Uploading is transferring the result file from your hard drive to a hard drive at the project.
Reporting is accessing the database at the project to tell the outcome of the task (correct, failed, other) and to signal it's ready to be checked by their validator. Reporting takes a lot of overhead (CPU, RAM) on the server, and reporting one task takes as much resources as reporting 100 tasks. So projects rather have you report as many as possible at the same time, rather than one at a time.

Some projects have one large data file on your drive and you don't upload any results, just report the outcome of the task just done before being told what to do next.

---

And at least one project uploads intermediate results. CPDN including testing site can upload anything from about 3 zip files at the end of a task to over 20 zips at intervals during the task run, depending on how many model months the task runs for. The zips with the actual data in them can vary between about 5 or 6 MB up to almost 1GB on a couple of testing batches.

It all depends on what the project is looking at. Prime Grid I assume doesn't have to send back much more than whether it has found a new prime number or not. CPDN sends back data on the model climate predictions which are saved from lots of points during the task run for atmospheric pressure, wind speed, humidity, temperature etc.