:xmastree: Well, we made it to 3 million bits before the end of this year -- thanks to everyone who has helped so far. The new [URL="http://www.mersenneforum.org/showthread.php?t=6808"]reservation page for 3-4 million bits[/URL] is up -- get yours. Hopefully, we will reach 1 million digits (or 3.33 million bits) by August 2007. There will be a FFT jump at about 3.12 million bits. We should also find a prime soon -- after all that is the purpose of this project.

We are at nearly 3.1 million bit tests and it is not the end of January yet :geek: That's about 80% towards a million digits. We are due a prime, but, fingers crossed, that prime will occur at just over a million digits :wink: But who knows? [URL="http://www.mersenneforum.org/showthread.php?t=6808"]Join us and chance your arm[/URL].

[QUOTE=paulunderwood;96826]We are at nearly 3.1 million bit tests and it is not the end of January yet :geek: That's about 80% towards a million digits. We are due a prime, but, fingers crossed, that prime will occur at just over a million digits :wink: But who knows? [URL="http://www.mersenneforum.org/showthread.php?t=6808"]Join us and chance your arm[/URL].[/QUOTE]

IF you find a million digit prime soon (which I hope you do), will you still continue to LLR to 5 M?

[QUOTE]IF you find a million digit prime soon (which I hope you do), will you still continue to LLR to 5 M?[/QUOTE]

Hardly, No. Personally, the main goal is a 1 million plus digit prime. The effort would have to be redoubled to find the following prime, most likely and that could be greater than 5M bits :cry: So it just a final push now. Help us :smile:

Will help soon! (As soon as I can free up a computer).

Thomas wrote:
[QUOTE](Just as a note: The next FFT length change is at n=3121158. Tests will take about 20% longer above that n-value.)[/QUOTE]

Thanks for the reminder. This new FFT size will last until n=3719191, i.e. for the rest of the year. 32-bit athlons can continue at the old length until n=3171158.

That is now 5 hours per test on a Pentium 4 at 3 GHz. Quicker on a Core 2 Duo.

[QUOTE=paulunderwood;98269]32-bit athlons can continue at the old length until n=3171158.[/QUOTE]

The above holds for version 3.5 (or below) of LLR. For LLR 3.6 and higher the 32bit Athlons will use the old length until n=3180158.

The make the things even more complicated: We found LLR 3.5 slightly faster than the newer versions on the Athlons (a fraction of a percent). So the Athlon testers might use version 3.5 up to n=3171158 and then switch to version 3.6/3.6.2/3.7...

Welcome back James!

I have asked Xyzzy to upload the next set of LLR input files, covering 3.15M to 3.3M bits, which will take us very close to 1 million digits (3.32M).

Tests take nearly 6 hours each on a Pentium 4 at 2.66GHz. An average file will take 2 weeks -- with 58 numbers. This will be the case until about 3.7M bits.

Thanks to Geoff for the new sieve, which Carlos and Thomas used to get the least divisor up to 245 trillion.

We are due to find a prime :smile:

If we find a prime now it will be 12th largest known.

Good luck all :grin:

Thanks Paul, it's good to be back.

I have just completed the last sub-3 million bit 321 candidate. Great work people. :tu:

Our next goal is to reach 1 million decimal digits by the end of May. After reaching this mile stone I hope we find another prime quicky.

In the long term, it would be great to get to 3.6 million by the end of this year and then a 4 million bits target for the end of 2008 will be easily attainable.

Our next prime is due at 4.1 million bits. However we do not know where it actually is. It could be much smaller or much larger.

Good luck all with finding our next 321 prime! :wink:

It looks as though we will miss the end-of-May target for all candidates below one million digits to be booked out, but we should easily get there by the end of June :smile:

For [I]all[/I] 0<n<3M, Thomas and I have been testing pre-project candidates or those missing numbers due to human error and those tested with PRP/PFGW, such that by the end of May we will have, uniformly, a divisor or a LLR residue :flex:

My effort will slow down in the hot summer months, but hopefully with people being interested in finding a mega-prime, the 3.6M goal is still on for the end of this year.

