LabWindows/CVI

cancel
Showing results for 
Search instead for 
Did you mean: 

CNV library memory leak?

Sure, you could upload a copy of it, and I could check it out.

 

When I said the memory allocation, I meant to show the memory allocation through the resource tracker. This way, it shows where the exact memory allocation is coming from and how much it is producing.

0 Kudos
Message 11 of 18
(1,586 Views)

Gotcha.  I'm re-running my test now with resource tracking enabled.  So far, nothing revealing.  From before my CNV calling to N minutes after, I'm not seeing anything new in there.

 

Beginning...

 

2016-07-25 07_04_16-Resource Tracking.png

 

After 1 hour...

 

2016-07-25 08_05_43-Resource Tracking.png

 

 

What is NI's FTP address these days?  ftp://ftp.ni.com/ isn't working for me.

0 Kudos
Message 12 of 18
(1,570 Views)

You should be able to upload to ftp://ftp.ni.com/incoming

0 Kudos
Message 13 of 18
(1,557 Views)

Sorry, today it's working.  😉

 

It's up there.  Look for my username.  It's set to run in debug mode, no hardware needed.

0 Kudos
Message 14 of 18
(1,550 Views)

Hello,

 

I just want to give you a quick update. When checking the known issues with CVI 2015, I noticed that there is currently a bug that has existed since CVI 8.5 that creates a small memory leak when creating network variables. This may explain the memory growing when running for hours. 

 

The following article explains this, and provides some workarounds. It should show up under issue 41414.

 

LabWindows™/CVI™ 2015 Known Issues

http://www.ni.com/product-documentation/52829/en/

 

 

0 Kudos
Message 15 of 18
(1,533 Views)

Thanks for the reply.  I'm not entirely clear how to go about implicitly creating those variables.  But regardless, in my app, I'm only creating and deleting the variables once.  They are resident for the entire duration of the app.

0 Kudos
Message 16 of 18
(1,521 Views)

You are right about that. Hm, I am going to have some programs run overnight to track the memory leakage. I will check to see if the issue is inherent to the CNV functions, or if something else might be happening in the project.

Message 17 of 18
(1,504 Views)

Wow, time flies!  I got busy doing other things and then returned to this issue just recently.  I had totally forgotten that I was looking into my memory leak four years (!) ago.

 

ShamWOW, you still around buddy?  Did you ever get a chance to test CNV related leaks?

An update on my end.  In a desperate attempt to cut out culprits, I added a local disk method of obtaining these temperature measurements. I now write the double value to an ASCII file and let my reader applications open that file in READ_ONLY mode, get the value and close the file.  This is way lo-fi, but works well and fast.  Therefore, no CNV to contend with.

 

And yet I saw my memory leak still happening over 70+ hours.  That leaves only a couple other libraries for me to focus on:

 

1) TCP (this one is vital, it's the connection method to my instrument, Watlow controller)

2) TDMS (this one isn't strictly needed, as I am writing temperature data to a TDMS file).

0 Kudos
Message 18 of 18
(536 Views)