Pages

Sunday 8 January 2012

BGI ... GPU ... NGS...???


First of all, let me state that I'm not a fan of video games. However it seems that video games are somehow starting to heavily influence the world of genomics and NGS. Last fall BGI scientists have been able to shrink from 4 days to 6 hours the time needed to completely analyze the sequence data describing a human genome. The result was achieved using servers built around graphics processing units, or GPUs, the sort of processors that were originally designed to draw images on personal computer and are heavily used by modern video games. However, don't get too excited just yet. The latest generation gaming PC in your apartment most likely is not powerful enough to analyze your grandma or grandpa genome, at least from what I see about the hardware used by BGI (see the NVIDIA(R) Tesla(TM) GPU-based server farm pictured below).
You can read the entire story in this article recently published on Wired. NVIDIA, the GPU chip-maker collaborating with BGI, also has a press release on this achievement. SEQanswers has a number of posts about GPU based systems for NGS.




1 comment:

Stefano Calza said...

Indeed I installed few months ago a Nvidia graphic card CUDA/OpenCL enabled (CUDA is the programming language for parallel computing on Nvida GPUs). Still under used here but definitely useful when coming to heavy computation. Only issue is that very few programs take advage of it. Unless you program C/CUDA yourself...