Which gives good performance? separate database vs separate schema

Hello Friends,
I have many instances of my software running on a server (Solaris SPARC). Each software instance requires some DB tables (same DDL for all instances' tables) to store data.
It essentially means that some processes from each instance of the software connect to these tables.
Now, should I put these tables in 1 Database's different schemas or in separate databases itself for good performance?
I am using libpq for connection.  

Pictorial Representation:

Process1 -> DB1.schema1.table1

Process2 -> DB1.schema2.table1

  Vs.

Process1 -> DB1.default.table1

Process2 -> DB2.default.table1

Which one is better?


thanks in advance