PCI-6703,MacOS 9.1,NI-DAQ 6.6.1,Labview 5.0

[jdwright]

I am using a PCI-6703 w/ MacOS 9.1, NI-DAQ 6.6.1, and Labview 5.0.
The computer screen freezes when I run a VI for extended periods of time.
How do I solve this problem?

[George M.]

Hi,

Are you able to reproduce this behavior on another computer? Can you regain control of the computer, or do you have to restart it? Is there an error code that is displayed? Has the code worked before? If so, what has changed? Any other information you can add to better define your situation would be appreciated. Thanks.

George

[jdwright]

I have reproduced this behavior in two different situations.

situation 1:
Beige G3 PowerMac computer, MacOS 9.2.2, Labview 5.0, NI-DAQ 6.6.1f1, PCI-1200 installed, PCI-6703 Main VI calls ~5 subVIs. The third VI (VI3) does the data acquisition. After a number of hours (~10-15), the computer screen freezes. Normally I would run the VI in debug mode to see exactly where it hangs. However, due to the long time required to see the problem, I cannot really take this approach. I can see that VI3 is running when the freeze ocurrs. Note, none of the subVIs actually use the PCI-6703 yet. The problem arises simply because the PCI-6703 is physically installed in a PCI slot of the computer (unlikely) or the problem arises because I changed from NI-DAQ 6.0.x to NI-DAQ 6.6.1f1 (my current working hypothesis). When I change from NI-DAQ 6.6.1f1 back to NI-DAQ 6.0.x the problem disappears. NI-DAQ 6.0.x isn't good enough though because it doesn't recognize the PCI-6703. I guess NI-DAQ 6.1.x already added support for the PCI-6703 so I could try that instead of 6.6.x. Because it takes a number of hours for the problem to appear I am
guessing that it has to the memory filling up. VI3 is continuously reading data from 6 analog channels and displaying the data on 6 waveform charts. The data from each channel is stored in a 2-d array. Each of those arrays is bundled by name into a bundle. The bundle also contains string and numeric values that encode the scan parameters. Although this named bundle is handy for passing the data from one subVI to another, I think it is cumbersome for memory management. Maybe NI-DAQ 6.6.x handles the memory differently from 6.0.x and that's why the problem is showing up. Crashes due to lack of memory have been a problem in the past for this Labview program.

situation 2:
Blue and White G3 PowerMac computer, MacOS 9.2.2, Labview 6.x, NI-DAQ 6.6.x, PCI-1200 installed. The main VI continuously reads data from 2 analog channels and displays the data on two waveform charts. The screen freezes after
a number of hours (~5). When I change from NI-DAQ 6.6.x to 6.0.x this problem goes away.

There should be people out there using NI-DAQ 6.6.x with the MacOS. Last I checked, 32% of users are using traditional NI-DAQ. On the other hand, only 1 in 10 (probably less) use traditional NI-DAQ with a mac. So that's 3.2% at best.
I might be trying to do something that nobody else has done.

[johnsold]

I have run LV 5 and LV 6 under OS 9 and NI-DAQ 6.6.x and do not recall any problems like yours. However, most of my programs did not run for hours. Typically when something crashes after hours, I tend to look for memory leaks or data overflows. Try running the profiler (I think it was available in LV5). Take a snapshot every 15 minutes or so and see if the amount of memory or time starts to increase.

I seem to recall that there were soem issues with NI-DAQ 6.6.x not being fully compliant with some DAQ hardware (in spite of documentation to the contrary). A PCI bridge conflict was one thing, although I could not find anything definitive about how to identify when it existed. We had problems the 6602 counters.

I do not have any comparable DAQ hardware or I would offer to try it.

Lynn

[jdwright]

Thank you for your reply. I will definitely try to use the "profiler". How do I open the profiler? I will look for it in the menu items but I don't really know where to look. The profiler sounds like it would be extremely helpful for diagnosing the problem. Up until now, I have been using the "About this macintosh..." memory usage bars in the finder, but that does not seem to show any memory leak.

I have simplified examination of the problem. I can reproduce the problem by simply continously running the "AI sample channels" VI that is a part of the main Labview library for data acquisition. Thus, it's not my bad programming that causes the problem. It's really a problem with NI-DAQ and the DAQ library. I continously scan 6 channels and the computer freezes after about an hour. I am currently whether the problem appears if I simply continously scan 1 channel. Anyone with a PCI-1200, NI-DAQ 6.6.x, any version of MacOS, any version of Labview, and any macintosh computer should be able to reproduce the problem. If not I'd really like to know. The problem might even show up for a different board besides the PCI-1200. But maybe the legacy nature of the PCI-1200 allowed it to slip through the testing cracks.

[johnsold]

In LV5 the Profiler is located at Project>>Show Profile Window. Check the boxes for Profile Memeory Use and Memory Usage and Timing Statistics. Press the Start button before starting your VI. Then from time to time press the Snapshot button and watch for trends. Obviously the total time and # runs will climb steadily but watch for changes in average time or memory usage that may be clues to your problem. If you are unfamiliar with the Profiler you may want to try a few runs with other programs to see how it works before commmitting to a long run with your "problem child."

I have LV 6 and a Lab-NB board in a 7100/66 with OS 7.5.5 in the lab. I may try the AI Sample Channels on that to see what happens.

Lynn

[jdwright]

OK, I found it. This utility will really help. Thank you.

[carolhuddy]

Did you ever find out what the problem was?  I am finding the same behavior with a PXI/ SCXI system after running it about 14 - 15 hours.