Recently, I've been using a sieve depth of 47 since the sieve has totally completed up to 2^47 ~ 140T.

But it may soon (or already) make sense to use 48. We have already crossed the 90% sieve point to 200T and much between 200T and 281T has been sieved as well.

MikeH or Nuri would know best here. Does somone know what percentage of the factors between 2^47 and 2^48 have been been found already by sieving? I would guess that even though less than 50% of the 2^47-2^48 range has been sieved, probably well over 50% of the total factors in that range have been found. If that's the case, I think the time to switch to the higher depth number is now.

Mike, how do you feel about adding a stats page somewhere showing the % of factors left to find in each power of 2 range? A page that showed something like:

.
.
.
45-46: 00.01%
46-47: 00.12%
47-48: 32.84%
48-49: 89.23%
.
.
.

Of course these numbers will nessisarily be estimates since you can't know in advance exactly how many factors are in each range but a rough idea is all we need. It could then conclude at the bottom what sieve depth to use for factoring by choosing the high end of the highest range that has a density less than 50% left (48 in my example numbers.)

Another reason to move to 48 is because we currently don't have enough computing power devoted to factoring to do an optimal level of work anyway. So if we put the sieve depth 1 higher, it will speed up (aka lower the bounds of) the tests making it possible to cover larger ranges before SB gets to them.

I look forward to hearing your comments.

Cheers,
Louie