View Single Post
Old 2020-05-06, 21:40   #104
charybdis
 
Apr 2020

5×23 Posts
Default

Here's some more duplication rate data:

Code:
qmin      raw        unique
500000    367310494  213177040
3000000   348579594  216387208
5000000   339080413  217856601
7000000   331657087  218727230
10000000  323179335  219755058
13000000  316605494  220579118
17000000  307823056  220314561
20000000  303177219  220593021
For each run I chose a range of Q that took roughly the same CPU-time to sieve, as guessed from the timing data in the logfile. As the clients don't send in their results in exact order of Q, there will have been some slight variations, which presumably explains the dip in unique rels with qmin=17M compared to 13M and 20M.

It looks as if 15M or 20M would be a good value to start at.
charybdis is offline   Reply With Quote