I still remember my excitement nearly nine years ago, when I first came to work at NVIDIA and I learned that NVIDIA had a super-secret project to build a GPU architecture with dedicated support for general-purpose parallel computing. I had been doing GPGPU research for a few years and I wasn’t sure of its future, but this investment showed that NVIDIA considered GPU Computing a key technology. That secret project became the Tesla architecture, launched in late 2006—the first architecture to support CUDA.
It’s hard to believe that the CUDA platform is already 5 years old, and boy has it come a long way. The performance and efficiency of software built on CUDA, combined with a thriving ecosystem of programming languages, libraries, tools, training, and service providers, have helped make GPU computing a leading HPC technology. The GPU Technology Conference, to be held May 14-17 in San Jose, will be a landmark occasion for CUDA: the announcement of CUDA 5. CUDA 5 and the Kepler GPU architecture don’t just increase application performance; they enable a more powerful parallel programming model that expands the possibilities of GPU computing, and language features that improve programmer productivity. I’m excited to have the opportunity to introduce CUDA 5 at GTC, where I’ll tell you all about these revolutionary features and give insight into the philosophy driving the development of new CUDA hardware and software.
So if you want to get the inside scoop on this revolutionary CUDA release, come see my talk, “CUDA 5 and Beyond”, at 4:00pm Tuesday, May 15 in Hall 1 of the San Jose Convention Center. Add it to your session schedule now. If you can’t make it, you’ll be able to listen to the recorded session afterwards.
By the way: CUDA 4.2 is now available, with support for Kepler GK104 (such as GeForce GTX 680). Download and try it today!