Allocation of prping resources

[tqft]

Ok this has probably been done ages ago but I could not see/find it.

If there is a link please let me know.

I assume some logic has been used in deciding how many clients to allocate to a particular k.

k Total Pending
4847 18034 436
5359 23394 616
10223 22467 507
19249 7431 156
21181 18698 430
22699 7517 185
24737 20142 454
27653 12342 264
28433 9964 275
33661 18660 426
44131 2891 prime
46157 338 prime
54767 4665 prime
55459 27545 640
65567 167 prime
67607 6917 156
69109 864 prime

Is it proth weight? Or is there some other logic logic/guesstimate/hunch about which one next driving the allocation?

It sure does not seem evenly distributed, I am looking at a snap shot but the numbers seem to out of kilter for anything else.

eg 28443 - 9964 complete tests 275 pending is very low compared to 55459 (27545 tests, 640 pending).

If the allocation of prping was waiting until more sieving/factoring results become available I would expect the higher k's to be disproportionately starved or prping but that does not seem to be the case. A naive guess says the higher k's would be expected to more likely have factors.

[ceselb]

No logic used except sorting by n.
The next test to be assigned is the one with lowest n. So you could say that proth weight plays an indirect role here.

[MikeH]

If you take a look at the "Remaining candidates (by k)" table on this page you should see a close correlation between the proportions of candidates remaining and those being PRPed.

[tqft]

I can see the pattern and the logic now.

Thanks.