05-04-2010 08:54 AM
I'm building a rather large project into an *.exe. This build takes 40 minutes for one of the smaller projects (3500 Vis *.exe size 37MB) and up to 3.5 hours for a larger project (5000+ VIs, *.exe size 52MB) using LabVIEW 8.6.1f
I've noticed that one of my computer processors is only being utilized to 25% (I have a total of 4 processors on the computer and only one is being used). Does anyone know how to get LabVIEW to use more of my available processor power? Any help/tips are appreciated.
05-04-2010 08:58 AM
Load in only one CPU indicates the code running may be single threaded.
Only using 25% indicates the work is not PCU bound but probably I/O bound.
Move to a solid state drive or RAID sub-system to speed-up the disk I/O.
Is all of your memory used up? If so more memory may reduce disk I/O requirements.
Ben
05-04-2010 09:11 AM
Hi, Dave,
Strange question from person who working with 5000+ VIs project...
Anyway, there are two threads which may be helpful for you:
Dual Processor Multi Core Parrell Processing Question
And here we was able to get 8 cores processing:
Interpolate 1d slow on 8 core machine
With best regards,
Andrey.
05-04-2010 09:22 AM
Hi, Dave, its me again,
Sorry, may be I not understand your initial question...
If you need to improve speed of building prcess, then you have nearby no chances other than change architecture a little bit and split your application into small modules (LLBs "plugins") and build these modules separately. So, every time when something changed its will be not nesessary to build whole application, but the only changed part.
Andrey.
05-04-2010 10:06 AM
This post/response explains why the application builder doesn't use more cores.
However, there may be something that is causing it to only use 25% of the one.
05-04-2010 10:10 AM
Exactly... the problem is not with the executable itself once it is running on the test station. The problem is getting the LabVIEW build process to use more of my available processing power on my computer so that the build can go faster. I can't find any settings anywhere in the development environment in order to optimize processor utilization during the building of the *.exe. Thanks for your help so far... more info regarding the LabVIEW building process optimization is what I'm looking for.
/Dave B
05-05-2010 02:39 AM
Hi Dave
If you read this thread you'll find that there is a work in progress to get the compiler to work faster, but that most of the build process needs to be serial and run in the UI thread. This means that normally only one thread can be used. In light of this the behavior you are seeing is actually expected, but like I said there is work in progress to speed things up.
Best Regards
David