Software Re-engineering
In-house software or Open Source software is reinvented into high-performance, high-efficiency software based on highly efficient methodologies and S/W engineering methodology.
The non-structed or object-oriented programming language is transplanted into a structured programming language, and then re-generated into the GPU code.
The non-structed or object-oriented programming language is transplanted into a structured programming language, and then re-generated into the GPU code.
Application of S/W Engineering Methodology
Apply Water-fall model
Requirement analysis → Planning → Design → Detailed design → Conducting a test → Maintenance
Quality assurance through verification is considered.
Requirement analysis → Planning → Design → Detailed design → Conducting a test → Maintenance
Quality assurance through verification is considered.
Code Migration
Rebuild In-house or Open-source code with code optimized for high-efficiency applications.
-Fortran C UDA/HIP
-C++ CUDA/HIP
-Matlab C CUDA/HIP
Various P/L transplants, such as Python and JAVA.
-Fortran C UDA/HIP
-C++ CUDA/HIP
-Matlab C CUDA/HIP
Various P/L transplants, such as Python and JAVA.
GPU Centric Methodology
To overcome performance degradation due to performance differences between CPU and GPU and data bottlenecks, the programming methodology computed only by GPUs ground.
Performance Improvement Case
SPH(3kinds) of fortran to C/CUDA : 200 to 1,700 times
OLED analysis of Matlab to C/CUDA : 10,000 times
Plasma analysis of C++ to C/CUDA : 200 times
OLED analysis of Matlab to C/CUDA : 10,000 times
Plasma analysis of C++ to C/CUDA : 200 times
