TDT 24: Parallel Environments and Scientific Computing

Fall 2008

Instructor:
Dr. Anne C. Elster

TIME & PLACE: Thursdays 14:15-16:00, IT-vest 454

Studiepoeng: 3,75

Final exam: Oral (muntlig) on Dec. 3, 2008

Innhold/Contents:

This module will cover the most well known programming environments, such as GPGPU programming, MPI, OpenMP and pthreads. The module will also cover numerical algorithms and methods reflecting the students' choice of Master pre-project.

SYLLABUS

"Proceedings of the IEEE, Vol 96, No 5"

except:

Convergence of Recognition, Mining, and Synthesis Workloads and Its Implications
Graphics Processing Units for Handhelds
Focusing on Regional Resilience

That is, the following papers from the above IEEE journal issue are covered:

Scalable Programming Models for Massively Multicore Processors (presented by Jan Christian Meyer) Jan Christian's presentation
GPU Computing (presented in part by Robin Eidissen and Elster)
Rise of the Graphics Processor (as above)
Challenges and Opportunities in Many-Core Computing (presented by Safrudin) Safrudin's presetation (to be added)
Self-Configuring Applications for Heterogeneous Systems (presented by Daniele) Daniele's presentation
Database Optimizations for Moden Hardware (presented by Even) Even's presentation

In addition the following material will be covered:

Master Project Utilizing GPUs on Cluster Computers by Leif Christian Larsen (including The Lapped Orthogonal Transform (LOT) presented by Daniel H)
A performance study of general-purpose applications on graphics processors using CUDA (presented by Rune H) Rune H's Presentation
Optimization Principles and Application Performance Evaluation of a Multithreaded GPU using CUDA (presented by Aasmund Herikstad) Aasmund H's presentation (to be added)
OpenCL - What's in store? (presented by Olav) Olav's presentation
From Dr. David M. Young's Iterative Solution of Large Linear Systems (published by Dover):
- Ch 1: Introduction (incl. Jacobi & Gauss-Seidel methods) (presented by Elster)
- Ch 2: Matrix Preliminaries (presented by Elster)
- Ch 3.3-3.5 Iterative solutions of large linear systems (presented in part by Owe) Owe's presentation
Link to further background material and PETSc:
- http://www-fp.mcs.anl.gov/petsc/
- "Numerical Libraries and Tools for Scalable Parallel Cluster Computing", Brown, Dongarra & Trefethen Avaialble at: http://www.netlib.org/utk/people/JackDongarra/PAPERS/libraries-tools1.pdf
"Tom, Dick and Mary Discover the FFT" , by D.R. Deller Jr, IEEE Signal Processing Magazine, April 1994 (presented by Aasmund Elduset) Eldhuset's presentation
From Numerical Simulations in Fluid Dynamics: A Practical Intro.: by Griebel, Dornseifer & Neunhoeffer, pp 4-41, SIAM 1997 (presented by Eirik and Henrik) Eirik and Henrik's presentation (to be added)
- See also Color figures from Griebel's home page at http://wissrech.iam.uni-bonn.de/research/projects/cfd

SCHEDULE

Mon. Aug 18 -- SEMESTER START NTNU
Thurs. and Fri. Aug. 21-22 Meetings with several students re. projects -- no class TDT24
Fri. Aug 29 -- No class, but reading material handed out, inlcuding IEEE journal issue to be emphasized in this course this fall.
Fri. Sep 05 -- Brief intro meeting; Tips on how to write a report
Thu. Sep 11 -- one-hr meeting -- set up course scheldue, brief intro to GPGPU
We generally met on Thursdays or Mondays, covering papers as indicated below.

Student Participants & main projects


	Eirik-Ola Aksnes -- Master of Science student, IDI, NTNU Master project topic: HPC oon GPGPU in collaboratio Center for Intergrated Operations in the Petroleum Indistry

	Åsumund Eldhuset -- Master of Technology student, IDI, NTNU Master project topic: HPC for GPU National IT-student (Årets IT-student) 2008

	Olav Fagerlund -- Master of Technology student, IDI, NTNU Master project topic: OpenCL

	Daniel Haugen -- Master of Technology student, IDI, NTNU Master project topic: HPC on GPU

	Åsmund Herikstad -- Master of Technology student, IDI, NTNU Master project topic: Sound processing on GPU

	Henrik Hesland -- Master of Technology student, IDI, NTNU Master project topic: HPC on GPU for seismic applications in collanborations with Center for Intergrated Operations in the Petroleum Indistry.

	Rune Hovland -- Master of Technology student, IDI, NTNU Master project topic: CPU - GPU performance metrics -- how to biult the best possible system configurations with modern components.

	Owe Johansen -- Master of Technology student, IDI, NTNU Masters thesis topic: Restarting large HPC seismic applications -- in collaborations with StatoilHydro

	Safrudin Mahic -- Master of Technology student, IDI, NTNU GPGPU project with Numerical Rocks (supervised by Prof. Lasse Natvig) >

	Even Skaland -- Master of Technology student, IDI, NTNU Project on MicroRNAs and transcriptional control (supervized by Dr. Pål Sætrom) details >

	Daniele Spampinato -- Master of Technology student, IDI, NTNU Master project topic: Linear Programming on GPGPU