Problem with these kind of question is that even a hypothetical scenario isn't really conceivable. The ends are just so far apart, it's hard to say anything.
Assuming the bandwidth is not an issue, if you take a pure html website, a single celeron from years past will have no problem serving millions and millions of requests. Comparatively, a faulty or even a very heavy processing script could take ages to complete a single request. So, the single greatest factor is the question of what you are running on the server.
You can use benchmarks to relatively compare one system against another in certain aspects. But even benchmarks can't tell you anymore than that.
If you want to be in a position where you can make numerical estimates of how much you can handle, you will need to run your application under load (test) and make expectations from it. That is the only way. That is, benchmark it against your application.