|
It's been six months since I've evaluated how much GFN sieving we need. I determine the optimal sieving level for the work we anticipate testing with genefer over the next five years. I had been using the work we've done from the time each subproject opened in BOINC to the present to gauge the future. This time I did it another way as well - looking at just the last six months worth of genefer work. As I expected, that gave much larger optimal sieving numbers and I've chosen to use those now. Here's the results:
GFN15 and GFN16 are oversieved and will not reopen. If at some point genefer can test beyond b=400P then those will be reevaluated.
GFN17 will cease sieving at p=140E (140000P). I had calculated the sieving needed for each of GFN17 low and GFN17 Mega separately and then added them. Really, I should have used whichever gave the higher value. It didn't matter before because that level was so much higher than where we were up to. But with GFN17 at around 109E as I write this, I used the 140E level of GFN17 low. I do not anticipate GFN17 sieving reopening again because we'll have tested so many more candidates by six months from now. Sieving covering candidates already tested is meaningless and drives down the optimal level.
GFN18's maximum p has been extended from 110E to 160E. It stays open and will probably extend further in six months.
GFN19 is reopened and the max p goes from 86E to 160E. I expect that will be extended further in six months.
GFN20 is reopened and the max p goes from 86E to 110E. We're not running nearly as much GFN20 on BOINC as the others, hence the lower (and counter-intuitive) optimal point. Yes, I did check to make sure it wasn't an error.
GFN21 stays open, the max p goes from 211E to 310E.
GFN22 is reopened. Max p goes from 217E to 360E. Under the old way I calculated this max p would have been 264E, so you can see how much difference the new formula made.
In addition, I've changed the sieving graphs so that they don't start at p=0 any more. The maximum width of the .png files created by PHP is 32767 pixels and I'd rather have more resolution horizontally to be able to catch any sieving anomalies. I've also changed all the graph colors so that the dots show in in black as it's easier to see the data that way.
One last change, the maximum reservation for GFN21 is now 400P, matching GFN22, for users who are able to make reservations that large. Any user with a good track record in sieving who might want to make 400P reservations for GFN21 or GFN22 should PM me. Many of you already can. | 
