Hello,
We have a question on the lab2, it is not clear how we should include the accelerator in the project.
From our understanding, we believe that :
-An accelerator is composed of 2 DMA (or one depending on how we consider a DMA) and these DMA talk to a code that makes the calculation we want.
-The goal of DMA is to work in parallel of the CPU without using it anyway (except to set the DMA how we wish by the slave part of it)
That said, from the software in C point of view, it means that the way we should use the DMA is :
1) give to the DMA the address in memory where we write the word to process
2) read the result of the process in another address (for instance the one just after)
We would like to know what should be accounted for in the profiling because we do not understand if we should add an IRQ or another real-time primitive in order to know when the process finished (this would in reality be very long compared to just do the calculation on the CPU and we then do not understand the point of the exercise). Maybe we should do something differently and it this case we would like more details about it.
Thank you in advance
Andy Piantoni