Course Homepage: GPGPU (Q2 2004)

Description

The evolution of consumer graphics cards in recent years has introduced the GPU (Grapics Processing Unit) as a flexible vector-processor capable of coloring, shading etc. in parallel. This evolution has been driven mainly by demands of more realistic effects by the computer gaming industry.
However, the built-in parallelism is also suitable for other tasks than visual effects: We will view the GPU as a general purpose vector processor and look at how parallel calculations (including physical simulation) are implemented. As examples we take real-time simulation of fluid dynamics and deformable virtual models of human anatomy.
The course provides an introduction to GPU-programming and traditional visual effects. However, the main focus will be on working with the GPU as a general purpose processor for physical simulation.

Credits:

5 ECTS

Evaluation:

Individual written report, 13 scale.

Compulsory programme:

Present solution to the weekly assignments in groups

Lecturers:

Jesper Mosegaard
Thomas Sangild

Place and time:

Tuesday 14-15
Thursday 11-13

Codd-121, Finlandsgade 24-26

Plan:

(25 October - 10 December)

Date	Topic (Thursday)	Material	Assignment (Tuesday)
Week 1: 26/10	Introduction slides	Code: Fixed function pipeline in cg Printed Handouts: The OpenGL fixed function pipeline A compilation of notes for the "Programming Graphics Hardware" course, Eurographics 2004 Additional recommended reading: Cg users manual pages 1-61, Appendix B. Tomas Möller, Real-Time Rendering: chapter 2, The Graphics Rendering Pipeline
Week 1: 28/10	GPU programming Introduction to GPU programming Per pixel lighting Parallel programming slides 01 02	Code: Per pixel lighting Simple bumpmapping Printed Handouts: Assignment for week 3 (09/11-04)
Week 2: 04/11	GPGPU programming resources Resources in hardware Invoking computation The memory model slides	Code: RenderTexture RenderTextureMisc Printed Handouts: GPUGems Programming Techniques, Tips, and Tricks for Real-time Graphics. Edited by Randima Fernando: chapter 28 and 37
Week 3: 09/11, 11/11	Higher level functionality Pointer, 1d and 3d arrays Scatter and reduction Sorting and searching GPGPU languages branching slides	Final Project info Bump-map discussion from last	Assignment from 28/10 Assignment from 04/11
Week 4: 16/11, 18/11	Spring Mass systems + Raytracing slides	Printed Handouts: The Ray Engine, Carr et. al Ray Tracing on Programmable Graphics Hardware, Purcell et al. A GPU Accelerated Surgical Simulator for Complex Morphology, Jesper Mosegaard and Thomas Sangild Sørensen	Assignment from 11/11
Week 5: 23/11, 25/11	Solving linear systems + fluid dynamics slides 01 slides 02	Printed Handouts: Linear algebra Operators for GPU implementation of Numerical Algorithms, Kruger J. and Westermann R. Sparse Matrix Solvers on the GPU: Conjugate Gradients and Multigrid, Bolz J et. al. Fast Fuid Dynamics Simulation on the GPU, Harris M.
Week 6: 30/11, 02/12	Follow up on last lecture (tuesday) slides	...	Assignment from 26/11 Assignment from 26/11 Remember, this is for thursday 2/12
Week 7: 07/12 09/12	Meso structure rendering (tuesday) slides Evaluation, Final Projects, unsettled business (thursday) slides 01 slides 02	Lifeng Wang et al. "View-Dependent Displacement Mapping" Terry Welsh. "Parallax Mapping with Offset Limiting: A Per-Pixel Approximation of Uneven Surfaces"

Participants:

19984373 Niels Jeppe Anders Nielsen
20001263 Rasmus Witting Larsen
20001467 Bent Bisballe
20001683 Søren Gjellerup Christiansen
20001995 Karsten stergaard Noe
20002177 Peter Gade Jensen
20002751 Lars Ole Simonsen
20022647 Michael Bræmer Nielsen
20023690 Jonas Larsen
20012938 Peter Trier
20011373 Thomas Mølhave
19982078 Nicolai Lundgaard
20023502 Søren Skov
19960197 Jesper Fruegaard Andersen
19970541 Aske Simon Christensen
19991875 Michael Westergaard
19971799 Allan Rasmusson
19990498 Troels Bjerre Sørensen
19992541 Daniel Nielsen
20002482 Niels Carsten Thrane

Acknowledgements

Some slide-material originate from:
SIGGRAPH 2004 GPGPU COURSE
Visualization 2004 TUTORIAL

Other courses on GPGPU related subjects:

Graphics Architecture, Spring 2004
Hacking the GPU

Local machines with Radeon 9800:

Zuse 114
nussebaum
hyldgaard
dickow

Zuse 127
All machines...