Sounds like you are doing a lot of guess work to identify the issue.

Run a server side trace capturing RCP:Completed and SQLBatch:Completed events with ( at very least the cpu column). With this run the output through ClearTrace http://www.scalesql.com/cleartrace/download.aspx

You will now know *which* actions are having the highest impact.

Once you have done that feel free to post back with your finding for better targeted advice