Alistair Rendell

Background info for 2009 honours projects using novel hardware.

IBM Cell

IEEE Micro article on IBM Cell Processor - good overview/starting point.
IBM Cell web page - lots of reference material. Look under publications.

NVIDIA GPU (we have the GTX8800)

CUDA home - programming guide available under "developing with CUDA" and documentation tab.
Nvidia developer site

General Papers/Material

Intro from Leach on maths for molecular modelling
DeFabritiis paper on accelerating molecular modelling calculations - I just happened to find this, I've not read it yet - thought it might be interesting.

Background on Molecular Mechanics/Dynamics (for Ben)

Rough aim is to develop an MD code for either the NVIDIA GPU or IBM PS3 that will run small systems very fast. Preference is for PS3 because of our links with IBM. We need terms for bonds, angles, torsions, van-der Waals and electrostatics. We will need to consider the various water models and have our system surrounded by a dielectric media - but more on that later....

Leach MD part1
Leach MD part2
Paper about NAMD on GPU - for reference at this point, ie not expected to read and understand.
The NAMD code that was used in the above paper

Background on Quantum Chemistry (for James)

Goal is to develop code for evaluation of the sort of integrals relevant to quantum chemistry on either the NVIDIA GPU or PS3. Most likely the PS3. First task is to understand enough quantum chemistry to know how the integrals arise and then to understand the algorithms used to evaluate them. We need to base our work around some existing code. Next task will be to assess the viability of using the Psi code. This is an opensource program written in C++ in an OO fashion - but I don't have experience with this yet - so we both need to check it out!

part of leach chapter on quantum chemistry - this will give you overview of quantum chemistry methods. I don't expect you to follow everything. We will focus on the Hartree Fock method. The general equations we solve iteratively are given on page 62. We are particularly interested in formation of J and K in eqn 2.158. This involves evaluation of the integrals over gaussian functions located on (upto) four centres (usually atomic nuclei). For details of how to evaluate the integrals see the Gill paper.
Gill paper. Lots of detail, I don't expect you to follow everything. Section 3.2 is the basic integrals of interest.
Yasuda paper on evaluation of integrals on GPU. This uses a slightly different algorithm to that given in Gill paper. This paper is just for reference at this point.
Martinez paper on evaluation of integrals on GPU. Also just for reference at this point.
The Psi code that we will consider using for this project
Shavitt paper on Guassian functions