NVIDIA's Jen-Hung Huang spoke this week at the GPU Technology Conference (GTC) on all things graphics processing, revealing that a new family of Tesla GPU units built on NVIDIA Kepler GPU computing architecture would be released inside 2012. What's being released specifically are the new NVIDIA Tesla K10 and K20 GPUs, both of them "built to handle the most complex HPC proclaims in the world. Kepler here is three times as efficient the architecture that came before it, NVIDIA Fermi, here once again establishing NVIDIA as a head in the whole world's computer processing environment.
First the Tesla K10, as Huang noted today at GTC, "is dedicated to seismic analysis where the most valuable resource is bandwidth - it has three times the single precision of today's Fermi Tesla and 1.8 times the memory bandwidth." On the second Tesla upgrade here with the K20, Huang noted that it had "3 times double precision of the Fermi, including Hyper Q and dynamics parallelism for stuff like physics, quantum chemistry, and computational finance.
Stick with us throughout the day for the rest of the action as Jen-Hung Huang and the rest of the NVIDIA crew reveal new technologies across the board for graphics processing. Earlier today Huang spoke on the explosive growth of CUDA for NVIDIA and how this will affect the future as well. Check out our NVIDIA timeline below and get pumped up for the future of NVIDIA architecture!