Thank you for the quick reply. From looking at the DW documentation, is it possible to quickly identify whether a particular API call will run on the CPU or the GPU?
Let me provide some background for you. Our team is planning to implement a non-neural net based object detection application on Drive PX 2. So we are going to have some sections of code that are sequential in nature, for which we plan to use the CPUs (Denvers + ARMs), while other sections of the code could be run in parallel, for which we plan to use the dGPUs. One of our major concerns is the inter-process communication and what DW can do to help with that. We could create our own source code to handle this, of course, but we were hoping that DW could help fulfill this need and allow us to field our application quickly. Do you know if this kind of thing is possible?