This demo utilizes the DMA engine to perform a memcpy. Initialization of DMA
is handled either on the first dma_memcpy() call, or explicitly via
init_dma_memcpy(). Using a preprocessor switch, the program will either
perform a simple systick-based comparison between dma_memcpy() and standard
memcpy() for a given buffer size, or it will launch a a single-producer,
single-consumer FreeRTOS-based program where data
is transferred between two threads via a shared buffer.
Currently the target architecture is ARM Cortex-M4, and the included demo is configured for TI's TM4C1294 Launchpad. For now it serves as a proof-of-concept, but eventually it will be put into a multi-platform library.