Currently there may be errors shown on top of a page, because of a missing Wiki update (PHP version and extension DPL3).
Navigation
Topics Help • Register • News • History • How to • Sequences statistics • Template prototypes

LL test distribution

From Prime-Wiki
Revision as of 10:14, 7 March 2019 by Karbon (talk | contribs) (navbox)
Jump to: navigation, search

The current distribution of LL tests can be viewed at http://www.mersenne.org/primenet

At any given instant, most of GIMPS's LL tests fall roughly into one of 2 categories. As time passes and GIMPS exhausts smaller exponents, both leading edges tend to move upwards, towards larger exponents.

Double Checking Leading Edge

Last update: 2008-09-01

This refers to the relatively narrow band in which most current double checks are taking place. As of Sept. 2008, most double checks are in the low 20 millions.

First time LL Leading Edge

This refers to the relatively narrow band in which most current first time LL tests are taking place. Mersenne prime discoveries almost always take place in this range. As of Sept. 2008, most first time tests are in the high 30 millions or low 40 millions.

Back when first time tests below 10 million digits were plentiful, there was a "10 million digit Leading Edge" as well, which originated at 33.2M and gradually moved upwards. But as first time exponents below 33.2M were gradually exhausted, this edge merged with the First time LL Leading Edge.

Other LL tests

A small minority of LL tests are well outside both leading edges.

A few LL tests are well below the double checking leading edge. The typical reason is that the tester is using very slow hardware, causing the test to take years, in which time the double checking leading edge moves past the test. As of Sept. 2008, the smallest ongoing LL test is around 18 million.

A few LL tests are well above the first time leading edge. This usually happens because the tester is very ambitious; such a strategy actually reduces a user's chances of discovering a new prime because large exponents take longer to test, so that less tests are completed per unit time. As of Sept. 2008, three exponents between 50.1 and 50.2 million have been checked out.

As of Sept. 2008, the largest Mersenne number ever LL tested is 2100,000,007-1, which has 30,103,002 digits. It was first tested by William Christian in 2006 and successfully double-checked the next year. It gave each user 79.417 CPU years (403.0413 GHz-days) of PrimeNet credit upon completion. However, an even larger LL test is already underway...