Methodologies and Performance Impacts of General

useumpireSoftware and s/w Development

Dec 2, 2013 (4 years and 5 months ago)


Methodologies and Performance Impacts of General
Purpose Computing on GPUs

Shinichi Yamagiwa


Graphics Processing Units (GPUs) was applied to graphics applications to
implement realistic perspectives of virtual scenes especially in entertainment
market. Due to the demands from the market for creating more high definition
scenes that simulates physics phenomenon correctly in cinemas or
visualization applications, the last decade has promoted drastic performance
improvement of GPUs. In the very near

future it will achieve TFLOPS
performance. Researchers from the high performance computing field focus
on the incredible high performance, and recently they are going to apply the
GPU horse power to general purpose applications. This field is called

General Purpose computing on GPUs (GPGPU). When we program any
application in a GPGPU environment, it is hard to extract the potential
performance of GPUs because originally it is designed to process graphics
applications. Therefore, the programmer sh
ould master the hardware
architectural and the processing conceptual aspects of GPU itself. This talk will
focused on the recent methodologies on programming in GPGPU environment
introducing advanced programming methods. It will also discuss performance
pacts presenting recent research results from Caravela Projects leaded by
the speaker.