<p>You have the fastest Intel/AMD processor in a 500 mile radius thanks to your custom built quad-core, liquid nitrogen cooled, overclocked 5.0Ghz CPU monster. Prepare to be summarily beat down, computationally speaking, by the kid next door who just bought the latest Nvidia GPU to play WOW at 80fps. Video cards, fueled by the gaming industry, have leap-frogged (pun intended) the processing power of the general purpose CPU for certain computational tasks. The rise of the multi-processor based general purpose GPU (GPGPU) platform is taking academia by storm due to its low costs and low barrier to entry into modern day supercomputing. The security community has already embraced the GPU for heavy lifting as have other fields especially when coupled with the sleek marketing efforts by Nvidia and their CUDA development environment, and competing GPU computing platforms from ATI and OpenCL. This 20 minutes session will chronicle the rise of the GPU in high performance computing and will highlight GPU vs. CPU benchmarks of well known security tools including: aircrack (10x speed-up), Pyrit (8x), CUDA Multiforcer, BarsWF MD5 cracker (3x), RainbowCrack multi-GPU CUDA version, and more. Finally, links and tips regarding implementing CUDA in BackTrack 4 are shared.</p>
<p>Collin Brack is a healthcare informatics and medical imaging consultant with experience in computational clusters. He works in academia where he focuses on high performance computing with medical physics researchers. His latest cluster is based on high-end graphics processors to achieve performance gains previously only available to multi-million dollar big iron. He has published and presented on the topics of system design, grid computing, and disaster recovery.</p>