Year of the sheep

A new year, a new series of challenges! As we're not just a bunch of sheep here at PrimeGrid we'll be trying a new concept for the first challenge of this series. From January 3rd 2015, 18:00 UTC until January 11th 2015, 18:00 we will be hosting the first Multi Project Challenge!

What is a Multi Project Challenge?
In this challenge units from multiple subprojects will count for your score on the leaderboard. In this case the included subprojects are: Seventeen-or-Bust (LLR), Prime Sierpinski Problem (LLR) and Extended Sierpinski Problem (LLR). Application builds are available for Linux, MacIntel, and Windows 32 and 64 bit.

NOTE: In your PrimeGrid preferences section, set "Send work from any subproject if selected projects have no work" to no to guarantee that no other work will be sent.


ATTENTION: The primality program LLR is CPU intensive; so, it is vital to have a stable system with good cooling. It does not tolerate "even the slightest of errors." Please see this post for more details on how you can "stress test" your computer. If your computer is highly overclocked, please consider "stress testing" it before the challenge. :)

The length of the WU's will vary wildly, based on your system and the subproject you select.
For the Extended Sierpinski Problem tasks will take about 6-8 hours on the fasted systems and 24+ hours on the slower systems.
For the Prime Sierpinski Problem wu's will take 45-55 hours on fast/newer computers and 150+ hours on slower/older computers.
For Seventeen-or-Bust, the fastest computers will take about 4 days, where as older/slower computers won't be able to finish a single set of WU's during the challenge.

Restricted airflow is one of the primary reasons for overheating. Take the time to monitor the fans and review the dust buildup. Please, please, please make sure your machines are up to the task.

Why a Multi Project Challenge?
There's two reasons to take this approach. For the full discussion see this thread, the short version is: this way also the subprojects with long work units can get a challenge and it's an elegant way to solve the "how to give a reasonable bonus"-problem we encountered last year with the dual GFN challenge.

Time zone converter:

The World Clock - Time Zone Converter

NOTE: The countdown clock on the front page uses the host computer time. Therefore, if your computer time is off, so will the countdown clock. For precise timing, use the UTC Time in the data section to the left of the countdown clock.

Scoring information
Scores will be kept for individuals and teams. Only work units issued AFTER January 3 2015 18:00 UTC and received BEFORE January 11 2015 18:00 UTC will be considered for credit.

In this challenge we will be using the same scoring algorithm as we use to determine BOINC credits, including bonuses!. Because the score bonuses are included in the challenge score, it is advantageous to run the longer tasks if you can manage to finish them within the 8 day challenge window. This will land you a higher position in the leaderboard and therefor a higher score on the overal leaderboard.

At the Conclusion of the Challenge
We would prefer users "moving on" to finish those tasks they have downloaded, if not then please ABORT the WU's instead of DETACHING, RESETTING, or PAUSING.

ABORTING WU's allows them to be recycled immediately; thus a much faster "clean up" to the end of an LLR Challenge. DETACHING, RESETTING, and PAUSING WU's causes them to remain in limbo until they EXPIRE. Therefore, we must wait until WU's expire to send them out to be completed.

Please consider either completing what's in the queue or ABORTING them. Thank you. :)

About the projects
The common denominator of these projects is the Sierpinski Problem.

Waclaw Franciszek Sierpinski (14 March 1882 - 21 October 1969), a Polish mathematician, was known for outstanding contributions to set theory, number theory, theory of functions and topology. It is in number theory where we find the Sierpinski problem.

Basically, the Sierpinski problem is "What is the smallest Sierpinski number" and the prime Sierpinski problem is "What is the smallest 'prime' Sierpinski number?"

First we look at Proth numbers (named after the French mathematician François Proth). A Proth number is a number of the form k*2^n+1 where k is odd, n is a positive integer, and 2^n>k.

A Sierpinski number is an odd k such that the Proth number k*2^n+1 is not prime for all n. For example, 3 is not a Sierpinski number because n=2 produces a prime number (3*2^2+1=13). In 1962, John Selfridge proved that 78,557 is a Sierpinski number...meaning he showed that for all n, 78557*2^n+1 was not prime.

