4-5M Reservations

[paulunderwood]

updated 5:55:56 AM, Tues, Sept 23, 2008

ALL DONE - PROJECT FINISHED

These mega-prime candidates have at least 1323327 decimal digits
They have been sieved for divisors less than 271.2 trillion
Here are 249 candidates contained in 4 input files ready for LLR testing

An average file of 58 candidates will take about three weeks to process on an average Pentium computer -- each number taking around 14 hours; 64-bit AMDs 15 hours per core; 2.4Ghz Core2Duo/Quad around 10 hours per core

The 321 Reservations System :
(0) Make ready your system, maybe by creating as many directories as the number of cores you want to use for this project
(1) Download and decompress the program LLR-3.7.1c (Linux or Windows)
(2) Choose a range from the list below that no one else has recently reserved further below
(3) Leave a message here to reserve your chosen range
(4) Download the corresponding input file(s) -- all done
(5) Run LLR with your new input file(s)
(6) When finished, email the results file(s) to paulunderwood@mindless.com
(7) To continue go back to step 2

(Alternatively to the above steps, just email me.)

Range of a 1000 n :: Candidates in file
4396000-4397000 :: 61
4397000-4398000 :: 60
4398000-4399000 :: 60
4399000-4400000 :: 68

[paulunderwood]

If you book out any of the remaining 4 files above, please ensure that you can complete them by the end of June.

You can always join PrimeGrid and crunch individual 321 numbers through the BOINC interface. Each will take about half a day and you might find a mega-prime

[Lennart]

4396000-4397000 :: 61
4397000-4398000 :: 60
4398000-4399000 :: 60
4399000-4400000 :: 68


I reserve the last 4 files.

/Lennart

[jrk]

Paul are you going to be adding more ranges?

[paulunderwood]

Originally Posted by jrk

Paul are you going to be adding more ranges?

No! I might have a few sub-n=4M if James can not do those himself.

I hope we can get all sub-n=5 done by the end of June

[mdettweiler]

Originally Posted by paulunderwood

No! I might have a few sub-n=4M if James can not do those himself.

I hope we can get all sub-n=5 done by the end of June

Once all n<5M is done, and sieving for 5M+ is complete (or at least ready for 5M-6M to be broken off), will files in the 5M-6M range be made available for manual testing, too, or will it be PrimeGrid only?

[paulunderwood]

Originally Posted by Anonymous

Once all n<5M is done, and sieving for 5M+ is complete (or at least ready for 5M-6M to be broken off), will files in the 5M-6M range be made available for manual testing, too, or will it be PrimeGrid only?

PrimeGrid only. PG are sieving n=5M-10M already. It will be up to them when they decide to release more (only on their server.)

[mdettweiler]

Originally Posted by paulunderwood

PrimeGrid only. PG are sieving n=5M-10M already. It will be up to them when they decide to release more (only on their server.)

Ah, I see. So, then, will the 321 project just die out, or will it pick a new k to work on? (Or maybe even a new base--such as 3*3^n-1?)

[paulunderwood]

Originally Posted by Anonymous

Ah, I see. So, then, will the 321 project just die out, or will it pick a new k to work on? (Or maybe even a new base--such as 3*3^n-1?)

"321search" is near its end, but PG will carry on searching for primes of the form 3*2^n-1 in conjunction with 3*2^n+1. "321search" will not continue.

[jrk]

Does PrimeGrid still assign double-checks at the same time as first-time tests? It was doing that a few months ago when I was checking it out.

[paulunderwood]

As far as I know, yes. Double checking becomes more important with the bigger numbers because there are many helpers needed, some of whom might have defective hardware, and because general hardware failures can occur anyway.