PXI RT RAM LabView 9

[Bill_Cr]

We are using 8145RT boards with 64M RAM in most machines and have just upgraded to LV9, with a bunch of difficulty.  I am finally past the major version changeover problems and back to development.  Does anyone know if LV9 is alot heavier on RAM overhead compared to LV8.6?  RT code that used to run before, and only slightly modified in LV9, now does not fully deploy and shows "Deployment completed with errors", but does not actually identify the errors with bold text like it normally does.  However, the bottom line says there is less than 200K of free memory out of 57M, so my guess is that the very last thing in the deployment didn't fully deploy because of a shortage of RAM.  When looking at MAX, it says only 5.4M is free even before I try to deploy, so it looks like LV9 overhead is taking up over 50M by itself.  I am about to start addressing this with memory management ideas from NI tutorials, but I wonder if a better way is to replace the RAM with 128M modules.  But, NI is showing like a 2-3week lead time on those modules (777885-128) and I don't really think I can wait that long.  These are almost impossible to find because they are a low profile form factor.  I have no idea what laptop(s) they are used in, but they are definitely not mainstream computers.  They appear to be a 144pin SODIMM at 133 speed (based on the Samsung chip designation on the 64M version we currently use), but they are only 1" tall and most laptop memory is 1.25" tall and won't fit on the board because of mechanical design issues with the 8145RT board.  If I am right about the speed, Swissbit makes a card that I have the model number for and possibly one or two others like Samsung or Crucial make one, but I can't find a source that actually has them in stock for sale.  The closest I found was a 66 speed with the right form factor.  Does anybody know if the 66 speed will work in a 133 slot (our PID loop closure rate is only 1.25kHz, so I guess we could probably use the 66 if it will work on the board without causing other problems in the RT system), or does anybody know of a supplier that stocks low profile form factor 144 pin SODIMMs that might have them?  My guess is that NI has to go back to Samsung or somebody to actually make them since they are not a mainstream product and that is why the big delay and high price tag.

[Smootastic]

Hi Bill,

Sounds like there are a couple of issues to address.  The first is the Deployment with Errors message.  Are you deploying shared variables?  What types of things in the project are you trying to deploy?  Can you run the VI directly from the project?

As far as memory is concerned, I'm not aware of any issues with using third party RAM.  I will advise against it, because it's probably untested and could cause issues with your controller. 

[Bill_Cr]

Thanks for the info.  Yes, we are deploying shared variables, but if you just deploy the Host SV.lvlib from the project by right clicking and deploying that one thing, you don't get an error, so I don't think it has anything to do with SVs.  You don't get the error unless you actually try to run the main RT code VI from the project (which deploys everything at the start).  It never seems to fail at an SV, it always happens on an NI vi/dll in the dependencies, so it's not in our code.  When you run something small like the NI RT benchmark example program with no other code in it, the benchmark code deploys and runs without error.  During the deployment of our RT code, the available memory gets very small before it errors out.  Because of all this, I am pretty sure the problem is memory and am looking at both options of (1) addressing RT memory management in the source per Developer Zone tutorials 4537 & 4430, and (2) doubling the physical RAM.