10-15-2008 02:12 AM
10-15-2008 10:00 AM
10-16-2008 12:57 AM - edited 10-16-2008 01:02 AM
Sean,
Jigg is right on. Unchecked memory allocation can cause degraded system perfromance in any environment. With TestStand, each time a step executes, by default, the step's Result container is copied to the ResultList for reporting later on. This list will, again, by default, persist until the execution completes. In the case of large interation loops, memory can be consumed quickly, which will eventually cause a slow down in your system's overall performance, since your operating system's resources are being monopolized.
The following settings are a few of the important settings pertinant to TestStand memory managment:
Check out the Developer Zone Tutorial: Best Practices for Improving NI TestStand System Performance for a more detailed and inclusive look at decisions that impact TestStand performance.
Evan Prothro
RF Systems Engineer | NI
07-24-2009 09:57 PM
I am going to use TestStand to implement test sequence also, and I am also have the same concern, because the test sequence could have more than 2147483648 steps. Is there any recommandation on how big of the computer memory should be?
Thanks
Peggy
07-25-2009 12:20 PM
07-25-2009 06:49 PM
This number is from previous somewhat similar program. This number includes measurements, stim cmd, action, and control steps. The initial thought when I ask this question is in the case if everything is going to be recorded. The actual measurement maybe in hundreds (say 100-300~). I think, the current computer configuration has 512MB of RAM on Windows XP, but there is also a debate on whether or not it should go up to 1GB or more. From the description of how the recording is, it sounds pretty memory demanding. So, it makes me questioning on whether or our current computer configuratoin has sufficient RAM.
07-27-2009 09:41 AM
The recording is as demanding as you make it. As mentioned, you can disable the recording of results and I suspect you will have to do that for the vast majority of the 2G steps. You could also modify what data is saved.
In any case, 512MB is pretty low and I would recomend at least the 1G. Memory's so cheap that 2M would also be reasonable.
07-27-2009 05:16 PM
Do you where I might be able to find a justification?!
Thanks.
Peggy
07-27-2009 10:00 PM
See the NI recomendations. For my applications, each sequence has roughly 1000 measurements. We run the parallel model with 2 UUT's so we just doubled the recomended RAM. The cost of the extra RAM was buried in the noise compared to the cost of the pc itself and measurement hardware.
07-27-2009 10:34 PM