C2000 DSP executation time calculation

Hello, I'm trying to measure the duration of my operation with TMS320F28379D C 2000 dsp. When I go step by step, it says there must be an atomic subsystem in step 4. What does atomic subsystem mean? I also want to learn how to send the block used for memory copying via serial communication. I have another question. At what stage will this measure the processing time, while the program is being sent to the dsp or online while the dsp is running? Is there any other way to measure this time?