In my application the tasks are partitioned to coreA and B. In a scenario, on coreA video is grabbed from a camera and displayed on a LCD (data is transferred by DMA) while CoreB performes comprehensive calulations (with programmed IO).
When i compare this scenario with a scenario where coreB performs the same calculations while coreA sits idle, the processing performance of coreB in the latter scenario is about 30 percent higher.
Of corse, both cores share the sdram, but as the memory sections of them do not overlap, i hoped that coreA and coreB could run independently with full performance each.
May this be a matter of a wrong configuration of the sdram-controller?
Would be greatful for any hints.