Has The Performance Of Multiple AI Inputs Anything To Do With Computer Speed/RAM Size? (Using Windows XP)

[Shun]

Hello,

I am using LabView-DAGmx and PXI-6220 to acquire voltage (differential) data. The application needs 8 channels which PXI-6220 should be able to support easily, but also must have 10  1-D Floating Point Arrays (8 elements in each) in order to keep on calculating the average, verifying the readings, allowing changing the set points, ... etc. I also use 8 elapsed counter VIs to display the times in the application. While I really didn't think the FP Arrays and the calculation are a lot to my workstation, I found from time to time the elapsed counters are not constantly updating the times (strings) and the CPU Usage has hit 80%. I then modified my VI to do only 4 channels instead of 8 channels, and the CPU Usage would remain below 40% (I think it's still very high.) At the point, the timers were still not able to update the time display fast enough, although I found every time a correct time will be displayed when it updates.

My question is, how much memory should be appropriate for this kind of application (LabView & My Multiple Channel Averaging)? I have now Pentium 4 3GHz and 1GB RAM. Is upgrading RAM to 2G or 4G going to help a lot in the situation? If we intend to use TestStand, what size of RAM would an expert recommand?

Thanks!

[TCPlomp]

Without any code or details (samplerate) it is hard to say anything.

What is normally the load on the CPU. What other processes are executing (we had a virus scanner that would hit 100% every 5 minutes).

Ton

[Shun]

The sampling rate is 40/s. I am using DAGmx Read (Analog 1D Wfm N-Channal 1-Sample) VI in the application.

[TCPlomp]

Why single sample? How is the timing done? Reading data in bursts at approx. 10 Hz works most of the time perfect.

Ton

[Shun]

Single Sample so that each reading can be added to the total for averaging. This is different from "Moving Average" which uses a fixed buffer to calculate the mean from of the number of readings.

The timing is done through the general while loop.

There is also a particular but required (for our application) reason that it has to be 40Hz for each channel.

Please see the attachment. Thanks a lot!

[TCPlomp]

I see several things in your code that can be improved: 

•  The calculations in the for loop should be done with auto indexing of the for-loop
• The memory you use inside the for loop could be stored with shift register, no need for the property nodes which influences your speed in a very bad way
• The calculations you do in the bottom part are all referenced via VI-server, this can be influencing your speed as well

Ton

PS you can attach pictures directly.

To attach pictures of your code I advice the Code Capture Tool. It is great for this purpose, you select the part you want to capture, start the tool (Tools->Code Capture Tool), review your setting (FP-none, BD-selected, Clipboard-FileName), finish the capture.
Then you switch over to your internet browser, in the file dialog you past the file name from the clipboard and the file is attached.
To download

[Shun]

Thanks, Ton! I shall use your expertises to improve the existing codes then come back for a further discussion.