Skip to content

Latest commit

 

History

History
45 lines (37 loc) · 1.34 KB

README.md

File metadata and controls

45 lines (37 loc) · 1.34 KB

CUDA Introduction

University of Pennsylvania, CIS 565: GPU Programming and Architecture, Project 1

  • Xinyue Zhu
  • Tested on: Windows 10, i5-5200U @ 2.20GHz 8GB, GTX 960M

MY README

Include screenshots, analysis, etc. ###Part1: *1)change the number of planets N： N=7000 20fps

N=5000 40fps

N=2000 60fps

while N=2000: while N=5000: while N=7000: From above we can see, when N goes up, usage of GPU goes up,fps goes down,running time goes up. Because when the number of planets increase the calculation of gravity increase. *2)change the block size: while N=7000, block size=56,128,256,gridsize=126,56,28:their fps are all around 20. and no much difference in the time. *3)at project2,because the size of the mat is too small block size do not affect much of the timeline and GPU block size=1 block size=1000 ###Part 2: If the large amount of data is simple and do not have much relevance with each other, the GPU will perform better. CPU has cache, it can store and use different kind of data to calculate.