Reading about Nvidia’s and ATI/AMD’s recent announcements about General Purpose GPU computing, it struck me that the WURST/SALAMI alignment algorithm would be a perfect match for the matrix multiplying monsters that are found on the latest graphics cards. Based on the performance of other algorithms in the literature, such as the Folding@home project’s 20x-30x speedup, I’m guessing that it should easily be possible to do a SALAMI search on the entire PDB in less than one minute if the alignments were computed on the GPU. Unfortunately, I can’t afford to buy one of the new graphics cards necessary to try it out.
-So if you’re interested in seeing seriously fast protein structure alignments, and you have a spare Nvidia 8xxx series graphics card, or you can give me a user account on a machine with such a card on a fast network, then please get in touch. I’m itching to give this a try…
Archive for May, 2007
Computing Protein Structure and Sequence Alignments on a GPU
Tuesday, May 15th, 2007new numbers of the beast.
Wednesday, May 2nd, 2007If they keep going like that, they’re going to have to start worrying about running out of keys soon…
