1 WO 2022179075 A1 WO2022179075 A1 WO 2022179075A1 CN 2021115789 W CN2021115789 W CN 2021115789W WO 2022179075 A1 WO2022179075 A1 WO 2022179075A1
Authority
WO
WIPO (PCT)
Prior art keywords
processing
array
data
weight
elements
Prior art date
Application number
PCT/CN2021/115789
Other languages
English (en)
Chinese (zh)
Inventor
周军
常亮
周亮
吴飞
Original Assignee
成都商汤科技有限公司
电子科技大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 成都商汤科技有限公司, 电子科技大学 filed Critical 成都商汤科技有限公司
Publication of WO2022179075A1 publication Critical patent/WO2022179075A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • 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
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features

Definitions

  • the present disclosure relates to the field of computer technology, and in particular, to a data processing method, apparatus, computer device, and storage medium.
  • the method before the determining of the target feature elements and target weight elements corresponding to the multiple processing cycles respectively, the method further includes: based on the size of the PE array, performing the processing on the original to-be-processed image feature matrix and the original weight matrix. The size is transformed to obtain the feature matrix of the to-be-processed image and the weight matrix.
  • the use of the PE array to perform a preset operation on the first operand and the second operand stored in the PE array to obtain intermediate processing data corresponding to the processing cycle Including: in the processing cycle, weighted summation is performed on each row of target feature elements in the first operand and each row of weight elements in the second operand to obtain the different weight matrices Corresponding intermediate data; based on the intermediate data corresponding to the different weight matrices, obtain the intermediate processing data corresponding to the processing period.
  • the original to-be-processed image feature matrix may be size-transformed, or not.
  • the size of the feature map is smaller than the size of the PE array, then only a part of the PE in the PE array will be used during the process of using the PE array, while Not all will be used.
  • the fourth column a1, a2, a3, and a4 are weighted and summed to obtain the intermediate data O4 1 corresponding to the weight matrix W4.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computational Mathematics (AREA)
  • Software Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Multimedia (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Algebra (AREA)
  • Databases & Information Systems (AREA)
  • Image Processing (AREA)

Abstract

Procédé et appareil de traitement de données, dispositif informatique et support de stockage. Le procédé consiste : à déterminer, à partir d'une matrice de caractéristiques d'image à traiter et de matrices de pondération, des éléments de caractéristiques cibles et des éléments de pondération cibles correspondant à une pluralité de cycles de traitement, ladite matrice de caractéristiques d'image correspondant à une pluralité de matrices de pondération ; en réponse à l'arrivée d'un quelconque cycle de traitement, à acquérir, par chaque moteur de traitement (PE) dans un réseau PE, un élément de caractéristiques cible correspondant et un élément de pondération cible correspondant du cycle de traitement, et à effectuer une opération prédéfinie, de manière à obtenir des données de traitement intermédiaires, pour tout cycle de traitement, des éléments de caractéristiques cibles dans le réseau PE comprenant des éléments de caractéristiques répétés, et des éléments de pondération cible correspondant aux éléments de caractéristiques répétés étant des éléments de pondération, correspondant aux éléments de caractéristiques répétés, dans différentes matrices de pondération ; et sur la base de données de traitement intermédiaire correspondant à la pluralité de cycles de traitement, à obtenir des données de résultat du traitement de ladite matrice de caractéristiques d'image.
PCT/CN2021/115789 2021-02-26 2021-08-31 Procédé et appareil de traitement de données, dispositif informatique et support de stockage WO2022179075A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110221235.3 2021-02-26
CN202110221235.3A CN112966729B (zh) 2021-02-26 2021-02-26 一种数据处理方法、装置、计算机设备及存储介质

Publications (1)

Publication Number Publication Date
WO2022179075A1 true WO2022179075A1 (fr) 2022-09-01

Family

ID=76275794

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/115789 WO2022179075A1 (fr) 2021-02-26 2021-08-31 Procédé et appareil de traitement de données, dispositif informatique et support de stockage

Country Status (2)

Country Link
CN (1) CN112966729B (fr)
WO (1) WO2022179075A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112966729B (zh) * 2021-02-26 2023-01-31 成都商汤科技有限公司 一种数据处理方法、装置、计算机设备及存储介质
CN113253336B (zh) * 2021-07-02 2021-10-01 深圳市翩翩科技有限公司 一种基于深度学习的地震预测方法和***

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106250103A (zh) * 2016-08-04 2016-12-21 东南大学 一种卷积神经网络循环卷积计算数据重用的***
US20190220734A1 (en) * 2016-10-11 2019-07-18 The Research Foundation For The State University Of New York System, Method, and Accelerator to Process Convolutional Neural Network Layers
CN112149047A (zh) * 2019-06-27 2020-12-29 深圳市中兴微电子技术有限公司 数据的处理方法及装置、存储介质和电子装置
CN112966729A (zh) * 2021-02-26 2021-06-15 成都商汤科技有限公司 一种数据处理方法、装置、计算机设备及存储介质

