Why replicate? I'm not sure what you want to do. The distributed transaction is less of an issue on the same server, with SELECT queries, and more with writes.

If you use a new filegroup, you do a few things. One is you keep the data together and consistent with backup/restore, recovery, simple transactions for changes, etc. Second, no code changes for queries. Third, you can get the same performance with a separate filegroup that you would get with a new database on the same server. Separate the IO out to different storage.