Another LV Bug? - A different chart memory leak?

[altenbach]

---

Matthew Kelton wrote:
Actually the waveform only has t0 and dt defined,  It is still assumed the points are equally spaced. 

---

No!, every (~)10ms you define a new t0. There is no guarantee that the points are equally spaced, ever! Your 10ms wait is not deterministic.

For example you can exaggerate the problem by radomizing the loop rate a bit like in the attached image to make the jitter more visible.

Message Edited by altenbach on 11-26-2007 11:48 PM

[chilly_charly]

---

altenbach a écrit: ... No!, every (~)10ms you define a new t0. There is no guarantee that the points are equally spaced, ever! Your 10ms wait is not deterministic.

For example you can exaggerate the problem by radomizing the loop rate a bit like in the attached image to make the jitter more visible.

---

Excellent !!!

[Ben]

Adding what little I know...

When using WF data types to a chart, the history length determines the number of updates and not the number of points.

If your WF has 100 data points and uses the default history size of 1024, then 102,400 points will be displayed before the earliest data is dropped.

Ben

[Matthew_Kelton]

Ben,

I just replied to a similar message from an NI employee in the original thread that spawned this one.

When I think of chart history, I guess I think data points.  So I thought that it would be 500,000 data points, so if I plotted 5,000 100 point waveforms I would fill the history, if I plot 500,000 1 point waveforms, I would fill the history.  But, since it is a history of 500,000 waveforms, I can plot 500,000 5,000 point waveforms or 500,000 1 point waveforms, or some mixture of the two.

I find this decision on NI's part to be puzzling.  The memory usage is completely non-deterministic, and, at least to me, it seems non-intuitive.


Message Edited by Matthew Kelton on 11-28-2007 01:18 PM

[Ben]

The WF data type is great when dealing non-periodic devices like a bunch of GPIB stuff since the t0 of each value can be distinct. I have also used them with DAQ devices (since about LV 6.X) and when the chart fills up, memory used to settle down (at least the last time i looked).

Ben