Building the best LabView 8.2 computer

[mjmattin]

Hello,

Please forgive me if I post this in the wrong section.

My company is trying update our lab computer but we are having a hard time finding the best replacement.

Background Info:

We run Labview 8.2 on Windows XP with 2 GB of RAM and a Pentium 4 Processor (3.4 GHz).  In our lab we do high power testing and thus we use LabView to control our 2.5 MegaWatt DC drive while also using LabView to sample data from a 60 channel data aquisition system.  We believe we are having issues with the PC not being able to keep up/bottlenecking. 

All of our LabView programs/code were created by consulting business, so we have no way to modify our code and we have very little knowledge of how the code is structured.  Even though some tests only require a few channels of data, we have no way to turn off the other 50 plus channels from being sampled.

What I would like is information for creating a powerful PC to run our LabView on.  We use multiple Labview programs at once and our lab PC just can't handle it.

I have seen the minimum and recommended system requirements, and I believe we meet the recommened requirements, however, I thnk our use of LabView in our test lab is forcing the program to do too much at once.

If you could build a custom LabView PC, what kind of processor, RAM, video card(s), etc, would give us a powerful PC with tons of headroom?  Would it make more sense for us to use multiple computers, one for control of the DC drive and one for data aquisition?

Our LabView programs/code also output graphs of the real-time data like voltage and current from the drive and we use four monitors, two monitors for who is running LabView, and two more monitors about 10 feet away for other Engineers in the lab for the testing.  Would a video card helps us?

I would be greatly appreciative of any feedback.

Thanks for taking time to read this,

Mike

[natasftw]

LabVIEW 8.2 is only compatible with the 32-bit version of XP.  You've already limited your "super" machine quite a bit with that decision.  If you're looking to increase power, you'd serve yourself better by taking the time to understand your program more than trying to max out your system (you're pretty close already).

Do you have the source code or just an executable?

[mjmattin]

Thanks for the reply! All of the code is locked.  It sounds like our best option would be to upgrade to a newer operating system?

[RavensFan]

Probably not.

If you have LV 8.2.1, you could updgrade your OS to Vista.  But that isn't supported by MS anymore.

http://digital.ni.com/public.nsf/allkb/B972242574D4BB99862575A7007520CB

[Hooovahh]

So while 8.2 isn't officially supported in Windows 7, I'd bet that it would be able to run.  There are some running LabVIEW 7.x in Windows 7.  But a much bigger struggle you are going to have is going to come with getting a new fresh machine, and having everything installed and configured right.  Did this company create an installer that installs all of the components?  What version of DAQmx is it running?  NI-VISA? CAN? XNet? FPGA? MAX? What hardware is it using, are the scales and channels configured in MAX?  Where is configuration stored?  There's lots of stuff involved in taking a new blank machine and setting it up to run.  I'd try to just get the system running on any other computer first, the task is going to be more difficult than you may think.

You say all the code is locked, does that mean it is running from source?  Is the company you worked with still around?  Can their services be used again?  This time I'd make sure you get what you are paying for, and get the source code to everything developed for you.

[billko]

---

@Hooovahh wrote:

So while 8.2 isn't officially supported in Windows 7, I'd bet that it would be able to run.  There are some running LabVIEW 7.x in Windows 7.  But a much bigger struggle you are going to have is going to come with getting a new fresh machine, and having everything installed and configured right.  Did this company create an installer that installs all of the components?  What version of DAQmx is it running?  NI-VISA? CAN? XNet? FPGA? MAX? What hardware is it using, are the scales and channels configured in MAX?  Where is configuration stored?  There's lots of stuff involved in taking a new blank machine and setting it up to run.  I'd try to just get the system running on any other computer first, the task is going to be more difficult than you may think.

You say all the code is locked, does that mean it is running from source?  Is the company you worked with still around?  Can their services be used again?  This time I'd make sure you get what you are paying for, and get the source code to everything developed for you.

---

Sometimes it's better to stand pat.  I think this is one of those times.  Go with what you have already.  No use trying to upgrade - the ROI just isn't there.

[Hooovahh]

That's probably true.  I'd invest in as much RAM as the computer can hold (sounds like 4GB if the OS is 32 bit), and an SSD (be sure and image from the current one and deploy to the SSD) then your machine should be identical to it is today, just with some added performance.  If the CPU is the bottle neck, then there isn't much that can be helped, but I've seen plenty of old crappy computers have new life with an SSD.  I've yet to try to upgrade an old tester and see if there is improvements but it's worth a shot and only costs $100-$200 and a couple hours of your time.  Oh and I'd uninstall, or disable any running programs or services that aren't needed, especially anti-virus and firewalls if the PC can be disconnected from the internet.

[mjmattin]

Thank you all for the responses.

The consulting company who provided us with the program/code is still around so we are going to cantact them and get all the information we can about whats going on "behind the scenes" of our PC.  We would like to upgrade the computer and OS but it looks like that is going to be alot trickier than expected.  In the meantime, we are going to get our IT to bump up the RAM to 4GB and possibly swap out the HD for an SSD.  I think the real problem is that we are having the PC control and sample too much at once.  Thank you again to everyone who has responded.  I am going to look into classes so we actually have some in house knowledge as well.

[Hooovahh]

Great idea, here are the standard training links you'll find on the forums.

NI Learning Center

NI Getting Started

-Hardware Basics

-MyRIO Project Essentials Guide (lots of good simple circuits with links to youtube demonstrations)

-LabVEW Basics

-DAQ Application Tutorials

-cRIO Developer's Guide

Learn NI Training Resource Videos

6 Hour LabVIEW Introduction
Self Paced training for students
Self Paced training beginner to advanced, SSP Required

But honestly it sounds like you may want to take some formalized training from NI or one of its partners, good luck.

[Bob_Schor]

Indeed, if the people who wrote the code don't know how to take advantage of LabVIEW's inherent parallelism, and they are doing a combination of data acquisition, processing, and control, presumably all in a single loop, there are all kinds of problems waiting to happen.

I "inherited" such a routine that was "doing too much" in a Real-Time loop.  It ran at 500 Hz, and many runs of 40-50 seconds showed "RTERR" flags, i.e. samples were missed.  I recoded it using multiple loops, with the time-critical loop doing nothing except acquiring data and shipping it out on an RT FIFO, and the routine now runs at 1KHz and never misses a sample (and we do a lot more with the data).

Bob Schor