As Robert says, things like sorts are pretty nippy already. If you are just merging two files, that's a pretty simple thing which doesn't leave much room for optimisation. It reads a record on file 1, reads a record on file 2, compares the keys, writes out the lower key, reads the file that had the lower key, repeats until end.
How often will the job be running?
How many key fields are defined?
What type of files are the input and output? Ensure optimal blocking.
Try to dig up a manual. I'm sure it will have advice on treating big files.
Don't expect any big savings, so consider the first question, which is asking "is it worth it"?