Unstructured grid master-slave dynamic parallel preprocessing method
An unstructured grid and preprocessing technology, which is applied in the computer field, can solve problems such as low efficiency and large preprocessing memory requirements, and achieve the effects of solving excessive memory requirements, saving memory, and solving low efficiency
Inactive Publication Date: 2021-03-05
JIANGNAN INST OF COMPUTING TECH
View PDF0 Cites 2 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] The purpose of the present invention is to provide a master-slave dynamic parallel preprocessing method for unstructured grids. Excessive memory requirements and low efficiency of large-scale unstructured grid preprocessing
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment
[0021] Embodiment: a master-slave dynamic parallel preprocessing method of an unstructured grid, comprising the following steps:
[0022] S1. The master process assigns tasks according to the order of physical blocks, and distributes the grid information of each physical block to each slave process;
[0023] S2. The slave process receives the task from the master process, and preprocesses the grid units in each physical block, that is, establishes the data structure required by the METIS block for the physical block. This data structure includes the grid unit serial number xadj and The serial number adjncy of the neighbor unit of each grid unit, the preprocessing steps are as follows:
[0024] S21. Initially set the serial number adjncy of the neighbor unit of each grid unit to 0, and initialize the number of neighbors of each grid unit to 0;
[0025] S22. Each slave process sets a grid unit preprocessor buffer according to the size of the CPU memory, for batch processing the...
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
Login to View More Abstract
The invention discloses an unstructured grid master-slave dynamic parallel preprocessing method, which is based on a ten-million-order and above-order grid formed by a plurality of physical blocks, and comprises the following steps: S1: a master process allocates tasks according to the sequence of the physical blocks, and distributes grid information of each physical block to each slave process; s2, a task is received from the main process from the process, and the grid units in each physical block are preprocessed; s3, the slave process sends a preprocessing result back to the master process,and applies for a next group of grids from the master process for preprocessing; s4, after the master process distributes all the grid groups, an end signal is sent to the slave process; and S5, after the slave process receives the ending signal, the slave process ends. According to the invention, the numerical simulation requirement of a ten-million-order and above-order grid scale can be met, and the problems that the large-scale unstructured grid preprocessing memory requirement is too large and the efficiency is low are solved.
Description
technical field [0001] The invention relates to a master-slave dynamic parallel preprocessing method of an unstructured grid, which belongs to the technical field of computers. Background technique [0002] The existing unstructured computing software generally adopts the relatively mature unstructured grid block software METIS for load balancing and task division. Due to historical reasons, the existing computational fluid dynamics software does not consider the need for ultra-large-scale unstructured grid simulation, and the preprocessing process is a serial program. When the grid size of the numerical simulation increases to the order of tens of millions and above, The preprocessing process requires too much memory, which often exceeds the memory capacity of a single CPU of a conventional high-performance computer, resulting in the failure of preprocessing or the low efficiency of preprocessing. [0003] In recent years, the application of unstructured grids in computati...
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
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06F9/50
CPCG06F9/5016G06F9/5072
Inventor 李芳刘鑫孙唯哲徐占何香王臻
Owner JIANGNAN INST OF COMPUTING TECH