P-1 and srfile

[henryzz]

when would you want to use:
srfile -p 2
what sort of project would need this

[henryzz]

why has no one answered is it because i didnt explain my question properly or because people dont know an answer
i will restate my question
why would a project need to be able to p-1 factoring optimal for if a factor is found several primality tests will be saved
when would this be necessary

[mdettweiler]

Originally Posted by henryzz

when would you want to use:
srfile -p 2
what sort of project would need this

Some k*b^n+-c projects can reap slight benefits from P-1 factoring; even though P-1 on those numbers is not nearly as efficient as it is on Mersenne numbers, sometimes it can still be worthwhile. Seventeen or Bust and Riesel Sieve would be good examples of projects that use P-1 factoring in this way.

Hope this helps!

Max

[henryzz]

when would finding a factor save several primality tests

[mdettweiler]

Originally Posted by henryzz

when would finding a factor save several primality tests

Depends on whether or not the project plans to do doublechecking, and the status of their doublecheck effort. For example, say you have a number that's farther ahead in the sieve range and hasn't been tested at all yet. A factor for that number will save two tests. Likewise, a number that has been tested once but not doublechecked yet will have one test saved if a factor is found. And, of course, if the number has been both tested and doublechecked, then the factor saves zero tests.

[axn]

What has srfile got to do with P-1 factoring?!

[mdettweiler]

Originally Posted by axn1

What has srfile got to do with P-1 factoring?!

Srfile has an command line switch that will let you choose to output a sieve file as Pfactor= lines in a Prime95-compatible worktodo.ini file, along with a number of more traditional sieve file formats.

[masser]

P-1 factoring will become useful for the Sierpinski/Riesel Base 5 project before we reach the end of our dat file.

The largest numbers in our dat file are of size 5^n, where n = 2,000,000. Sometime after our testing queues reach the n>1,000,000 mark, P-1 factoring will become useful.