What's your upper limit on run time, and number of elements? The
most memory efficient method would probably be to write the data to a
binary file, pair timstamps with data points (or just a time stamp at
the beginning and extrapolate data point times based on offset, if
you're continuously recording data), and write a .vi to retreive the
data based on the time you enter to review. For 1 data point per
second, this shouldn't be too tough to impliment.
If I
had to impliment this functionality, I'd be using a lossy queue as my
FIFO, with the "overflow element" output of the lossy enqueue vi
written to the binary file. To replace the plotting of the real-time
data array as a waveform, simply wire the new queue element to a strip
chart and set the history length to whatever you want it to be. Though,
if you go with the strip chart method the buffering of the data is
handled by LabVIEW, and you wouldn't really need the queue/array at all
unless you're doing some data analysis.
