View Single Post
Old 2020-05-11, 14:32   #107
VBCurtis's Avatar
Feb 2005
Riverside, CA

450310 Posts

We surely can try mfb0 = 59, but we may as well try that on a 31/32 job that we already expect to be quicker.

Let's start a new thread for 165-170 digits. I expect this will be on the cusp where 2LP and 3LP are about the same speed.

The short answer (compared to C178 settings): Cut poly select by 75% by halving admax and halving P.

Use the 31/31 settings you just tried, but with A=28. Given the lower duplicate rate you saw on A=28, I'd drop Q-min to 10M. You could drop lim's a bit, perhaps 10M each, but it doesn't matter much. You already have strong grasp of how to control rels_wanted, but I expect to need 10% fewer relations for 10 digits smaller.
VBCurtis is online now   Reply With Quote