We need to support both NVIDIA CUDA (e.g., NVDIA SHIELD tablet) and OpenCL (Microsoft Surface Pro 3). However, presently, apparently the OpenCLLink only supports CPU, not the GPU, on Intel chipsets. Are there any beta or open-source work-arounds that would allow us to use the GPUs on both platforms?
We have a high-volume medical image processing and DNN application that needs to runs on a variety of devices.