How many more or less RTTs is required for a :

* non-persistent connection with 4 parallel connections sending 8 objects with a 1 RTT transmission delay per object, than for a

* persistent connection with pipelining sending the same 8 objects with the same transmission delay?

Assume you can have no more than 4 parallel connections or pipelined objects per request. Your answer should be a positive integer if non-persistent requires more RTTs, a negative integer if it requires fewer.