Anyone who wants to make a start on mega-prime tests should email me.

Here are some relevant project links:
[URL="http://www.mersenneforum.org/showthread.php?t=6874"]Current status[/URL]
[URL="http://primes.utm.edu/bios/page.php?id=479"]Prime Pages Bio for 321[/URL]
[URL="http://www.mersenneforum.org/showthread.php?p=50474#post50474"]Contributor Statistics[/URL]
[URL="http://www.mersenneforum.org/showthread.php?t=6808"]Reservations[/URL]

:help:

Thanks to Xyzzy for uploading the latest 321 input files ready for LLR crunching, covering from n=3.3M to 3.45M. :tu:

We are also grateful to Thomas for further sieving these to 271.2 trillion. :bow:

This batch will take us over 1 million decimal digits (3.32M) :w00t:

Grab yourself some and maybe you will find a mega-prime :love:

Also we have covered all n from 0 to 3M with a factor or an LLR residue. After some more analysis of errors, we will make all public...

We are now testing candidates for a prime number with over 1 million digits, a so called mega-prime :w00t:

Thanks to all who have helped over the last 4 years to get us here.

Join us!

We have 34 files to be booked out to reach the end of September goal of getting to 3.45 million bits, in order to be on track for 3.6 million bits by the end of the year. join us!

A cursory inspection of (~7k) double checked 321 numbers revealed 1 unknown mismatch, but not prime :wink: I am still planning the script to scan for double checks...

Please keep up the good effort; We could find a mega-prime any day soon :unsure:

Reminders:[LIST][*]321 is dense[*]321 sequence is not erratic[*]321 LLR is the quickest[*][URL="http://www.mersenneforum.org/showthread.php?t=6874"]321 status page[/URL][*][URL="http://www.mersenneforum.org/showthread.php?t=6808"]321 reservations[/URL][*][URL="http://www.mersenneforum.org/showthread.php?t=3410"]321 statistics[/URL][/LIST]Good luck!

Thanks again to Xyzzy for uploading and hosting the new input files, which will take this project to 3.6 million bits. If we can do these by the end of 2007 that would be wonderful.

There will be about a further 100 files at the current FFT size; At 3.7 million bits I will reduce the input files to a range of 500, each containing about 29 candidates.

Although we expect the next 321 prime to be about 4 million bits, it could be found sooner. Give it a go -- you never know your luck! :wink:

We have reached 3.5 million bit candidate testing. Thanks to all participants for the help :tu: With more help, we could be the first "[URL="http://primes.utm.edu/bios/top20.php?type=project&by=ScoreNormal"]project[/URL]" to find a "[URL="http://primes.utm.edu/glossary/page.php?sort=Megaprime"]mega[/URL]-llr" prime. As a "[URL="http://primes.utm.edu/bios/top20.php?type=person&by=ScoreNorma"]person[/URL]" you could receive the credit. Join us, if you have not already.

Welcome PrimeGrid
 

We are proud to announce that PrimeGrid will be helping 321search from today.

This is how "credit" and "points" will be allocated:
[LIST=1][*]PrimeGrid will be treated as a "person" or individual with respect to [URL="http://www.mersenneforum.org/showthread.php?p=50474#post50474"]321's internal points system[/URL]
[*]If a "321 prime" is found by PrimeGrid credit will be shared between 321search and PrimeGrid at the Prime Pages database [I][URL="http://primes.utm.edu/bios/top20.php?type=project&by=ScoreNormal"]as teams[/URL][/I]
[*]If a 321 prime is found by anyone and PrimeGrid have reached at least 3rd position in 321's internal scoring system, then PrimeGrid will share Prime Pages points with 321search [I]as teams[/I].[/LIST]
Note that individual people who make a reservations on 321search's "[URL="http://www.mersenneforum.org/showthread.php?t=6808"]reservation system[/URL]" will get full credit for a prime as a "[URL="http://primes.utm.edu/bios/top20.php?type=person&by=ScoreNormal"]person[/URL]" at the Prime Pages; It has been left up to [URL="http://www.primegrid.com/"]PrimeGrid[/URL] as to how they share the Prime Pages points, if PrimeGrid finds a prime.

