This is free, open-source (GPL v3 license), Multiple Back-Propagation neural network trainer. It’s an incredible use of CUDA, with claim of 175X speed-up (GTX 280 vs. unknown Intel CPU).
Great work!
This is free, open-source (GPL v3 license), Multiple Back-Propagation neural network trainer. It’s an incredible use of CUDA, with claim of 175X speed-up (GTX 280 vs. unknown Intel CPU).
Great work!
DARPA are giving $25 millions to nVidia to research exascale computer, that should offer 1000X performance-level of actual existing supercomputers in 2018, using future generations of nVidia’s Fermi GPU.
Again, this show the incredible momentum that GPGPU is gaining, US government funding nVidia to research new architectural evolutions, to raise performance-level, reliability, and programmability of their GPU.
Fixstars announced Yellow Dog Linux distribution for CUDA Development. Free for University, but $400 for any other use. It’s sad it’s so expensive.
I played with the idea to do a similar distro, based on Ubuntu, at some point. Maybe it’s a project I should re-consider, but with a FREE distribution!