Family Cites Families (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106447037B (zh) * 2015-10-08 2019-02-12 上海兆芯集成电路有限公司 具有多个可选择输出的神经网络单元
US10489479B1 (en) * 2016-09-12 2019-11-26 Habana Labs Ltd. Matrix multiplication engine
US10515302B2 (en) * 2016-12-08 2019-12-24 Via Alliance Semiconductor Co., Ltd. Neural network unit with mixed data and weight size computation capability
CN108229645B (zh) * 2017-04-28 2021-08-06 北京市商汤科技开发有限公司 卷积加速和计算处理方法、装置、电子设备及存储介质
CN108805275B (zh) * 2017-06-16 2021-01-22 上海兆芯集成电路有限公司 可编程设备及其操作方法和计算机可用介质
CN109213962B (zh) * 2017-07-07 2020-10-09 华为技术有限公司 运算加速器
US10671349B2 (en) * 2017-07-24 2020-06-02 Tesla, Inc. Accelerated mathematical engine
CN108665059A (zh) * 2018-05-22 2018-10-16 中国科学技术大学苏州研究院 基于现场可编程门阵列的卷积神经网络加速***
CN110659445B (zh) * 2018-06-29 2022-12-30 龙芯中科技术股份有限公司 一种运算装置及其处理方法
CN109635944B (zh) * 2018-12-24 2020-10-27 西安交通大学 一种稀疏卷积神经网络加速器及实现方法
CN109740115A (zh) * 2019-01-08 2019-05-10 郑州云海信息技术有限公司 一种实现矩阵乘法运算的方法、装置及设备
US11379555B2 (en) * 2019-06-28 2022-07-05 Amazon Technologies, Inc. Dilated convolution using systolic array
CN110705687B (zh) * 2019-09-05 2020-11-03 北京三快在线科技有限公司 卷积神经网络硬件计算装置及方法
CN111414994B (zh) * 2020-03-03 2022-07-12 哈尔滨工业大学 一种基于FPGA的Yolov3网络计算加速***及其加速方法
CN111582467B (zh) * 2020-05-14 2023-12-22 上海商汤智能科技有限公司 人工智能加速器和电子设备
CN111967582B (zh) * 2020-08-07 2022-07-08 苏州浪潮智能科技有限公司 一种cnn卷积层运算方法及cnn卷积层运算加速器
CN111897579B (zh) * 2020-08-18 2024-01-30 腾讯科技（深圳）有限公司 图像数据处理方法、装置、计算机设备和存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106250103A (zh) * 2016-08-04 2016-12-21 东南大学 一种卷积神经网络循环卷积计算数据重用的***
US20190220734A1 (en) * 2016-10-11 2019-07-18 The Research Foundation For The State University Of New York System, Method, and Accelerator to Process Convolutional Neural Network Layers
CN112149047A (zh) * 2019-06-27 2020-12-29 深圳市中兴微电子技术有限公司 数据的处理方法及装置、存储介质和电子装置
CN112966729A (zh) * 2021-02-26 2021-06-15 成都商汤科技有限公司 一种数据处理方法、装置、计算机设备及存储介质

Also Published As

Publication number Publication date
CN112966729B (zh) 2023-01-31
CN112966729A (zh) 2021-06-15

Similar Documents

Publication Publication Date Title
US11341399B2 (en) Reducing power consumption in a neural network processor by skipping processing operations
US20210117810A1 (en) On-chip code breakpoint debugging method, on-chip processor, and chip breakpoint debugging system
US20210224125A1 (en) Operation Accelerator, Processing Method, and Related Device
CN109003132B (zh) 广告推荐方法及相关产品
US10943167B1 (en) Restructuring a multi-dimensional array
US11775430B1 (en) Memory access for multiple circuit components
CN109522052B (zh) 一种计算装置及板卡
US9886377B2 (en) Pipelined convolutional operations for processing clusters
US20230026006A1 (en) Convolution computation engine, artificial intelligence chip, and data processing method
TW202026858A (zh) 在深度神經網路中利用啟動稀疏性
US11500811B2 (en) Apparatuses and methods for map reduce
WO2022179075A1 (fr) Procédé et appareil de traitement de données, dispositif informatique et support de stockage
WO2022179074A1 (fr) Appareil et procédé de traitement de données, dispositif informatique et support de stockage
JP2018120549A (ja) プロセッサ、情報処理装置及びプロセッサの動作方法
JP2015197702A (ja) 情報処理装置、情報処理方法
US11579921B2 (en) Method and system for performing parallel computations to generate multiple output feature maps
CN110991630A (zh) 一种面向边缘计算的卷积神经网络处理器
CN111353598A (zh) 一种神经网络压缩方法、电子设备及计算机可读介质
US20210142144A1 (en) Multi-size convolutional layer
US11467973B1 (en) Fine-grained access memory controller
WO2021082723A1 (fr) Appareil d'execution
WO2021082747A1 (fr) Appareil d'exploitation et produit associé
KR102372869B1 (ko) 인공 신경망을 위한 행렬 연산기 및 행렬 연산 방법
CN111382835A (zh) 一种神经网络压缩方法、电子设备及计算机可读介质
KR102311659B1 (ko) 컨볼루션 신경망 모델에 기초한 컴퓨팅 장치 및 그 동작 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21927502

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE