nVidia’s nVision visual computing event only lived to see its inaugural outing in 2008. Earlier this year, nVidia announced that it had cancelled the show for 2009, citing the recession and bad economic times as being "inappropriate". Our take was that nVidia was forced to cancel the event as a responsibility to stockholders, but in all honesty – that decision was a bad one.

According to the information we have at hand, nVidia shelled out around six million greenbacks to host its allegedly annual conference. Its replacement is something that will be appreciated by journalists and people alike – the LAN party is out, and only the serious business part remains. The name of the conference is somewhat uninspiring, at least coming from the master of renaming – GPU Technology Conference.

GTC is basically three events held at the same time: The Emerging Companies Summit, GPU Developer Summit and the nVidia Research Summit. During nVision 08, there were so many things to address that we barely got an opportunity to take a glance at the Emerging Companies Summit, a conference where a lot of young companies showcased fascinating stuff. This time around, there will be plenty of time to give deserved media attention to small companies that are solving the questions of tomorrow – all using GPU technology.

nVidia ditching CUDA?

Personally, I can’t wait for GPU Developer Summit, with tons of technical presentations and panels targeting developers of consumer, professional and HPC applications using DirectX, OpenCL and OpenGL APIs, C++, FORTRAN and other programming languages. Notably absent from the press release was any mention of CUDA, nVidia’s Compute Unified Device Architecture.

Under the new chief scientist at nVidia, you can expect a lot of moves such as these in order to clear the confusion about just what CUDA is. In the past, we saw nVidia’s own press releases sending a wrong message, putting CUDA in the same bracket as DirectX Compute and OpenCL, which was well – plain wrong. Bear in mind that CUDA is nothing else but a C languange with nVidia’s own extensions and the code is being compiled mostly through PathScale’s Open64 compiler e.g. optimized for GPGPU or GPU Computing architecture. CUDA is NOT an API targeting DirectX Compute, OpenCL etc. Translated into plain English, CUDA’s SDK will remain intact after the arrival of OpenCL and DirectX Compute and other Application Programming Interfaces, e.g. APIs. ;)

The GPU Technology conference is being held at the Fairmont Hotel in San Jose, California. The official conference pages are now up: nVidia’s GPU Technology Conference.