Sieving efficiency statistics?

[dudlio]

Has anyone looked through the project database to figure out how much sieving has helped (avoid) PRPing? For example, each sieved factor should fall into one of these categories:

* Useful: Sieved factor has avoided a PRP test. PRP test was queued and then dropped when sieved factor was found.

* Possible/Unknown: Sieved factor may avoid a PRP test, but PRPing hasn't gotten that far yet.

* Obsolete: Sieved factor has already been PRP tested.

I think this categorization would put a lot of this "is sieving worth it" debate to rest. Can we keep track of this metadata, if not retroactively, at least from now on?

[shauge]

It is my understanding that this is exactly what the sieving stats are based on. You can read these numbers from the stats.

In addition you can also see how many factors that just saved a PRP test in the double check run, but not in the main run.

[dudlio]

Thanks, I hadn't seen these stats before.

So let me get this straight: For 12M<n<13M, twenty thousand PRP's were/will be issued (first run) and fourteen thousand tests were saved by sieving. Without sieving, thirty five thousand PRP's would have been issued.

Is this correct?