View Single Post
Old 2005-06-28, 09:23   #33
dave_0273
 
dave_0273's Avatar
 
Oct 2003
Australia, Brisbane

1D616 Posts
Default

I have often thought, and have often been asked why we don't do p-1ing ahead of the leading edge of LL testing. Basically those two reasons are why I have never tried it.

There is no way (with the current amount of people in mersenne-aries) that we could keep ahead of the leading edge. We could only do a small percentage.

The other thing that we would need is for another group to be working with us doing the trail factoring before we do the p-1.

However, one the biggest reasons why I have never bothered to go ahead of the leading edge of LL testing is because if people really wanted to do that sort of work, they could always do it the semi-automated way. They could reserve 50 or so exponents, do the p-1 and then release them again. Because the work is therefore technically assigned through primenet, they wouldn't be duplicating work and they wouldn't be stepping on each others toes.

This doesn't work with mersenne-aries because you just can't get enough work this way. The majority of exponents are p-1ed on the first LL test. Even when I tried to reserve 50 exponents, I would find that I would get less than 5 that would require a p-1 test.
dave_0273 is offline   Reply With Quote