A GPU Parallel Particle Swarm Optimization Method Based on Adaptive Warps
A particle swarm optimization and thread warp technology, applied in concurrent instruction execution, machine execution device, program control design, etc., can solve problems such as enlargement, and achieve the effect of reducing power consumption, saving hardware costs, and shortening the time to solve problems
Active Publication Date: 2019-01-18
NANJING BEIDOU INNOVATION & APPL TECH RES INST CO LTD
View PDF6 Cites 2 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
This will undoubtedly increase
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
[0047] Such as figure 1 As shown, it is a kind of GPU parallel particle swarm optimization method based on adaptive warp of the present embodiment, comprising the following steps:
[0048] Step 1: Initialize the problem function parameters and initialize the particle swarm parameters;
[0049] Step 2: Define three CUDA kernel functions, which are used to calculate the velocity and position of the particle in parallel, the fitness value of the particle, the best fitness value found by the next generation particle itself and its corresponding solution, and the entire particle swarm up to now The best fitness value found so far and its corresponding solution;
[0050] Step 3: Calculate and initialize the BlockNum and GridNum of each kernel function according to the adaptive warp algorithm;
[0051] Step 4: Call the kernel function to iteratively update the velocity and position of the particle swarm in parallel, and find the current best fitness value and its corresp...
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 a self-adaptive thread based GPU parallel particle swarm optimization method. The self-adaptive thread based GPU parallel particle swarm optimization method comprises the following steps that 1, problem function parameters are initialized, and particle swarm parameters are initialized; 2, the CUDA kernel functions are defined and are respectively used for parallelly calculating the speed, position and fitness value of particles, the best fitness value found by next-generation particles themselves and a solution corresponding to the value and the up-to-now found best fitness value of the whole particle swarm and a solution corresponding to the value; 3, according to a self-adaptive thread algorithm, Block and Grid parameters of the kernel functions are calculated and initialized; 4, the kernel functions are invoked, the speed and position of the particle swarm is iterated and updated, and the current fitness value and the solution corresponding to the value are calculated; 5, the step 4 is repeatedly executed till the set finishing conditions are reached, and a GPU outputs a calculation result. The parallel solving time of the particle swarm algorithm on the GPU can be remarkably shortened, the power consumption can be reduced, and the hardness cost can be saved.
Description
technical field [0001] The invention relates to a particle swarm optimization method, which belongs to the field of computer data processing, in particular to a GPU parallel particle swarm optimization method based on an adaptive thread warp. Background technique [0002] Particle Swarm Optimization (PSO) algorithm is an evolutionary computing technology, because of its simple concept, easy implementation, and strong global search and convergence capabilities, it has been rapidly developed and widely used. . Various parallel PSO algorithm versions have appeared. Among them, for the CUDA parallel architecture, there are two main allocation schemes for threads: 1) a thread corresponds to a particle; 2) a thread corresponds to a dimension, and a block corresponds to a particle. Although the first coarse-grained parallel method has achieved a good speed-up ratio, because each dimension corresponding to the particle in each thread is still executed serially, the degree of parall...
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 Patents(China)
IPC IPC(8): G06F9/38G06N3/00
CPCG06F9/3885G06N3/006
Inventor 何发智张硕
Owner NANJING BEIDOU INNOVATION & APPL TECH RES INST CO LTD