NVIDIA Seeds OpenCL Driver to Developers

NVIDIA just released an OpenCL* driver (for GPU computing) to developers today. Developers will have access to the driver to validate and test their current projects. Previously, it was possible to use NVIDIA GPUs for crunching math with CUDA*, a C-like language. The issue was that NVIDIA was pretty much the only party to have a CUDA driver, so many developers were waiting for a more “open” solution like OpenCL or DirectX Compute (for gaming). We guess that Apple was a driving force behind NVIDIA’s progress towards an OpenCL driver as the next Mac OS is expected to heavily use GPU Compute. At this point, NVIDIA is basically the only game in town, when it comes to GPU Computing.

*OpenCL is an API defined by many companies, including NVIDIA, AMD and Intel.

*CUDA used to be the name for a C-like language that NVIDIA created/promoted, but it is now used to name the GPU Compute “Architecture” (the hardware changes required for general purpose computing). The definition change is most likely driven by Marketing as OpenCL and DirectX compute might evetually take over CUDA (the language) in popularity. By naming the architecture “CUDA”, the name will stick for quite some time, and along with it, the brand that was created will remain a (psychological) differentiator.