02-10-2011 09:22 AM
I remember a demo from one of the LabVIEW classes which benchmarked the speed of writing directly to an indicator compared to a local variable and property node.
I thought that writing to a local variable was very slow compared to writing to a terminal. With 2010 there is no difference that I can see. Maybe I am not remembering correctly but I know the NI gurus did a whole bunch of performance optimizations.
I no longer have 2009. Can someone with 2009 or earlier tell me what the speed difference is between writing to a local vs terminal in the attached vi?
Solved! Go to Solution.
02-10-2011 09:37 AM
NI implemented a back-dorr to allow the local to update fast. Not as fast as the terminal but fast.
The draw-back cmoes into effect when you have multiples.
If you have a large array you are updating (large array will make it easier to benchmark) and there are 20 places that read via the local, then the dat has to be copied to all of those other locals.
Ben
02-10-2011 09:51 AM
Ah. So this back door was put in for 2010?
I was trying to prove to someone that writing to local variables is much slower than writing to the terminal. So I wrote the benchmark vi and was surprised by what I saw.
02-10-2011 09:54 AM
No, I suspect the back-door was implemented as long ago as LV 6 or prior.
I think one of my tags point to a discusion where I read that. Don't have time to chase it down at the moment.
Ben
02-10-2011 10:09 AM
Sit down with a roll of duct tape and start wrapping your head in and by all means DO NOT read post #17 until you are done wrapping your head.
But don't stop at post #17 keep reading, it gets better.
Ben
02-10-2011 10:22 AM
02-10-2011 03:43 PM
Thanks. I ignored your advice about the duct tape and read through #17 and the whole thread. I will now go find some duct tape to see if that helps stop my head from spinning.