Sorry to to be vague, but the answer really depends on what you are trying to do. If you are trying to do reads from different channels, you know you can use the ':' and ',' operator while creating the task and then use one read or write VI.
Well said. In addition, using serial would create data flow dependencies, so the reads could not happen at the ame time, where parallel would remove the dependencies.
Consider error handling. If one task fails to initialize, are the others still 'safe' to run?
I use parallel DAQ a lot, but I sync their 'parallelism' on the hardware layer (PXI or RTSI). Code (at the hardware abstraction layer) is about catching the errors correctly.
