Visualization1

From SuperMe

Jump to: navigation, search

Contents

High-resolution Data Visualization

Project by: Jason Monk, University of Maine
Advisor: Dr. Bruce Segee

Abstract

Displaying a 3D object on a 2D screen takes a large number of algebraic computations, so over time GPUs (Graphics Processing Units) have developed into processors that can do lots of calculations in parallel, similar to the design of a supercomputer. CUDA is the architecture recently implemented in NVidia graphics cards that allows users to compile and run C/C++ code in parallel on their GPU. The added computational power from the GPU(s) allows problems that could normally require a supercomputer to run on a graphics card in comparable time. The goal of my research will be to find ways to increase the performance of the high-resolution display wall found in the innovation center by harvesting unused GPU computational power. Of particular interest will be a CUDA enhanced version of virtualGL allowing faster rendering at high resolutions.

Weekly Reports and Schedule

  • Week 1 (06/1-06/07):
    • SuperME Program Orientation
    • Introduction to Supercomputer programming and concepts
  • Week 2 (06/08-06/014):
    • Research in current display wall system and software
    • Testing on memory read/write speeds using CUDA [1]
  • Week 3 (06/15-06/21):
  • Week 4 (06/22-06/28):
  • Week 5 (06/29-07/05):
  • Week 7 (07/13-07/19):
  • Week 8 (07/20-07/26):
  • Week 9 (07/27-08/2):
  • Week 10 (08/3-08/07):
    • Written report due - PDF
    • Poster due - Thursday, August 6 - SLA (Scribus) - PDF
    • Final symposium - Friday, August 7




CUDA Tutorial

Presentation - PPT - PDF
Fireplace Example - ZIP