Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Method and system for constructing time overhead model based on kernel function

A time and model technology, applied in the direction of electrical digital data processing, multi-program device, program control design, etc., can solve the problem of not considering the thread block allocation time in the kernel function, the parallel performance of the algorithm cannot be maximized, and the thread block subjective problems such as sex and arbitrariness, to achieve the effect of improving operating efficiency, comprehensive and reasonable structure, and ensuring effectiveness

Inactive Publication Date: 2019-08-02
CHINA AGRI UNIV
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The current time overhead model only involves the execution time of the kernel function, and does not consider the allocation time of thread blocks in the kernel function, which affects the user's estimation of the running time of the parallel algorithm to a certain extent; in addition, when transplanting parallel programs, Due to the subjectivity and arbitrariness of the choice of thread block size, the parallel performance of the algorithm will not be maximized.

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
  • Method and system for constructing time overhead model based on kernel function
  • Method and system for constructing time overhead model based on kernel function
  • Method and system for constructing time overhead model based on kernel function

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are the Some, but not all, embodiments are invented. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

[0029] see figure 1 , this embodiment discloses a method for building a time overhead model based on a kernel function, including:

[0030] S1. Construct the warp memory access parallelism WMPD according to the hardware structure of the GPU and the parallel program to be transplanted, and calculate the thread block block in the kernel function before the parallel prog...

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 Kernel function based time cost model construction method and system. The method includes: dividing the cost time of a parallel program Kernel function into distribution time of thread blocks and execution time of the Kernel function, analyzing the distribution time and the execution time, and constructing a corresponding time equation, wherein the execution time fully considers overlapping between calculation and access and storage, and various possible situations, and the execution time of the Kernel function is represented by two parallelism degree parameters; and finally, performing superposition on the distribution time and the execution time to acquire the final Kernel function time cost model. A user can acquire the running time of an application program on a GPU(global block scheduler) through the model, and then mass users can timely analyze a program to be transplanted and reasonably arrange work, effective utilization of the GPU is facilitated, and economic and time loss can be avoided.

Description

technical field [0001] The invention relates to the field of parallel and distributed computing, in particular to a method and system for constructing a time overhead model based on a kernel function. Background technique [0002] With the continuous improvement of GPU performance and programmability, GPUs are increasingly used in general-purpose computing, and GPUs with such characteristics are called general-purpose GPUs. There are more and more systems accelerated by general-purpose GPUs, and this heterogeneous system in which CPUs and general-purpose GPUs cooperate has received more and more attention in the field of high-performance computing. [0003] The application development and execution efficiency of user programs on heterogeneous systems have always been the bottleneck restricting its development. The performance of the GPU is closely related to many factors, including the characteristics of the program itself and the characteristics of the GPU. The characteri...

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): G06F9/48G06F9/50
CPCG06F9/4843G06F9/5044G06F2209/5018
Inventor 王剑秦王佳弘林枝宏孙瑞志
Owner CHINA AGRI UNIV
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
Eureka Blog
Learn More
PatSnap group products