Reducing deployment time on cRIO

[D. Ackermann]

Hi

I am currently involved in a project with a test sequencer based on actor framework executed on cRIO platform. One major problem that arose, is very long deploy times for the code to the cRIO. Meaning the start LabVIEW code in debug mode. It takes up to 8 minutes to get the code started.

I want to create a strategy on how to optmize the project to reduce the deploy times. And would be glad to receive some work experiances from the field.

We use LV2020. LVOOP is heavily used.
We do not use precompiled libraries yet. We heard from another company, that they had problems using it with actor framework, but cannot say what exactly was the problem. Are there any?

Thank you in advance.

Dermot

[Jordan Kuehn]

That deployment time does seem excessive, at least for deploying the application. Running interactively is a mixed bag in my experience. I have a few cRIO AF applications of various sizes at the moment. One crashes LabVIEW when I attempt to launch it interactively, another similar one works just fine. I use the debug tools to acquire queues (in a conditional disable structure) for the main Actor Core that has a small debug UI included with it.

When deploying are you able to see what libraries or code seems to take so long while watching the items scroll past? Is it code you have written or 3rd party libraries?

[D. Ackermann]

I have to say, I did not work on it myself yet. I will receive a cRIO tomorrow to start to work on this problem.

I just wanted to start to collect information beforehand on what can be done to reduce the time. I will post more details then.

Thanks!

 

[D. Ackermann]

Ok. Now I have my hands on it. I have a cRIO-9047, which should be quite fast.

To get my test to work fast, I had to strip down the current project a lot. I kicked out EtherCAT master, Softmotion, PID control and logging, which was all programmed in actors.
With this reduced code on my PC it takes about 3 minutes to have the program running after I hit start in LabVIEW. I then put most libs of this project in a wrapper lib and turned this into a packed project lib. This reduced the deploy time to about 1,5 minutes. I noticed then, that I see a lot of toolkits VIs in the deploy process window. JKI this, OGTK that.. So probably using a lot of Toolkits on RT is not really optimal..

I think it will require a redesign with project packed libraries, while also checking which toolkits are really necessary, to be able to work efficiently in this project