Unlock instant, AI-driven research and patent intelligence for your innovation.

Deep dependency problem parallel method based on instruction sequence and message sequence guidance

A technology of instruction sequence and message sequence, applied in concurrent instruction execution, program control design, machine execution device, etc., can solve problems such as acceleration and inability to achieve effective

Inactive Publication Date: 2021-03-19
JIANGNAN INST OF COMPUTING TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

After the actual calculation example test, the data under the unstructured grid often have deep dependencies, and the existing Shenwei platform cannot achieve effective acceleration for such problems

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
  • Deep dependency problem parallel method based on instruction sequence and message sequence guidance
  • Deep dependency problem parallel method based on instruction sequence and message sequence guidance
  • Deep dependency problem parallel method based on instruction sequence and message sequence guidance

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0038] Embodiment: a parallel method for deep dependence problems based on instruction sequence and message sequence guidance, oriented to non-network structure, comprising the following steps:

[0039] S1. According to the calculation of load balance between slave cores, the solution vector is equally divided into blocks, and the number of each vector block is determined, and the dependency between elements in the solution vector is converted into the dependency between vector blocks;

[0040] S2. Completing the calculation of a vector block is called an update operation, and this update operation requires the following three steps:

[0041] S21. Partial update: the current vector block receives the data sent by the previous vector block through register communication, and the initial vector block can be solved directly;

[0042] S22. Self-update: the current vector block uses the data transmitted by the previous vector block to solve the elements in the current vector block,...

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 deep dependency problem parallel method based on instruction sequence and message sequence guidance, which comprises the following steps: S1, averagely dividing a solution vector according to blocks, and converting dependencies among elements in the solution vector into dependencies among vector blocks through the division; s2, completing calculation of one vector block to be called as one-time updating operation, wherein the one-time updating operation needs the following three steps: S21, partial updating: receiving data sent by a preceding block; s22, self-updating: calculating and updating the elements in the block; s23, finishing updating: sending elements obtained after the vector block is solved to the dependent subsequent block; s3, calculating corresponding vector blocks by each calculation core in the many cores in sequence, and repeating the updating operation step in the S2 to realize the calculation of the whole assembly line work; and S4, designing a string of instruction streams, namely an instruction sequence, for each computing core. Mutual cooperation of many-core internal communication is improved, memory access time expenditure is reduced, and effective acceleration is achieved.

Description

technical field [0001] The invention belongs to the technical field of numerical calculation, and in particular relates to a parallel method for deep dependence problems guided by instruction sequences and message sequences. Background technique [0002] The depth dependence problem is a common type of problem in the CFD field. For example, the solution of the common sparse lower triangular equation has a deep dependence relationship between the solutions, that is, the subsequent solution depends on the previous solution to be solved, so the solution of this problem is a very stringent row operation. Even such a serial operation can discover the parallel part. Currently, for the problem of deep dependence between the solution vector elements in the unstructured grid, the hierarchical parallel idea is mainly used, that is, whether the solution vector x is related to each other. For layered processing, unknowns belonging to the same layer can be solved in parallel, and the la...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G06F9/38G06F9/30
CPCG06F9/30036G06F9/3838G06F9/3869
Inventor 陈鑫陈德训刘鑫李芳徐金秀孙唯哲郭恒王臻
Owner JIANGNAN INST OF COMPUTING TECH