Distributed System Manager Historical Trend

[viScience]

I have found that the DSM historical trend display will become fragmented (looks like data points are not connected together) after rewriting about 1/2 the screen.  If I zoom in and then out, the display will refresh correctly and continue properly until a while and then back again to the disjointed display.  This does not seems to occur if I am looking at a smaller time span.

[DiscoBall]

Hi Sachsm,
 
Thanks for bringing this up. Seeing the disjointed/fragmented display is likely caused by graphics settings on your computer. Getting Windows to redraw the image (by zooming in and out) does indeed fix the issue. Do you have your graphics card set on full acceleration? Do you believe you are losing data and are experiencing more than a drawing issue?

[viScience]

Joshua,

Thanks for your response.

I turned off all GPU HW accel but the problem still persists.

[viScience]

BTW,  I do not seem to be loosing any data points.

[DiscoBall]

Hi Sachsm,
 
Having a low hardware acceleration for the graphics is the problem and turning it off, as you have shown, makes things worse. Can you check to see if there is an update driver for your graphics card?

I haven't been able to reproduce this issue with DSM on my machine. Out of curiosity, does this fragmentation only occur in the second historical trend, or does it also occur in the default? Also, please post a screenshot of the properties for the historical trend.

[viScience]

Hi Joshua,

Actually the 2nd screen shot I sent was with the GPU set to full acceleration, it was just another more pronounced example of the problem.  As I said, I did turn off all acceleration and there was no apparent change in the disjointed display.  Also I have attached 2 more shots, one of the Historical Trend, which shows the problem and one of the same channel of data displayed in the Auto View which never exhibits this problem.  It would seem that if it were a graphics problem I would be seeing it happen in both cases.  It seems more like some sort of problem with the way the Historical graph is handling the incomming timestamps of the data.  In my system, all of the data is coming in from a cRIO which has a GPS 1588 timestamp.  The windows system time is unfortunately not GPS sync'd, and is set ~ 20s behind the GPS time.  This turns out to be neccesary since the Axis span of Historical graphs is based on windows systems time not the timestamp of the data set.  (I still do not understand why is works that way)  If the windows system time is ahead of the data set then it does not show up on the display at all.