Most number theorists believe that 78,557 is the smallest Sierpinski number, but it hasn't yet been proven. In order to prove it, it has to be shown that every single k less than 78,557 is not a Sierpinski number, and to do that, some n must be found that makes k*2^n+1 prime.

The smallest proven 'prime' Sierpinski number is 271,129. In order to prove it, it has to be shown that every single 'prime' k less than 271,129 is not a Sierpinski number, and to do that, some n must be found that makes k*2^n+1 prime.

Seventeen or Bust is working on the Sierpinski problem and the Prime Sierpinski Project is working on the 'prime' Sierpinski problem. The following k's remain for each project:

• Waclaw Franciszek Sierpinski (WIKI)
  
• Sierpinski number / The Sierpinski problem (WIKI)
  
• The Sierpinski problem (Prothsearch)
  

What is LLR?

The Lucas-Lehmer-Riesel (LLR) test is a primality test for numbers of the form N = k*2^n − 1, with 2^n > k. Also, LLR is a program developed by Jean Penné that can run the LLR-tests. It includes the Proth test to perform +1 tests and PRP to test non base 2 numbers. See also:

• Lucas-Lehmer-Riesel test (WIKI)
  
• Download LLR by Jean Penné
  

(Edouard Lucas: 1842-1891, Derrick H. Lehmer: 1905-1991, Hans Riesel: born 1929).
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

We note that an astronomical new moon will occur at 2015, February 19, 07:47, Chinese time (that is February 18, 23:47, UTC). Therefore the New Year, the start of year of the Goat (or Sheep), is February 19. /JeppeSN

With both SoB and PSP, there is a possibility that our existing supply of tasks may be exhausted. I don't think we'll actually crunch that many, but during challenges a lot of tasks get downloaded but not crunched. It's possible enough tasks will be downloaded to consume the existing pool of untested candidates.

We're in the process of preparing additional tasks -- even if they're not needed during the challenge, it's likely they'll be needed later in the year. For SoB, the current tasks are in the n=27M range. The next batch of test are n=31M. The new tasks take about 30% longer to crunch than the existing tasks.

The range for the next batch of PSP tasks is still being determined.

ESP is not affected. Unlike SoB and PSP, ESP is purely a PrimeGrid project which doesn't need to be coordinated with external projects. We're the only ones crunching ESP and we'll keep crunching upwards without gaps.
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Looks like I got finally 12 SOB wu's. Now we shall see how long it will take to finish them. The forecast seems to be for 84 hours each.

It is trying to run 12 and thus PPS Sieve wu's are taking over 9 minutes each and they used to be about 8 minutes a few seconds each. Perhaps I should abort 6 of the SOB's?

After 32 minutes of running 12 SOB's if the early average holds up it would take 283 hours each. Too long for the contest.

SOB has now been changed to a 45 day deadline, which affects new workunits generated from this point on. There will still be some 60 day deadlines sent out over the next few days as the already generated tasks are sent out, and any additional tasks sent out for older units will also have the older 60 day deadline.

Even if we load the larger n=31M SOB tasks before the challenge ends, they'll also have a 45 day deadline towards the goal of having a shorter cleanup. Sometime after the challenge ends we will consider whether 45 or 60 days is more appropriate for the larger tasks on a permanent basis.
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

So, we're hunting prime sheep next year? I found one for you:



Merry Christmas! ;)
____________

Seems to be that running 6 SOB tasks at once after 41:29 hours I am at 26.155% done. This means my average expected finish will be taking 158 hours. Will finish in time for the Challenge start off.

Seems to be that running 6 SOB tasks at once after 41:29 hours I am at 26.155% done. This means my average expected finish will be taking 158 hours. Will finish in time for the Challenge start off.

Most likely not a good idea to run 6 SOB tasks at once... Due to the huge exponent for those tasks, you are likely creating memory bandwidth bottleneck.

I will be crunching 1 SOB and 2 ESP candidates at one time on my i5... leaving a core open.

Neo
AtP

It sure is running a lot longer than I thought originally. But as of this morning and almost 3 days after the start I have reached over 44% done, it is still predicting that it will get done on about 30 hours, but I think on my average speed so far, it will be 96 hours to go.

It sure is running a lot longer than I thought originally. But as of this morning and almost 3 days after the start I have reached over 44% done, it is still predicting that it will get done on about 30 hours, but I think on my average speed so far, it will be 96 hours to go.

