Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A method for improving computing speed by using mpi and openmp mixed parallel for smp cluster system

A computing speed and cluster system technology, applied in the direction of concurrent instruction execution, machine execution devices, etc., can solve problems such as parallel optimization of algorithms, achieve the effect of reducing synchronization, reducing the number of barriers, and improving computing speed

Active Publication Date: 2018-05-04
INST OF SOFTWARE APPL TECH GUANGZHOU & CHINESE ACAD OF SCI
View PDF2 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in addition to the sparse matrix-vector multiplication, the conjugate gradient method also has calculation steps such as the multiplication and summation of multiple vectors and vectors, scalars and vectors, and the solution of linear equations. These calculations are still calculated using serial methods, which cannot maximize optimize the algorithm in parallel

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
  • A method for improving computing speed by using mpi and openmp mixed parallel for smp cluster system
  • A method for improving computing speed by using mpi and openmp mixed parallel for smp cluster system
  • A method for improving computing speed by using mpi and openmp mixed parallel for smp cluster system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0087] In this embodiment, a method for solving large-scale linear equations is solved by using MPI and OpenMP mixed and parallel to improve the calculation speed for the SMP cluster system. The preconditioned conjugate gradient method is an iterative method for solving symmetric positive definite sparse matrix linear equations. It is widely used in engineering and scientific computing. Its algorithm is as follows:

[0088] take x (0) ∈ R n , calculate r (0) =b-Ax (0) , let p (0) =r(0)

[0089] For k=0, 1, 2, ..., compute

[0090]

[0091] x (k+1) =x (k+1) +α k p (k)

[0092] r (k+1) =b-Ax (k+1) = r (k) -α k AP (k)

[0093] like Then output x′≡x (k+1) , to stop the calculation. otherwise,

[0094]

[0095] p (k+1) = r (k+1) +β k p (k)

[0096] Among them, in large-scale engineering and computing problems, x is a vector to be solved, b is a known vector, and A is a coefficient matrix, which is usually a large sparse matrix. A sparse matrix is ​​on...

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 method for improving computing speed by adopting MPI and OpenMP mixed parallel for SMP cluster system. According to the number of computing nodes and the number of available CPU cores in the nodes, the number of MPI processes and the number of OpenMP threads that can be called are determined; There are sub-sparse matrices, sub-initial vectors, block vectors and maximum calculation tolerances; each process enables multi-threaded compilation instructions; performs cyclic calculation of the preconditioned conjugate gradient method on each process, and there are only three OpenMP barriers in the cyclic calculation. If the calculated error is less than the allowable value, the loop ends, otherwise the loop continues; the calculation results of each process are reduced, and the solution to the problem is output; in parallel computing, the MPI process is started first, the problem is decomposed into multiple processes, and the inter-node communication is started. In parallel, each MPI process is assigned to a computing node, and information is exchanged between processes using message passing; then in each MPI process, a set of threads are created using OpenMP guidance instructions and assigned to different processors of the computing node in parallel implement.

Description

technical field [0001] The invention relates to a parallel computing technology, in particular to a method for improving computing speed through parallel computing. Background technique [0002] The iterative method is currently the mainstream method for solving large sparse linear equations. The preconditioned conjugate gradient method in the iterative method is a method that reduces the number of iterations of the conjugate gradient method through preprocessing technology and can accelerate convergence. It is used in engineering and It has been widely used in scientific computing. The conjugate gradient method is a method for solving the numerical solution of a specific linear system, where the coefficient matrix is ​​a symmetric and positive definite real number matrix. With the increasing scale and complexity of scientific and engineering problems, the serial conjugate gradient method has been difficult to meet the requirements of the scale and speed of solving sparse l...

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 Patents(China)
IPC IPC(8): G06F9/38
Inventor 罗海飙廖俊豪
Owner INST OF SOFTWARE APPL TECH GUANGZHOU & CHINESE ACAD OF SCI
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products