HI enigma,
I think the first place to start is that the graph is just a way to represent an array of data. The array "lives" in the block diagram, and you represent it on a graph. So when you say you want to calibrate a graph, you need to start with the array(s) of data, and then "massage" the data in the array, and then push the array into the graph.
I am a little confused on your exact application, but it sound like you have two acquisition channels. One before the sensor, and one after the sensor. The data for these two channels will come out of a NI-DAQ Read VI as column arrays (2 col, and N rows, where N = number of scans to read at a time). You can wire the 2D array directly into a waveform graph (right click the graph and select Transpose Array) for viewing. If you want to "calibrate" the data, you could subtract ch1 from ch 0 (or vise versa) to get kind of a differential value, and then wire the resulting 1D array into the graph. Try the array tools for indexing indiv columns for your math operations.
For the "memory" need, that will probably be harder. If the acquisition is triggered, then you can expect point N to have a certain value (as long as the test is the same for each test run). but if the acquisition is not triggered, then a certain value may occur at x index n on one test run, and at index m on another. You would need to do use some type of sliding window to recenter the data with each new acquisition...
As for the real time, the data streaming will be function of the DAQ Read returning. As long as your data comes in a couple of times a sec, you won't be able to notice any delay while the buffer fills up and the DAQ read is waiting to retrieve the "number of scans to read" parameter that you gave it.
Let me know if you have any questions.
Joel