Thank you Neo. Looks like it is crunching about 16-17% a day, with some cpu's running slightly faster than others. I wonder if before the start of the challenge I also have to empty my PPS Sieve queue, although that will not be part of the challenge this time?

What to expect regarding task lengths.

My faster computer is one of the faster machines right now, although definitely not the fastest. It's an i5 Haswell running at 3.6 GHz with dual channel overclocked 2400 MHz DDR3 memory. I'm running Windows and have a fair amount of junk running in the background, and I also use this as my normal use computer, so it's not crunching at 100%.

The challenge runs for 192 hours.

The current SoB tasks take about 105 hours on this computer. I'll be able to get one set of four done during the challenge.

The current PSP tasks take about 40.5 hours. After the SoB tasks I expect to be able to complete 2 sets of PSP tasks.

ESP takes about 6.25 hours. I might, maybe, if I'm lucky, be able to squeeze in one set of ESP tasks before the challenge ends.

My slow computer will be able to crunch one ESP task during the entire challenge.

Expect a moderate increase in run times if you have 1600 or 1333 MHz DDR3. Also expect a moderate increase if you have AVX rather than FMA3. Expect a substantial increase without AVX (which includes all AMD processors.)

Quad channel memory will improve performance, and I suspect DDR4 will improve performance.

NOTE: It's possible, although not certain, that both SoB and PSP may move to a larger range during the challenge. If so, SoB's tasks will increase in duration by about a third. It's not yet been determined what will happen with PSP.
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

I tried to run a few wu's on my middle faster computer and the ESP-Sieve 44 took about 45 minutes each, per cpu.

Michael (or any of the mods!), will you be able to warn us when the SoB WUs are going to change to the n=31m batch? I just mean by a post on here, of course.

Since there are three subprojects listed for the next challenge, how can I designate the one I would like to work on, since my slower computer may not be able to finish in time two of the longer projects?

Michael (or any of the mods!), will you be able to warn us when the SoB WUs are going to change to the n=31m batch? I just mean by a post on here, of course.

Miklos,
From the BOINC GUI manager, you can assign different "venues" to your different computers, allowing them to run different subprojects. From BOINC, hit the "Your Computers" button. That will open a web browser page with a row for each computer you have. Toward the left side of each row, there will be a "Details" link to click on. Click it, and the next page, near the bottom, will have a "Location" menu, having choices for "default" ("---") home, work, school, or the various planets. Pick one for each computer, and hit "Update" next to the menu. You can make each computer the same or different.

Then, back on the normal subproject selection page, you can select different (or the same!) subprojects for "home", "work", "school", et cetera, as you see fit.

For this challenge I'll be running SoB on my fast boxes, then maybe PSP. My slower boxes I'll have on ESP the whole time.

Good luck!
--Gary

Just a thought - where will a prime found in the n=31m range of SoB rank in the top 10? Obviously I'm aware that it will change over time, but just say the first candidate (or whatever example is easy for you to use) turns out to be prime.

Looks like the ESP LLR takes about 9.5 hours on my fastest computer, when I do 6 at a time. I wonder if running 12 at once would take twice as long each.

Here's some times I've done over the last few days. The 3570K @ 3.7Ghz is a recent ASUS MB (P8Z77-V LK) I got used and for some reason the CPU multiplier is missing from the BIOS even though it's a K CPU. It also will not boot with memory above 1600. It's the EXACT same ASUS moodel and BIOS as my other 2 Intel MBs. I'm still researching possible causes. There are other people with the same problem with this model as well as other MBs.
But it does let me show closer to stock times.


Intel Sandy Bridge 2600K @ 4.4Ghz (4 cores running) with 1866 memory 9-10-9-24 timings:

SoB: ~125.5 hrs
PSP: ~56 hrs
ESP: ~7.5 hrs

Intel Ivy Bridge 3570K @ 3.7Ghz with 1600 memory 8-8-8-24 timings:

SoB: ~145 hrs
PSP: ~59 hrs
ESP: ~8.5 hrs

Intel Ivy Bridge 3570K @ 4.4Ghz with 2400 memory 10-12-12-31 timings:

SoB: ~118 hrs
PSP: ~50 hrs
ESP: ~7 hrs

AMD Phenom II 970BE @ 3.7Ghz with 1333 memory 7-7-7-21 timings:

PSP: ~124 hrs
ESP: ~18.5 hrs
____________
"Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety."

