I am very new to CUDA so I am sorry if I am asking about things which are obviously impossible to someone more well versed with CUDA.
I need to use a GPU and CUDA to do some basic calculations on a huge amount of data which is coming into the machine either over a fiber connection or a GigE connection. Is there anyway to have this data be streamed directly to the GPU’s memory and executed on instead of having it come into the machine stored to main memory in some sort of buffer and then be transferred to the GPU’s memory to be worked on. This process of getting the data and running the computations needs to happen in real time. So to recap, data will be streaming in at a very high rate and needs to have some simple calculations done on it how can I best go about making this happen.
Any topics or general advice anyone has would be awesome.