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

Many-core application performance evaluation method based on domestic supercalculation micro-architecture

A technology with application performance and super-computing technology, applied in computing, instrumentation, electrical and digital data processing, etc., it can solve the problems of lack of system and comprehensiveness, and achieve the effect of good versatility, high accuracy, and reduced difficulty in use.

Active Publication Date: 2021-12-24
SHANDONG COMP SCI CENTNAT SUPERCOMP CENT IN JINAN
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In view of the above problems, some specifications in the existing technology can realize accurate and effective evaluation of the performance of some codes of the program, but there is no systematic, comprehensive, and effective performance evaluation model for general many-core codes; especially in the many-core In the optimization process, there are various optimization methods, and sometimes specific analysis of specific scenarios is required, which brings great challenges to the establishment of performance evaluation models

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
  • Many-core application performance evaluation method based on domestic supercalculation micro-architecture
  • Many-core application performance evaluation method based on domestic supercalculation micro-architecture
  • Many-core application performance evaluation method based on domestic supercalculation micro-architecture

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0074] A performance evaluation method for many-core applications based on domestic supercomputing micro-architecture, the specific method is as follows:

[0075] 1) Model input includes Lat_SPM, n_v, n_scalar_i, n_scalar, n_arr_i_get, n_arr_ge, n_arr_i_put, n_arr_put, d_size, n_iasm, n_iasmi, n_dasm, n_ddiv, n_cmp, n_1d, n_2d, n_3d, n_i_siz, n_size, n_cpe_m, n_com, , T_cpe, n_cmp, n_gsize, n_psize, T_g1, T_p1, n_core;

[0076] Among them, Lat_SPM is the delay of accessing from the core office memory, and the unit is cycles; n_v is the number of array variables; n_scalar_i is the number of integer scalars; n_scalar is the number of floating-point scalars; n_arr_i_get is the read-in integer array n_arr_get is the number of floating-point arrays read in; n_arr_i_put is the number of output integer arrays; n_arr_put is the number of output floating-point arrays; d_size is the size of a single array; n_iasm is the loop Outside, the number of integer addition, subtraction, and mul...

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 many-core application performance evaluation method based on a domestic supercalculation micro-architecture. Aiming at main processes such as thread group task distribution and ending, data transmission, data calculation, slave core synchronization and communication and the like in an Athread parallel programming mode of a Shenwei domestic super computer system, a corresponding time operation model is fitted according to a large amount of actual example operation data of different scales, and the error is within 10%, so that the time operation efficiency is greatly improved. And the operation performance of the many-core application program can be effectively evaluated.

Description

technical field [0001] The invention specifically relates to a many-core application performance evaluation method based on a domestic supercomputing micro-architecture, and belongs to the technical field of supercomputer performance evaluation. Background technique [0002] With the development of domestic supercomputers, more and more attention has been paid to the performance of "applications" on domestic supercomputer platforms. Therefore, effective many-core performance evaluation methods are urgently needed to assist developers in supercomputer performance optimization. [0003] Each core group on the CPU chip of domestic Shineway series supercomputers contains 1 computing control core (master core) and 1 computing core array (slave core array), and the computing core array is controlled by 64 computing cores (slave cores), array The physical space of all core groups is uniformly addressed, and the operating frequency is 1.45GHZ. Both the computing control core and th...

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): G06F11/34
CPCG06F11/3447G06F11/3452Y02D10/00
Inventor 杨美红郭强曾云辉庄园张海红
Owner SHANDONG COMP SCI CENTNAT SUPERCOMP CENT IN JINAN
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