When mapArrayToAccelerator is called, the base address of arrays is
often still zero; we haven't parsed the trace yet and the base address
stored here is the trace address, not the simulation address. That means
we can't check for overlapping memory regions.
When mapArrayToAccelerator is called, the base address of arrays is often still zero; we haven't parsed the trace yet and the base address stored here is the trace address, not the simulation address. That means we can't check for overlapping memory regions.
This is related to (but does not fix) https://github.com/harvard-acc/smaug/issues/102.