Can I specify the range to search the Mersenne Prime?

[Dubslow]

Originally Posted by Unregistered

Just only test the M47, how long the time it will take? One day , or a week?

See Brian's post above; keep in mind that as he said, current exponents handed out by PrimeNet are much higher than M47. (M47 would take 8-9 days on one core of an Intel Core i7-2600K slightly overclocked, and something in the 55M-58M range as you're likely to get from PrimeNet would probably take 3-3.5 weeks.)

[jasong]

Possibly stupid question, but have we found 47 Mersennes? It sounds like the person doesn't speak English as a first language, so might not sufficiently understand what the Prime95 program does.

[Batalov]

You remind of a nerdy anecdote:

Einstein, Newton and Pascal were playing hide-and-go-seek. It was Einstein's turn to seek, and as he began counting Pascal ran and hid behind a bush. But Newton picked up a stick, used it to draw a metre square on the ground, and sat in the middle of it. Einstein finished counting, turned around, and immediately said "Aha! Found you Newton." but Newton replied "Ah, but a newton per a metre squared is a pascal!

[LaurV]

Originally Posted by jasong

Possibly stupid question, but have we found 47 Mersennes? It sounds like the person doesn't speak English as a first language, so might not sufficiently understand what the Prime95 program does.

No, we don't need to find Mersenne numbers, we know all of them, and there is an infinite amount of them. We were able to prove primality for 47 of them only. There are no stupid questions... :P

@Batalov: ROFL, you made my day... hehehe...

Originally Posted by LaurV

No, we don't need to find Mersenne numbers, we know all of them, and there is an infinite amount of them. We were able to prove primality for 47 of them only. There are no stupid questions... :P

@Batalov: ROFL, you made my day... hehehe...

Yet! Exactly, I started from this wiki page, then asked a stupid question.
Is there anybody thinking I am searching for the Mr. Mersenne or the Mersenne number, not for the Mersenne prime?

[science_man_88]

Originally Posted by Batalov

You remind of a nerdy anecdote:

Einstein, Newton and Pascal were playing hide-and-go-seek. It was Einstein's turn to seek, and as he began counting Pascal ran and hid behind a bush. But Newton picked up a stick, used it to draw a metre square on the ground, and sat in the middle of it. Einstein finished counting, turned around, and immediately said "Aha! Found you Newton." but Newton replied "Ah, but a newton per a metre squared is a pascal!

you could also make it Einstein, Joule, and Pascal with Joule hiding in a 1 cubic metre box.

[Dubslow]

Originally Posted by Unregistered

Yet! Exactly, I started from this wiki page, then asked a stupid question.
Is there anybody thinking I am searching for the Mr. Mersenne or the Mersenne number, not for the Mersenne prime?

If you're using PrimeNet, you're fine.

[Christenson]

Originally Posted by Unregistered

Thank you , your answer (the second part) is what i want!

OK, a little more detail:

"smaller" Mersenne numbers (less than 2^10,000,000-1) or so have ECM curves run on them that sometimes find factors. These are relatively quick; a few hours apiece...for a few percent chance of finding a factor.

Around 2^27,000,000 or so, begin a large number of Mersenne numbers that have been reported as composite by a single Lucas-Lehmer test. Some of us work that range of exponents, since the chance of a computer hardware error in such long Lucas-Lehmer tests is up to a few percent, depending on just how hard your hardware is being pushed. These are the "double checks".

There's also trial factoring; to do this effectively, you need a GPU running CUDA or OpenCL. These cards outperform CPUs on this job by 100:1. Trial Factoring (TF) eliminates numbers from LL(Lucas-Lehmer) tests by finding factors; these can run in an hour or two, but it may take a day or three to find a factor, and the problem gets exponentially harder as the bit level goes up.

Then there's P-1 testing. This is a sophisticated factoring method that can find factors of a certain form, and can run for several days on a given exponent, with typically 3-7% chance of finding a factor on a typical primenet assignment. Such assignments run a day or two.

Around 2^50,000,000-1 or so are the first time LL tests. If you are really nuts, you can have your PC spend a year or more completing a test around 2^333,333,333 in the search for the first 100-million digit prime number. But it's not for the faint of heart; the probability of an error in your calculations goes up another order of magnitude or two, making errors a near certainty.

Since noone else has said it, unregistered user, I will: Your efforts on behalf of primenet are appreciated, whatever part of the problem you decide to set your computer(s) to working on, however casually. Thank you for joining GIMPS and contributing.

[Dubslow]

Heh, so many bad word choices/terrible puns...

A few notes: The first time LL tests are closer to 2^55,000,000-1 or 2^60,000,000-1, and depending on the particular settings, I can usually do a P-1 in less than a day.

What is most useful for GIMPS is P-1, at the moment, which gets the least work relative to the other work types. You do however need (typically) at least 1GiB of RAM available to do that.

If you don't want to do that, then just do either first time primality tests (LL tests) or double checks. At the moment, less double checking goes on compared to first time tests, so some of us prefer to work on those instead. (Because the double check exponents are ~ half the size of first time tests, they run 4-5 times faster.)

Originally Posted by Dubslow

Heh, so many bad word choices/terrible puns...

A few notes: The first time LL tests are closer to 2^55,000,000-1 or 2^60,000,000-1, and depending on the particular settings, I can usually do a P-1 in less than a day.

What is most useful for GIMPS is P-1, at the moment, which gets the least work relative to the other work types. You do however need (typically) at least 1GiB of RAM available to do that.

If you don't want to do that, then just do either first time primality tests (LL tests) or double checks. At the moment, less double checking goes on compared to first time tests, so some of us prefer to work on those instead. (Because the double check exponents are ~ half the size of first time tests, they run 4-5 times faster.)

hahaha, you are scared? because i want the second part of ......

[davieddy]

Originally Posted by Unregistered

Can I specify the range to search the Mersenne Prime?

The short answer is "yes".
If you want the long one you'll have to consult George (whoever he is) or Chalsall.

Don't ask me: I'm bard.

David

PS If worse comes to worst, try Bob.