Does the file gateway cache need to be at least as large as the largest file being uploaded?

From the FAQs: "writes could fail if there is no free cache space to store data locally pending upload to S3." This seems to suggest that the file needs to sit in that cache space before it can be uploaded into S3, which makes me think that the cache needs to be at least the size of the largest potential file (I guess in my customer's case, 5TB).

My customer cares less about the performance benefits of the cache (because they are using file gw for backups that they don't plan to access) and more about the limitations that it introduces for file uploads. The documentation says that the "cache disk size must be tuned to the size of the active working set." Basically the only active working set for my customer will be the files that need to be uploaded. They are not anticipating a need for continuous retrieval.

So my question is- does the file gateway cache need to be at least as large as the largest file being uploaded? Or larger potentially? Any suggestions on how to choose the size of the cache?