Do you know a more efficient way to store and retrieve a waveform?

[3d0g]

I just created a waveform containing cosine and sinusoids, each with 2.5 million points, using 26 such f(t) binary files as discussed, above, and construction (summation,) which included 26 f(t) data loads, of course, took maybe a second and then full plots of the magnitude and phase of the FFT took another second.  Indeed, binary files are the way to go.

Now, following one other thing, storing the FFT is next.

ASCII proved the concept, and binary sent it all to warp speed.  (By the way, no LabVIEW slowing yet either.)

Again, thank you all for your help/conversation.   

[Kevin_Price]

Just to chime in again with one feature that can make TDMS preferable to raw binary: it's a defined format that's internally self-describing.  So any program that can interact with the TDMS format will be able to decode the binary data in it.

When you save to a regular binary file, there is no specific standard so other apps are very unlikely to be able to interpret it correctly.   Is it DBL or SGL precision?  Is it waveforms or is it X data and Y data?  Are the X and Y interleaved or sequential?   Is there a header to include metadata?  How long is it? What format is it in?

None of these things are inherently defined in a regular binary file.  The reader needs to know exactly what the writer did to have any chance of recovering the data.  But in a TDMS file, *all* of these things will be defined inherently for any TDMS-capable program to recover.

I think you'll find that TDMS is of similar speed to regular binary, perhaps not *quite* as fast, but in the same ballpark.  At least give it a try, even if you don't use it this time around.

-Kevin P