How does the "Read TDMS" function work when reading part of a data file?

[jcannon]

Hi everyone, I have stored some rather large .lvm files that causing me a a big headache when trying to read/analyse the data.... As a work around I first convereted these files to .tdms.... Now the files are much smaller (50%) and can be load/plot within a couple of seconds rather than 10's of seconds. Specifically, the files include 7 channels of data that were sampled at 2kHz for about 8-9 mins..... So, each file contains about 700,000 data points or elements.... I am using a producer/consumer architechure to read/plot all the data, then using cursors to identify a data subset, which is send via a queue to the consumer where analysis occurs.... Given the size of the files, when subset is quite large I get "Out of Memory" errors and LabVIEW crashes. What I am thinking is that I can down-sample or decimate the data in the producer back to 1kHz (I need to ms resolution so I can see the spikes and can appropriately selected the subset) then index back into the original tdms file to send the undeciated data to the consumer. With the "Read TDMS" function, the help states: COUNT: Specifies the maximum number of data elements to read from the .tdms file for each channel. The default is –1. OFFSET: Specifies the number of data elements into the .tdms file at which the function begins analyzing the .tdms file. The default is 0. I am confused by the term "elements". Does this mean actual "data points" as in elements within the 2D array? Or bytes? I am hoping array elements... Would this then mean that given my 7 channels of data, each element value that is a multiple of 7 would index into a new line of data? I also assume this index is zero based... So with no offset, a count value of 0 would return all the data elements at time 0.0000s for all channels (regardless differences in the precision of each channel, unlike bytes), a value of 7 would return all the data elements at time 0.0005s, a value of 14 would return all the data elements at time 0.0010s, etc etc.. Also, if anyone has any example code for indexing into a file using this method they could shareI would be very appreciative... I would assume a for loop could also be used to read the data subset into the consumer in smaller chunks to reduce the memory load? Many thanks, Jack

[crossrulz]

You should have a different channel in the TDMS file.  So you are reading X elements from a channel starting at element Y.  Yes, those are elements in the array, not Bytes.  NI did a good job of making the API easy to use.