View Single Post
2019-12-16, 02:22   #4
kriesel

"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

592410 Posts

Quote:
 Originally Posted by jshort Interesting that the p-1 test isn't used straight away, but rather a trivial Sieve of Eratosthenes. Anyways, I'd imagine that the bounds B and B' must be awfully small.
The order of TF first, lowest bits first, P-1 later is the outcome of doing first, what gives the most factored per unit effort, and so maximizing time saved.
https://www.mersenneforum.org/showpo...23&postcount=6
P-1 bounds are typically optimized to maximize probable time saved, also.
https://www.mersenneforum.org/showpo...7&postcount=23
An appropriate set of P-1 bounds is a rather lengthy computation compared to the beginning TF levels.

Last fiddled with by kriesel on 2019-12-16 at 02:23