Here is an update. It seems that we can compress losslessly from 32-bit to 24-bit using 4472B in LabWindows/CVI.
Assume that:
1, x channels, y sample/channel and so total z = x* y samples to record.
What I did:
I. Save the lossless compression data:
1, Set the channels to lossless compression mode.
2, Set a pointer to the array to store the data as raw_data = (unsigned char *) calloc (4*z, sizeof (unsigned char)). Note that the array size is 4*z bytes, which is the same as the non-compression mode array size.
3, Use ReadRaw to read the accquired raw data. Checking the array (raw_data) returned by the function in CVI debug environment, one sees that the first 75% of the array is occupied by data (if the inputs are not zeros) and the last 25% are all zeros. If no compression mode is set, then the array of the same size are full of non-zero data.This seems to be consistent with gaving's comments, i.e., the driver packs the data if the lossless mode is set.However, the driver doesn't automatically drop the last 25% of the array.This also explains the array size in #2.
4, Use fwrite (3*z, sizeof (unsigned char), FILE * file_pointer) to write the first 75% of the raw_data to a file.
II. Load the lossless compression data:
1, Since every datapoint is now composed of 3 bytes of raw data in the file. So read 3 bytes everytime and form it to a long number.
2, I am not sure why, but it seems that the most significant byte comes first, which seems to be contrast to the LittleEndian setting. So I use
val = (val<<8) + *(raw_data+i*byte_per_number+j); to convert the 3 bytes of raw data to one data point. Here i is an index number from 0 to the total number of the data points I am expecting and j is the index number looping from 0 to 2 to load 3 bytes from the raw data file.
3, Scale the raw data with the calibration coefficients. More details can be found in this post: http://forums.ni.com/ni/board/message?board.id=100&thread.id=680&view=by_date_ascending&page=1
Again, thanks a lot to gaving and Todd. I do look forward to more explanations and comments to what was described above, especially II-2.
XR
