I have few queries regarding the memory allocation methods provided using the cuda APIs
Does cudaMallocHost or cudaHostAlloc APIs allocate contiguous physical memory from the Host.
Is there any limitation with the amount of memory that can be allocated with
Can we provide the memory address obtained through cudaMallocHost() and cudaMalloc() to copy the stream from the camera device connected through PCIe interface using some DMA controller . Here this means that the address provided to the DMA controller is Physical address and the allocated memory is contiguous.