NI VISA and Parallel connections with Serial Devices

[Kenny_K]

I am trying to figure out what is causing our code to not be able to run truly in parallel when using several different Serial Port adapters, and trying to communicate with multiple devices at the same time.  I have tried a MOXA Ethernet to 4 port RS323, a MOXA USB to 4 port RS232 and a StarTech USB to 8 port RS232, and they all show the same behavior.

In the "real code", I have cloneable DQMH modules that communicate with a test nest, containing the device under test and serial ports to communicate to each device individually.

I made a simple VI, that will communicate with a device on each serial port, just to try to get it down to its "simplest" form.

When looking at the execution times of the commands, I would expect them to be similar in times (under 2 seconds for each serial port), but I am seeing that is appears that only 2 of the serial ports can communicate at the same time.

Notice how two of the devices are under 2 sec, and the other 2 take considerably longer.

All of the VIs are set to Shared Clone.   Some of the VIs internally are Pre-allocated subroutines, but otherwise are Shared Clones.

- I have tried running the Read/Writes as both Asynchronous and Synchronous, I didnt notice any difference.

- I have run the VI Analyzer to check for re-entrant issues, and found none.

- I tried adding in the In Place element structure where there was an unblundle/bundle (per VI analyzer recommendation)

- I have the high resolution relative seconds sprinkled around to gather some timing stats.  It is a Pre-allocated subroutine, but I wrap it in a Flat sequence structure to control data flow.

- Replaced the VISA Configure Serial Port with the actual code on the block diagram, since that VI is not re-entrant

I am currently using LabVIEW 2024 Q3, Patch 2 (64 bit), but I have seen the same behavior with other versions of LabVIEW as well.

NI VISA is also 2024 Q3.

Sorry, I cannot post the code due to company IP.

[Spruce]

Did you try setting your VIs to preallocated clone? That always seems a better option to me. RAM is cheap!

[Kenny_K]

Unfortunately I cant, due to using Dynamic Dispatch in the class.

[RTSLVU]

I have tried similar things in the past with VISA Serial and multiple serial ports. I believe even with multiple device on multiple ports (even different ports like Serial, GPIB, VX11) VISA does not access them in parallel regardless of how your program looks, VISA accesses them one at a time in random order like a race condition. 

[Kenny_K]

Ok, an update.  When using the simple VI, if I move each serial port into its own FOR loop, I can achieve the expected parallel performance

Times are much faster, and way more consistent.

I also did create a quick and dirty Cloneable DQMH module, and the times are still matching the parallel FOR loop in the code above.

So that means that the issue is in my actual test code (which is a lot more complicated than the simple module shown here.

I guess ill keep trying figure out what is causing the loops to not do the communication in parallel.