Randomly Slow Copy & Paste?

[RoryOF]

How big is the Spreadsheet? How complex are the calculations?

[solid1]

This one is 36kb, calculation was as I stated simple multiplication from two cells.

[jrkrideau]

This one is 36kb, calculation was as I stated simple multiplication from two cells.

Unless you have Data > Calculate > AutoRecalulate unchecked, I believe that AOO will recalculate Everything every time you make a change or add a new operation. So while you are just adding 5 + 7 you are also recalculating every formula in the entire spreadsheet. Hence, Rorys' question.

If you are doing operations on an entire 50K vector or same rather complex calculations, the overhead builds up very quickly Try unchecking AutoRecalulate and only do a manual Recalculate (F9) periodically as seems appropriate. I think it will help.

Regarding plotting ~50K cells, I am not sure that I am reading this correctly. Is that 50K per column or 25K in one column and 25k in the other or some other arrangement? I, originally, read it as 50K & 50K giving a total of 100K cells and then realized I, probably, am misinterpreting it.

In any case I generated samples of 100K and 50K normally distributed random numbers and tried plotting. I am running a very basic laptop without much RAM and when I tried to generate a plot I had problems.

With N=100K the program drew a preliminary graph and froze. After half-an-hour I forced a quit. With the N =50K, again I drew a preliminary graph and the program froze. After some unknown time, perhaps 10 minutes or so, the program resumed operations and seemed fine.

Presumably with enough time the N = 100K plot would work and both would work faster with a faster CPU and more RAM.. I have not used a spreadsheet for plotting for many years and never with that many data points so I do know what one should expect. Hopefully some one with more experience can comment.

I will note that I tried plotting a simple x-y graph of the 100K sample in a couple of dedicated graphing programs and results ranged from less than a second in Gnuplot to perhaps 3 seconds in [R].

If you do not need the plots as an integral part of the spreadsheet but rather for further analysis or reports, one of these or perhaps something like Python with Plotly would be a good alternative. It is trivial enough to insert the resultant plot into the spreadsheet, if needed.

In my experience both Gnuplot and [R] produce much higher quality graphs than any spreadsheet and with much greater ease of use. It appears that the same can be said for Python.