Andrey Dmitriev wrote:
It means, that sometimes (but not always) AutoIndex is more preferred than ShiftRegister.
As of LV 6i, the game changed radically. Prior to LV 6i, it was faster to work in the SR. There were a bunch of enhancement introduced in LV 6i that improved performance.
Also keep in mind the "allocate" array takes some time and may (not sure) force a call/check to the Memory manager.
It IS possible to predict when and where LV will duplicate or copy data, but its not easy to pass that info in a short rule other than
"LabVIEW's in-placement algorithm will attempt to schedule code execution to minimize the number of data copies."
Applying that end goal to the countless number of code constructs possible... quoting Ross Perot
"The devil is in the details."