-Benjamin Franklin

With both SoB and PSP, there is a possibility that our existing supply of tasks may be exhausted. I don't think we'll actually crunch that many, but during challenges a lot of tasks get downloaded but not crunched. It's possible enough tasks will be downloaded to consume the existing pool of untested candidates.

We're in the process of preparing additional tasks -- even if they're not needed during the challenge, it's likely they'll be needed later in the year. For SoB, the current tasks are in the n=27M range. The next batch of test are n=31M. The new tasks take about 30% longer to crunch than the existing tasks.

The range for the next batch of PSP tasks is still being determined.

ESP is not affected. Unlike SoB and PSP, ESP is purely a PrimeGrid project which doesn't need to be coordinated with external projects. We're the only ones crunching ESP and we'll keep crunching upwards without gaps.

594665184 416387957 29 Dec 2014 | 13:04:17 UTC 1 Jan 2015 | 0:02:47 UTC Completed, marked as invalid 98,767.78 92,065.61 0.00 Extended Sierpinski Problem v6.24
Too bad after so many hours.

594665184 416387957 29 Dec 2014 | 13:04:17 UTC 1 Jan 2015 | 0:02:47 UTC Completed, marked as invalid 98,767.78 92,065.61 0.00 Extended Sierpinski Problem v6.24
Too bad after so many hours.

This scheme is clearly feasible... I even thought about it myself a while back. But, I thought of a few downsides:

- With suspended tasks no other PG work will d/l. This could be a fuss if you wanted to keep using a GPU.
- I believe with LLR, when you suspend a task, you will lose some progress, as it will drop back to the last checkpoint when resumed.
- You have to be able to be at keyboard to do the suspend/resumes at the right time. Maybe not a problem for some, but for me, yeah.
- In the (unlikely) event that a prime is found, if it was one that you had suspended, you're a lot more likely to be double-checker rather than initial finder.

Just some more food for thought.

--Gary

This scheme is clearly feasible... I even thought about it myself a while back. But, I thought of a few downsides:

- With suspended tasks no other PG work will d/l. This could be a fuss if you wanted to keep using a GPU.

- I believe with LLR, when you suspend a task, you will lose some progress, as it will drop back to the last checkpoint when resumed.

- You have to be able to be at keyboard to do the suspend/resumes at the right time. Maybe not a problem for some, but for me, yeah.

- In the (unlikely) event that a prime is found, if it was one that you had suspended, you're a lot more likely to be double-checker rather than initial finder.

I do this in a lot of challenges and it does work. It is a lot of micro managing. Thinking things through is the key. One minor drawback is you don't show up in the challenge scores until near the end.


____________
"Those who would give up Essential Liberty to purchase a little Temporary Safety, deserve neither Liberty nor Safety."

-Benjamin Franklin

Besides, the odds of one of your 6 or 7 tasks being prime is rather small.

How small? What are the (8 day challenge) odds for finding a SOB/PSP/ESP prime(s) in their respective ranges?
Looking at the top 100 prime list: quite a few are first timers without any other (non-Mega) finds. I'm searching for my first prime.

The challenge has started! This is our first ever SoB challenge, so it will be interesting to see how many SoB tasks get done.

This is how many tasks were sent out during each minute in the first 10 minutes of the challenge:

I wonder if I should be running 6 only or 12 wu's at once. It sure is less time consuming as far as attention to details to run the 12 at once and not keep suspending a whole bunch except the 6 running. Any suggestions, please?

Damn you day light savingstime, missed the start bij half an hour! :( Lets hope it doesn't matter too much in the end :)
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

I'm waiting until after 19:00 to return queue sizes back to normal in case there's another surge of daylight savings time computers. :)
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Based on the first hour's average, it looks like running 12(since I could not get to run 6 only it seems) it will take 16-21 hours for the 12 to get done.

That projected time sounds somewhat optimistic.

Yes, only ESP LLR. But perhaps when the next set starts it will switch to running six only. I would gladly settle for these numbers for the time, though.

Oops. Sorry, guys [sheepish grin] - I assumed SoB.

I'm not sure if it's been mentioned lately, but the longer tasks (PSP, and even more so, SoB) have credit bonus and produce more credit per hour than do ESP tasks. That also means you get more challenge points per hour.

Of course, you do actually have to be able to finish the tasks and return them before the challenge ends, which may not be possible for some slower computers with the longer tasks. So choose wisely!

____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

I started with four PSP tasks on my Q9400, and I was amazed to see they didn't complete 1% in 2 hours! That meant they would take over 200 hours each! I aborted two and am trying the other two opposite ESP to see if memory bandwidth is improved.

Edit: Nope, if anything memory bandwidth got worse.
____________

I started with four PSP tasks on my Q9400, and I was amazed to see they didn't complete 1% in 2 hours! That meant they would take over 200 hours each! I aborted two and am trying the other two opposite ESP to see if memory bandwidth is improved.

With just under three hours into the challenge, the server has sent out:

1258 SoB tasks
1420 PSP tasks
7420 ESP tasks

The first of those ESP results should be coming back in another 3 hours.
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Damn you day light savingstime, missed the start bij half an hour! :( Lets hope it doesn't matter too much in the end :)

Day light savingstime? In winter? :o:

That's a new trick! You gotta get a patent on it! Lol! :D

Looking at the database, it looks like it's possible to run ESP in a bit over 4 hours. I'm not sure how (and it may be a recording error). We might start seeing results sooner than I expected.
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Michael (or any of the mods!), will you be able to warn us when the SoB WUs are going to change to the n=31m batch? I just mean by a post on here, of course.

We will try. It's conceivable that a surge in task downloads may cause the transition to happen when we're not expecting it, but we'll do the best we can.

Yes, I expect a lot will change in the standings once the first PSPs and SoBs will start coming in. SoB's will take another 3 days or approximately, PSP could be a bit sooner.
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)

After 1 day:

1690 SoB tasks sent out
2168 PSP tasks sent out
12161 ESP tasks sent out and 896 returned

This is a rare challenge where the task deadlines (for ESP) are shorter than the challenge duration. All of the ESP tasks sent out during the first two days of the challenge will either be returned or time out before the challenge ends.

____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

According to data from my systems after first day, it seems that standard or moderately overclocked Haswell CPUs will require about 100 hours to finish one set of SoB tasks. I.e. if they try to run second set, they will miss the deadline of the challenge by very small amount (4-8 hours). So be aware.

My PSP tasks are going quite well, headed to be completed at the 45-50 hour mark. fastest task is 58.5% complete after 26 hours and 20 min into the task. The slower one is at 53.5% complete after the same time as the first one. i5-2500k @ 4.5GHz
____________

275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew

After almost 2 days...

Challenge update:

SoB: 2010 tasks sent
PSP: 2805 tasks sent, 2 tasks returned
ESP: 17288 tasks sent, 1173 tasks returned

SoB, PSP, and ESP are quite different in size, so merely looking at the number of tasks doesn't tell you which of the three is more popular. If we handicap the raw statistics by multiplying the number of tasks by the relative run-times, we get this:

SoB: 32803
PSP: 17952
ESP: 17288

Not surprisingly, SoB seems to be the project of choice, at least for those computers fast enough to complete them in 8 days.

Happy crunching, and good luck to you all!
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

The first PSP task to be returned was crunched on an i7-4790 in 141,315 seconds (or 39:15:15).

The second PSP was crunched on an i5-2500K in 160,746 seconds (44:39:6).
____________
My lucky number is 75898^524288+1
Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

The second PSP was crunched on an i5-2500K in 160,746 seconds (44:39:6).

After almost 2 days...

Challenge update:

SoB: 2010 tasks sent
PSP: 2805 tasks sent, 2 tasks returned
ESP: 17288 tasks sent, 1173 tasks returned

SoB, PSP, and ESP are quite different in size, so merely looking at the number of tasks doesn't tell you which of the three is more popular. If we handicap the raw statistics by multiplying the number of tasks by the relative run-times, we get this:

SoB: 32803
PSP: 17952
ESP: 17288

Not surprisingly, SoB seems to be the project of choice, at least for those computers fast enough to complete them in 8 days.

Happy crunching, and good luck to you all!