[quote=paulunderwood;118785]We are proud to announce that PrimeGrid will be helping 321search from today.

This is how "credit" and "points" will be allocated:[LIST=1][*]PrimeGrid will be treated as a "person" or individual with respect to [URL="http://www.mersenneforum.org/showthread.php?p=50474#post50474"]321's internal points system[/URL][*]If a "321 prime" is found by PrimeGrid credit will be shared between 321search and PrimeGrid at the Prime Pages database [I][URL="http://primes.utm.edu/bios/top20.php?type=project&by=ScoreNormal"]as teams[/URL][/I][*]If a 321 prime is found by anyone and PrimeGrid have reached at least 3rd position in 321's internal scoring system, then PrimeGrid will share Prime Pages points with 321search [I]as teams[/I].[/LIST]Note that individual people who make a reservations on 321search's "[URL="http://www.mersenneforum.org/showthread.php?t=6808"]reservation system[/URL]" will get full credit for a prime as a "[URL="http://primes.utm.edu/bios/top20.php?type=person&by=ScoreNormal"]person[/URL]" at the Prime Pages; It has been left up to [URL="http://www.primegrid.com/"]PrimeGrid[/URL] as to how they share the Prime Pages points, if PrimeGrid finds a prime.[/quote]
Cool! :cool: I already crunch for PrimeGrid, so when work is available for 321 I'll simply check the box in my preferences! (I was already hoping to do some 321 work on the one "core" [actually just a hyperthread] of my system that I reserve for non-BOINC stuff, but it's all booked out right now with other work, so this is especially great for me right now.)

BTW, is that why jmblazek reserved that humongous range yesterday? :wink:

It should be live already: [url]http://www.primegrid.com/apps.php[/url] :smile: but only Windows so far...

[QUOTE]BTW, is that why jmblazek reserved that humongous range yesterday? [/QUOTE]

:lol: I am grateful to John for negotiating between Rytis and me over the PG-321 partnership.

[quote=paulunderwood;118806]It should be live already: [URL]http://www.primegrid.com/apps.php[/URL] :smile: but only Windows so far...
[/quote]
It isn't shown on the home page in the box that lists how much work is available for each app, though...usually that means that it's still in testing. (But since the LLR app has already been developed there, it shouldn't be in testing long, I would think.)

I've already got it selected in my project preferences, though, so when it does go live I should get some work.

P.S.: Rytis said the estimated date is in 2 or 3 days:
[url]http://www.primegrid.com/forum_thread.php?id=789[/url]

Update: 321 work on PrimeGrid is now available.

[quote=Rytis (on PrimeGrid forums, link to thread is in my last post)]
Work should be flowing now. Validation is off for the moment, also no frontpage status update yet.
[/quote]

[quote=paulunderwood;118806]It should be live already: [URL]http://www.primegrid.com/apps.php[/URL] :smile: but only Windows so far...



:lol: I am grateful to John for negotiating between Rytis and me over the PG-321 partnership.[/quote]


LLR (3*2^n-1)
PlattformAktuell

Microsoft Windows (98 or later) running on an Intel x86-compatible CPU 5.08 18 Nov 2007 15:57:07 UTC

Linux running on an Intel x86-compatible CPU 5.07 24 Nov 2007 11:53:55 UTC


I am running now on Linux Ubuntu 64

Lennart

[quote=paulunderwood;96856][quote]IF you find a million digit prime soon (which I hope you do), will you still continue to LLR to 5 M? [/quote]Hardly, No. Personally, the main goal is a 1 million plus digit prime. The effort would have to be redoubled to find the following prime, most likely and that could be greater than 5M bits :cry: So it just a final push now. Help us :smile:[/quote]
I noticed this post up a ways in this thread...which prompted the question: If 321 finds a 1 million digit prime before they reach 5M, what is to be done with the rest of the sieve file? Is it going to be made available publicly in case another person or project wants to use the rest of it? Or, would all the sieving work from the point at which the project is stopped, up to 5M be wasted?

