blocking vs async using guzzle

I am implementing code to write out a copy of my collected data to Amazon S3. S3 uses Guzzle and allows the calls to send the data either sync or async (promises). If each instance of my php program is truly a separate process, then synch should not be a problem. However, if these are really being handled by a single multi thread process then I must do async. 

Does anyone have a deep understanding of how Apache and php interact or can you point me to a white paper? Clearly synch is much easier to implement.

proof that an old dog can learn new tricks