Hi,
what is the maximum theoretical memory throughput on C6678 when doing calculations on the different memories (L1, local L2, L2 from another core and external memory), assuming a continuous iteration over an array of values (e.g. floats). How do the operating frequencies influence this value? Is there an influence on this value once there are multiple simultaneous requests from different cores? Is there a difference between reading and writing?
I could simply measure this value by implementing a simple loop and measuring the time for that loop, but in my case I'm interested in the theoretical limits.
Thanks,
Roman