I'm thinking that even if the 321 project doesn't want to continue searching after they find a 1 million digit prime, it would be nice if maybe PrimeGrid still continued the search up to 5M--it would be a shame to waste all that sieving work. And since PrimeGrid is already planning to do tests that big in the future with their Cullen/Woodall prime searches, it's not like tests of that size will be bigger than they'd otherwise be doing.

I plan to hand over the remaining files to PrimeGrid, if and when we find the next prime. I have already discussed this John in emails between us.

PrimeGrid has two advantages. The first is they double-check results -- larger numbers are more error prone. The second advantage is people can just point their Linux and Windows boxes at the server -- Mac Intel too?

[quote=paulunderwood;119456]I plan to hand over the remaining files to PrimeGrid, if and when we find the next prime. I have already discussed this John in emails between us.

PrimeGrid has two advantages. The first is they double-check results -- larger numbers are more error prone. The second advantage is people can just point their Linux and Windows boxes at the server -- Mac Intel too?[/quote]
PrimeGrid doesn't have a MacIntel app available yet, though I imagine it will be available in the somewhat near future, considering that a MacIntel build of LLR now exists.

I thought I had better write something about how things are going for the 321search project.

We have no numbers less than 1 million digits left for first time testing.

PrimeGrid have been on board for a couple of months and they have increased the search's throughput greatly. :smile:

We have started testing at 4 million bits. The new batch here for manual testing takes much longer, nearly 50% longer, to crunch compared to the previous batch because the new ones are much longer -- PrimeGrid and James Burt are doing the intermediate ones -- and use a bigger FFT window. The effect of worse caching and bus contention must kick in at some length -- whether we are there yet I do not know.

My thanks goes to Xyzzy for uploading and hosting the new batch of input files.

Candidate primes start at n=4.35 million or 1,309,480 decimal digits. There is a new FFT size for this batch, so expect individual tests to take about 20% longer than before.

Good luck! :smile:

The plan now for 321search is to continue with PrimeGrid to n=5M. Thereafter PrimeGrid will take over. They are currently sieving to n=10M. I guess there are 2 or 3 more primes below that.

There is a chance we will find another megaprime below n=5M -- I am hoping for some bunching now in the sequence of 321 primes :wink:

With PrimeGrid's help we should be able to finish to n=5M within 6 months :whistle:

PrimeGrid have been running "[URL="http://www.primegrid.com/forum_thread.php?id=928"]Drive for Five[/URL]" for the last week. It will be running until the end of this month. Even if they fail to reach n=5M by then, the time to completing our task will have been greatly shortened -- anything remaining will be mopped up within a month or two :w00t:

Maybe we will have a bonus prime too :wink:

PrimeGrid and many of its members have done a great work to get to 5M. Although there are hundreds of results to come in, they have exhausted their 321 input :surprised

James Burt has a few outstanding ranges below n=4M. Many of them will be put on the PG server soon. I am still hoping all n=sub-5M will be done by the end of June, but my guess is that there will be a few left over to do then.

Unfortunately, no prime has showed up yet :mellow:

:smile: I happy now with the coverage of 321 numbers below n=5M. Before this project is achived, I will create a new thread with all the results, including:[LIST][*]factors[*]llr residues[*]primes[*]double checks[*]errors[/LIST]

A change of plan: I have handed all the 321 factors and LLR results over to PrimeGrid. John says they hope to make a public database in the future.

I would like to thank the following for their help with "321search" over the last few years:

* Thomas Ritschel for his major contributions with sieving, crunching LLR, scripts, encouragement, discussion and ideas.
* The many loyal sievers and LLR crunchers who are listed on this subforum and at [URL="http://primes.utm.edu/bios/page.php?id=479"]The Prime Pages[/URL]
* PrimeGrid users and its organizers for their huge boost to throughput towards the n=5M goal
* Programmers: Jean and George for LLR; Paul Jobling and Geoff for their sieves; Phil Carmody for some Perl wizardry
* Xyzzy for allowing and managing this great message board

Anyone wanting to do some "321" crunching can head over to [URL="http://www.primegrid.com/"]PrimeGrid[/URL], where their help will be welcomed. :tu: