Being pedantic (and a smartass) the same, "real-time" original definition is that some calculation/reaction of the system is guaruanteed to be always bounded an

Being pedantic (and a smartass) the same, "real-time" original definition is that some calculation/reaction of the system is guaruanteed to be always bounded and within the system rqequirements (usually for it not to fail). So depending in the use case and application, real-time can be sub-second, or even seconds/minutes. E.g. if the instagram folks require likes to happen in real-time for all users around the world that is fine to be within seconds, but a safety critical application in a car must respond within milliseconds.. "you don't need REAL-time, just sub-second latency" in general does not make much sense, sounds more like this application real-time requirements just lie in the sub-second resolution region?!