Hi,
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
cudaMallocHost()
cudaMalloc() -
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.