AMD's only big argument against CUDA and NVIDIA's way to compute was that NVIDIA's hardware doesn't support OpenCL 1.0, an open specification from the Khronos Group. So as of today, NVIDIA officially supports it, and we've reported on many occasions that NVIDIA does and wants to support any way of parallel computing. And OpenCL, Open Computing Language, is a new computer API that allows developers to program their software for massive parallel computing and use the enormous power of a GPU to help speed up the task. And NVIDIA started this parallel computing with CUDA, although technically AMD was the first to use its GPU to encode video with Avivo and this late in 2005. - FudZilla

NVIDIA Can Handle OpenCL 1.0 Parallel Programming

Now NVIDIA is putting a lot of effort into parallel computing, as this is a brand new market that is very profitable, and if NVIDIA makes more people use GPU for non graphics stuff, this will mean that the company's market will immediately expand. NVIDIA's CUDA has support for other industry standard languages such as C, Java, Fortran and Python. NVIDIA also claims that 100 million of its GPU's are CUDA and/or OpenCL capable and can boost power to new hights. Maju Hegde, coming from AGEIA who is now general manager of CUDA at NVIDIA, is happy with its arrival and is happy that even Apple has embraced OpenCL.

He believes that with Apple's recognition that the GPU will now play an essential role in these consumer applications is a significant milestone in the history of computing. Though Neil Travett from NVIDIA, who is now leading their embedded content, holds the position of chair of the OpenCL working group at Khronos. We just hope that OpenCL won't end up as open standard OpenGL, that has a single supporter called Jon Carmack in the gaming world. We just fear that Microsoft will again play a judging role and we are not aware of what Microsoft wants to support today, but let's hope they do support it as well.