What T level should we sieve to for n<20m

[vjs]

Does anyone know what T level we should search upto for n=20m.

As we go into a higher and higher T range we get deminishing returns for our computational efforts find more excluded (previously found factors, duplicates, etc.).

There was a site somewhere where it was discussed that we should reach T=300T before n=6m. Sieveing beyond 300T for n<6m would be ineffective and your better off doing P-1 or prp until the n reaches 6m. So since we are using a dat file that reports only upto n=20m at what point should we stop sieveing for n<20m?

I know it's way in the future but is it 1000T 10,000T or 600T?

[royanee]

Seems like the resource allocation model says that it's valuable to us to go until we hit 5800T or we finish the test with n < 20m. Of course, that figure is for 16.5m, so at least 5.8P.