How to Optimize SCXI 1600 for speed with Thermocouples

Question

I'm working on a data acquisition system for my engineering firm and I'm trying to find a way to use our new thermocouple system as fast as possible.
The requirements for the DAQ process are:
Read 32 voltage channels from a PCI-6071E card
Read 32 thermocouple channels from a SCXI-1600 with an 1102C accessory
Complete the entire operation in under 5ms (this is so other parts of the program can respond to the incoming data quickly and trigger safety protocols if necessary)
Using LabVIEW 7.1 and MAX 4.4, I've got the voltage channels working to my satisfaction (with traditional DAQ VIs) and the rep rates I measure when I run the program are around 1ms (I do this by putting the DAQ code in a loop and reading the millisecond timer every time through that loop, then calculating the average time between loop executions).  I have been trying to get similar performance from the thermocouple channels using DAQ Assistant and DAQmx.  Some of the problems I've encountered are:
Very slow rep rates with 1-sample and N-sample acquisition modes (300-500ms)
Good rep rates when I switch to continuous mode, but then I get buffer overflow error -200279.
When I attempted to correct that error by setting the DAQmx buffer to overwrite unread data and only read the most recent sample, the calculated sample rate went to 20ms.  It was around 8ms when I left the error unhandled and continued acquisition.
At this point I'm out of ideas and am just looking for something to try and optimize the DAQ process for speed, as much as is possible.
Thank you for any help.

Answer 1

I guess I would be interested in checking out your code to see if there is anything I can recommend on changing.  However, I do have a few general Ideas of how to improve your performance.  These recommendations are purely based on what you could be doing to slow down the speed of the program because I am not sure how exactly you have everything set up.  
-Are you setting up the task and closing the task each time that you read from your daq card?  the way to get around this is to only have the DAQmx read vi in the while loop so you do not have time alloted for opening and closing the task each time.
-Try using a Producer/Consumer architecture.  This architecture uses queues and splits the aquisition with the post processing.  Here is a link to how to set up this architecture and some information on when to use it.
Application Design Patterns: Producer/Consumer
http://zone.ni.com/devzone/cda/tut/p/id/3023 
Message Edited by Jordan F on 02-06-2009 04:35 PM
Regards,
Jordan F
National Instruments