Before we proceed about QueryLocator class and let's see the limits of it.
I wrote a sample Debug statement to see the limits of the number of rows returned by query locator and Normal Square bracket  query
System.debug(Limits.getLimitQueryLocatorRows()); // 10000
So in a Synchronous transaction when you use
Database.getQueryLocator(10,000) you get less number of rows than normal
 Square bracket query(50,000).
Now, let's see how to use QueryLocator. QuerryLocator is nothing but a query cursor. Which is pretty useless on its own, but it can help us get an Iterator. And that iterator can help us iterating through that set of records.
Database.QueryLocator ql = Database.getQueryLocator([SELECT id from Account order by CreatedDate desc Limit 10000 ]);
Database.QueryLocatorIterator iterator = ql.iterator();
Account acc =(Account) iterator.next();
//Work on your list
Bit tedious aint? Such a pain....
Why would one want to use QueryLocatior instead of Normal Square bracket querry?
The answer lies in HEAP. As you can see from the code above, the Iterator can get only 1 record at a time, thus taking less memory, where as the SOQL  gets the big List which can sometimes cause HeapLimitExceeded exception. Thus if you are querying large number of fields and processing them and have concern that you can reach heap limitations, it's better to use Querrylocator instead of normal SOQL.