You will not need a date dimension always to show a month wise trend neither will it significantly improve your query performance as you can have a month object in your universe using a to_char(table.date,'MM/YYYY') and use this object for your trending in the webi report..
How much data size are we talking of here.. you can also create a aggregate table just with monthly data and use aggregate awareness function in universe in the database which can significantly reduce run time...
Also please make sure indexes are built on the column you are filtering on and if there is any joins which are there with high volume tables use the key or indexed column and defined index awareness on them..