Matrix multiplication calculating device and matrix multiplication calculating method both oriented to programmable algebra processor

A matrix multiplication and computing device technology, applied in the field of matrix multiplication computing devices, can solve the problem of low execution efficiency of large-scale and small matrix multiplication, and achieve the effect of improving computing efficiency

Active Publication Date: 2014-07-02
上海思朗科技有限公司
View PDF3 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0018] The technical problem to be solved by the present invention is the problem that the existing matrix mult

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Matrix multiplication calculating device and matrix multiplication calculating method both oriented to programmable algebra processor
  • Matrix multiplication calculating device and matrix multiplication calculating method both oriented to programmable algebra processor
  • Matrix multiplication calculating device and matrix multiplication calculating method both oriented to programmable algebra processor

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0027] In view of the existing situation, the present invention proposes a method and device for efficiently calculating matrix multiplication, which is particularly suitable for large-scale small matrix multiplication.

[0028] The large-scale small matrix multiplication vectorization device proposed by the present invention includes: a multi-granularity parallel memory 10, a data cache device 20, a data broadcast cache device 30, and a vector operation device 40. among them:

[0029] The multi-granularity parallel memory 10 is used to store multiple matrices to be multiplied, broadcast indexes, and results after matrix multiplication. The broadcast index is used to broadcast each data in the matrix. The read and write bit width of the memory is consistent with the bit width of the register file 201 of the data cache device, the bit width of the relevant register in the data broadcast cache device, and the operation size of the vector operation device 40, denoted as P.

[0030] Th...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a matrix multiplication calculating device and a matrix multiplication calculating method. The matrix multiplication calculating device comprises a multi-granularity parallel storer, a data caching device, a data broadcast caching device and a vector operation device. Parallel optimized processing is performed on matrix multiplication which is small in matrix size and large in operation quantity in actual application by adopting a programmably-configured DSP (digital signal processor) chip and combing a high-efficiency vectorized matrix multiplication scheme, and the matrix multiplication calculating device and the matrix multiplication calculating method have the advantages of being high in operation speed, high in parallel granularity and small in access number.

Description

technical field [0001] The present invention relates to the technical field of data processing, and more specifically, to a matrix multiplication computing device and method based on a programmable algebraic processor. Background technique [0002] Matrix multiplication is a basic operation in scientific computing, which is widely used in complex computing fields such as signal processing, image processing, radar, sonar, communication, etc., and because its computational complexity is O(n 3 ), making matrix multiplication often become the most time-consuming operation in the calculation process of the algorithm, which in turn affects the performance of the entire algorithm. Matrix multiplication operations are divided into large matrix multiplication and small matrix multiplication. Large matrix multiplication has an exponential increase in the amount of operations due to its huge row and column values. In recent years, it has received widespread attention. A general process...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06F17/16
Inventor 郭晓龙王晓琴王伟康吴军宁林啸郭璟张森赵旭莹
Owner 上海思朗科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap