Author |
Message |
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
We have entered the "optimal sieve zone". Actually, we've been floating around it for some time since the last prime was found back in June. Please consider this the first notice to anyone on the cusp of their next badge level (20K, 200K, 1M, 2.5M).
It is currently unknown "exactly" when the sieve will end. However, we can guarantee another four weeks (REVISED sooner: to at least 13 Oct 2010). In the meantime, we'll be investigating to pinpoint an exact stop date. And yes, we are taking the Calendula Challenge into consideration so people won't be forced to make a choice between the two. :)
This is a major accomplishment for both the Prime Sierpinski Problem and the Seventeen or Bust projects. The PSP/SoB sieve at PrimeGrid was announced on 13 Oct 07. That date this year might be a very fitting day to end the sieve. :)
The official classification of this will be a suspension and not a termination. Should "improved" sieving methods come up in the future, this sieve will be reviewed for continuation. EDIT: And of course, should there still be k's remaining as we approach 50M, we'll need to resume sieving for n>50M. :)
We'll update everyone as soon as we know more.
____________
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13524 ID: 53948 Credit: 244,649,750 RAC: 315,520
                          
|
It is currently unknown "exactly" when the sieve will end. However, we can guarantee another four weeks. In the meantime, we'll be investigating to pinpoint an exact stop date. And yes, we are taking the Calendula Challenge into consideration so people won't be forced to make a choice between the two. :)
This is a major accomplishment for both the Prime Sierpinski Problem and the Seventeen or Bust projects. The PSP/SoB sieve at PrimeGrid was announced on 13 Oct 07. That date this year might be a very fitting day to end the sieve. :)
Color me confused. Today is the 18th of September, you're saying there's at least 4 weeks to go (that takes us to the 6th of October), and you're saying (I think) that you're extending that by 13 days to allow for the Calendula challenge. That takes the earliest possible end date to October 29th.
Then the next paragraph says it might end on the 13th.
So I'm confused.
____________
My lucky number is 75898524288+1 |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Color me confused. Today is the 18th of September, you're saying there's at least 4 weeks to go (that takes us to the 6th of October), and you're saying (I think) that you're extending that by 13 days to allow for the Calendula challenge. That takes the earliest possible end date to October 29th.
Then the next paragraph says it might end on the 13th.
So I'm confused.
Sorry for the confusion. The 13th of October was an after thought once I discovered that the project was initially started on that date. As for the Challenge, I did not intend to imply an additional 13 days to be added.
By suggesting a symbolic end date, I should have revised the initial guarantee. The guarantee is at least until the 13th of October, 2010.
____________
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13524 ID: 53948 Credit: 244,649,750 RAC: 315,520
                          
|
By suggesting a symbolic end date, I should have revised the initial guarantee. The guarantee is at least until the 13th of October, 2010.
Ah, that's unfortunate. I was hoping the 29th was the date. I've got about 25 days worth of crunching left to get a silver badge in the PSP/SoB sieve, which means if I drop everything else and run only PSP sieve, I have just enough time to get it done.
So much for the PPS LLR call-to-arms and the Calendula challenge. :(
The ironic thing is the PSP sieve is what I had been crunching until the warning about the 321 sieving coming to and end (albeit in about 6 months time.) If I hadn't switched to 321, I would have had the silver in PSP already. Oh well.
____________
My lucky number is 75898524288+1 |
|
|
Scott Brown Volunteer moderator Project administrator Volunteer tester Project scientist
 Send message
Joined: 17 Oct 05 Posts: 2180 ID: 1178 Credit: 9,069,802,017 RAC: 13,168,358
                                      
|
I am wondering about how "optimal" is determined across all sieves. I assume that it is generally the same method...namely that the rate of factors being found falls below the rate of computational time to find actual primes with LLR, etc.
However, this leaves me with a question...if you will permit me the use of an old cliche in analogy...
Let us assume that we are dealing with the old "...forest for the trees" analogy where each "tree" is a number that is or is not prime. The prime "trees" are much rarer than non-prime "trees", and we have two methods for observing this difference. One method is to look directly at the primeness of each "tree"; in the analogy perhaps by walking to each tree and counting the branches. The second way is to observe the non-primeness of the tree in some manner (i.e., the sieve process); for the analogy, let's say by observing some characteristic of the leaves--say their color. The way that I understand the "optimal" decision is being made would be, in the analogy, to see when it takes the same amount of time to count the branches as to discern the leaf color of each "tree"?
If this is correct, then I wonder if it is really optimal for all sieve conditions. For example, the time of calculation of each workunit in the TRP LLR and sieve applications is similar, essentially like walking to each "tree", and thus the time to assess each "tree" by each method can be compared directly for efficiency. However, in the case of the PSP/SoB sieve vs. PSP LLR and SoB LLR, these times are drastically different (about 35minutes on average for the sieve vs. 90 hours and 286 hours for the LLR's. respectively). In the analogy, this would seem to be like the difference of walking to each "tree" to see if it is prime vs. flying over the "forest" in a plane and observing whole sections that are not prime. More directly said, in BOINC, I believe that it is fair to say that individuals are more likely to chose to crunch smaller/shorter work (at least with a preference for no more than a 1-day/24-hour workunit). Thus, even if the actual "tree by tree" rate of sieving is no faster than the LLR approach, isn't there an efficiency of more individuals being willing to sieve shorter work (essentially eliminating whole sections of the "forest")? Or has this already been factored into determining what is "optimal"?
Edit: or said more mathematically, let's say that you have 100 crunchers. All 100 are willing to crunch (and have the equipment to handle) the short sieves. If all 100 are also willing and capable of crunching the much longer LLR, then the direct application/computational efficiency can be directly compared. However, if only 25 are willing and/or have the capability to LLR, then wouldn't the optimal sieve depth need to be 4 times the actual application/computational efficiency?
____________
141941*2^4299438-1 is prime!
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13524 ID: 53948 Credit: 244,649,750 RAC: 315,520
                          
|
Conversely, however, people do want to find prime numbers, and sieves never will. That makes the LLRs more desirable. Finding one of these primes would be a big, big event. So there's also some pressure pushing crunchers towards the longer WUs.
If I wasn't playing "find the badge,", I would most likely be running some combination of SoB, PSP, and ESP LLRs.
____________
My lucky number is 75898524288+1 |
|
|
|
It's a shame, another project where i don't earn satisfaction badge. It's becoming just irritating.
____________
Polish National Team |
|
|
mackerel Volunteer tester
 Send message
Joined: 2 Oct 08 Posts: 2468 ID: 29980 Credit: 449,499,878 RAC: 300,908
                           
|
toms83, looking at your computer list it would only take some days to get a bronze badge if you want this one. |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
isn't there an efficiency of more individuals being willing to sieve shorter work (essentially eliminating whole sections of the "forest")? Or has this already been factored into determining what is "optimal"?
In general, we are comparing the average time to find a factor vs the average time to do an LLR test. Even if we were able to break the LLR test into smaller WU's equivalent to the sieve WU, we are still entering the "optimal sieve zone." Same thing applies even if there are more people willing to sieve. For anyone "not willing and/or have the capability to LLR", we encourage participation in the other sieves (321, Cullen/Woodall, PPS, TRP) where their computing power can be more effectively applied.
If PG had only one sieve, then your scenario would play a larger role...unless of course, the sieve WU time was increased to match the duration of an LLR WU. ;)
____________
|
|
|
|
Well I've only got 2 (64-bit) cores @2.9GHz, and am barely over the bronze badge threshold. At about 24 minutes per WU, it'll take me at least 45 days 24/7 to get to 200K. Realistically that means more like 90 days. >>punt<<
I certainly support shutting things down if the science warrants it; I just wish I had poured some more time in to this earlier. Oh well.
--Gary
|
|
|
|
toms83, looking at your computer list it would only take some days to get a bronze badge if you want this one.
I'll fight for silver, :) but I wanted all badges in ruby. :P
____________
Polish National Team |
|
|
mackerel Volunteer tester
 Send message
Joined: 2 Oct 08 Posts: 2468 ID: 29980 Credit: 449,499,878 RAC: 300,908
                           
|
Fair enough, I'm doing same actually, making a quick push for silver before resuming LLR elsewhere... but there's already closed projects where new joiners can't get a badge at all. We can only crunch what's there. |
|
|
|
Fair enough, I'm doing same actually, making a quick push for silver before resuming LLR elsewhere... but there's already closed projects where new joiners can't get a badge at all. We can only crunch what's there.
Unfortunately yes.
____________
Polish National Team |
|
|
|
I wonder if John's initial guarantee for 13th of October 2010 took under consideration that "quick push" by so many of us...
Two quick questions:
1.During The Calendula Challenge PSP sieving project will be available or not?
2.What other subproject could face suspension or termination in upcoming months ?
____________
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13524 ID: 53948 Credit: 244,649,750 RAC: 315,520
                          
|
I wonder if John's initial guarantee for 13th of October 2010 took under consideration that "quick push" by so many of us...
Two quick questions:
1.During The Calendula Challenge PSP sieving project will be available or not?
2.What other subproject could face suspension or termination in upcoming months ?
The first question is easy: Yes. You can crunch whatever you want. Nothing should be shut down on the BOINC side. They've shut down the PRPNet side to reduce server loads sometimes, but I'd be very surprised if that happened this time. I'd expect server loading to drop during this challenge since everyone will be running longer WUs than the average.
As for the second question, the 321 Sieve also is running up to a stopping point -- although the last I heard about it there's still a few months to go before it shuts down. See this thread for details.
Additionally, John answered your specific question here when the AP26 was found.
Edit: Considering the slow pace of finding the remaining SoB, PSP, ESP, and Riesel primes, although in theory they could all end today, we probably have years to go on all of those. 321 Sieve is the only predictable project that they've said will be ending in the near term.
____________
My lucky number is 75898524288+1 |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
13 October is approaching
This will be the last insertion date for PSP (Sieve) before it is suspended. Until then, the queue will be kept full for anyone wishing to complete their last effort on this project.
____________
|
|
|
|
Are you considering to use the new version of LLR, and take advantage of the 33% speedup to further increase the progress of this project?
KEP |
|
|
|
Are you considering to use the new version of LLR, and take advantage of the 33% speedup to further increase the progress of this project?
KEP
Yes. When this is fixed
http://www.mersenneforum.org/showpost.php?p=229781&postcount=2
Lennart |
|
|
|
Oh, I actually saw that thread yesterday, but as with all the other bugs with the new GWnum libraries, I assumed that these bugs also only affected non-base2 bases. Anyway, I think that you can actually get very far next year, even this year, when these bugs are fixed.
Take care
KEP |
|
|
|
SILVER is mine. :)
____________
Polish National Team |
|
|
|
It'll be a tight timeline for me to get silver. My estimates put the 200,000 credit achievement at about 12 hours into October 13th. Hope the reservoir can stay filled that long... |
|
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 11,549,356 RAC: 43
                    
|
It'll be a tight timeline for me to get silver. My estimates put the 200,000 credit achievement at about 12 hours into October 13th. Hope the reservoir can stay filled that long...
Have you set your cache to 10 days of work? If you are only doing PG/PSP, that might allow you to grab the extra amount you need. Reporting after the Oct 13 time will probably be allowed.
____________
Murphy (AtP)
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13524 ID: 53948 Credit: 244,649,750 RAC: 315,520
                          
|
It'll be a tight timeline for me to get silver. My estimates put the 200,000 credit achievement at about 12 hours into October 13th. Hope the reservoir can stay filled that long...
Just make sure you set your cache to buffer enough work units past the 12th so that you'll make it. If you're only half a day short there should be no problem since you can easily set BOINC to cache up to 10 days worth of work.
____________
My lucky number is 75898524288+1 |
|
|
|
It'll be a tight timeline for me to get silver. My estimates put the 200,000 credit achievement at about 12 hours into October 13th. Hope the reservoir can stay filled that long...
Just make sure you set your cache to buffer enough work units past the 12th so that you'll make it. If you're only half a day short there should be no problem since you can easily set BOINC to cache up to 10 days worth of work.
10 day ? all of my boxes are are set to a max 300 WU / box
2 core box @ 3.00 Ghz - 300 WU's
4 core box @ 2.83 Ghz- 300 WU's
8 core box @ 2.67Ghz - 300 WU's
16 core box @ 2.3 Ghz - 300 WU's
What about the 10 days worth - Is it per core or just 300 per box?
It is about 1 days worth not 10 I would like a gold here and 10 days worth would just make it
Steve Martin
____________
From the High Desert in New Mexico
|
|
|
Michael Goetz Volunteer moderator Project administrator
 Send message
Joined: 21 Jan 10 Posts: 13524 ID: 53948 Credit: 244,649,750 RAC: 315,520
                          
|
Hmmmm.
Well, you can set a 10 day cache on your computer, but, yes, the server can be set to limit the number of WUs, so depending on your machine you may not be able to download 10 days worth without some fudging around.
The question, however, had to do with missing by about 12 hours, so 300 WUs should be enough unless you're crunching more than 25 WUs/hr on your machine(s).
If you want to fudge it and get more (John, don't look!), you can set up virtual machines on your mutli-crore crunchers. They'll appear as separate computers to the server, so the VM and the host machine can both get 300 WUs. You might even be able to run multiple VMs at once, but I've never tried that. In theory, you could run one VM for each core, each with 300 WUs. You'll need about 500 meg of ram per VM, however.
____________
My lucky number is 75898524288+1 |
|
|
|
Thanks all for the buffer suggestion. I completely forgot about it. I'm so used to crunching PPS with a tiny buffer. All I needed was a 1 day stash of units. It's 8.30a here and I have 400 more credits to go, but it'll be clear sailing. I was overly cautious, though: there are still well over 100,000 units to draw from if I somehow fell short today. |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
A few hours left
This will be the last insertion for PSP (Sieve) before it is suspended. Increase your cache if you're not going to make it. However, at the rate of the pass couple of days, there will only be 2-3 days of work left after this last insertion.
Congratulations to everyone who has participated in the sieve. It's a wonderful milestone to reach.
____________
|
|
|
|
Aw fudge pockets. I only have 14 days until my Barney badge.
My fault, I didn't get to it n time.
Oh well, off to SGS.
____________
|
|
|
|
Wrong thread... sorry |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
PSP Sieve Update
The buffer is quickly heading towards 0. Currently, there are less than 90K tasks remaining. Get'em while you can. :)
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
PSP Sieve Update
Currently, there are less than 20K tasks remaining. Less than 24 hours of work at our current rate.
EDIT: And now as of 18 Oct 2010 20:26:12 UTC there are none! :) Available: PSP/SoB (Sieve) 0
All that's left now are the resends. Currently there are 32681 tasks in progress.
____________
|
|
|
|
Anyone wanting to dump your PSP/SoB Sieve tasks, please feel free. I will clean up any and all WUs left.
____________
|
|
|
|
Thanks for the encouragement from Tim and for whomever put those last few units back in the queue. I was able to get my badge last night. It was close but fun watching the race between points and units yesterday.
Thanks Rick
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
PSP Sieve Update
As of 18 Oct 2010 20:26:12 UTC there are 0 tasks remaining in the work buffer! :) Available: PSP/SoB (Sieve) 0
All that's left now are the resends. Currently there are 518 tasks in progress.
History: work in progress
29 Oct: 518 tasks
28 Oct: 1028 tasks
27 Oct: 2415 tasks
24 Oct: 3317 tasks
23 Oct: 6419 tasks
22 Oct: 7546 tasks
21 Oct: 12604 tasks
20 Oct: 13396 tasks
19 Oct: 15262 tasks
18 Oct: 32681 tasks
____________
|
|
|
|
Well, it doesn't look like I will earn my Barney badge. With only 7500 WUs left, I will need 3620 to make 2 million points. A little more than half ALL outstanding WUs.
I want to thank all the guys who aborted WUs so I could get more.
I will continue to run PSP Sieve until all WUs have been completed.
EDIT: I had calculated the credits per WU by using my own numbers. John has corrected me and has given a number double what I had. This makes my calcs erroneous by half.
This is great news to me, it gives me much more of a chance to get my Barney badge.
I still want to thank everyone who gave up any WUs whether to help or just get rid of a near dead project.
____________
|
|
|
|
It looks like it's over for me. My quad has run dry and all my remaining work will only add up to 13k points. It looks like I will fall 9k short.
____________
|
|
|
|
It looks like it's over for me. My quad has run dry and all my remaining work will only add up to 13k points. It looks like I will fall 9k short.
I was going for gold
Bronze on Aug 3
Silver on Aug 21
Gold =? Well about 25k short
I didn't find out PSP sieve ending until the challenge was under way. So did my best.
Looks like the next project to fall will be 321 sieve. Have silver now
Gold and a Pink are next after a Ruby PSA
Good Luck, Steve
____________
From the High Desert in New Mexico
|
|
|
|
I think I should just stop making any kind of prediction. It now appears, I have enough work to just barely make my Barney badge. I don't know how or where the work is coming from, I am just grateful it showed up.
So, if anyone has aborted work recently or moved on to another project and let your PSP sieve work time out, I thank you. And if there has been a little gnome assisting me in this endeavor, I thank you also.
BARNEY BADGE OR BUST!!!!
____________
|
|
|
|
I just got my Barney badge, hurray for me!!
I still have quite a number of WUs left, so if anyone wants me to abort the remainder so they can attempt to get their next level badge, please speak up.
____________
|
|
|
|
Don't ask, just do it!! :)
I need 150 wu with time live over 29.X, so please abort for over crunchers.
I will abort my all wu after my silver badge.
Thank you. |
|
|
|
Done, there all yours.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
PSP Sieve Update
Currently there are 0 tasks in progress.
History: work in progress
03 Nov: 0 tasks
02 Nov: 4 tasks
30 Oct: 354 tasks
29 Oct: 518 tasks
28 Oct: 1028 tasks
27 Oct: 2415 tasks
24 Oct: 3317 tasks
23 Oct: 6419 tasks
22 Oct: 7546 tasks
21 Oct: 12604 tasks
20 Oct: 13396 tasks
19 Oct: 15262 tasks
18 Oct: 32681 tasks
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
PSP (Sieve) is officially suspended
The last few remaining tasks are now complete. The official classification of this will be a suspension and not a termination. Should "improved" sieving methods come up in the future, this sieve will be reviewed for continuation. EDIT: And of course, should there still be k's remaining as we approach 50M, we'll need to resume sieving for n>50M. :)
This is a major accomplishment for both the Prime Sierpinski Problem and the Seventeen or Bust projects. In a little over three years (13 Oct 2007 to 3 Nov 2010) we helped advance the sieve from 1.5P to just over 83.1P. Also, the PSP (Sieve) was the inaugural project in the Challenge Series starting in 2008 and was a mainstay each year with the Ides of March Challenges. Now that it has joined Caesar, we'll have to find another project to replace it. :)
Thank you to all who contributed to this project.
____________
|
|
|
|
Since this is currently finished, should it not be removed as a choice for crunching, just in case someone thinks they can use it as a filler before a project since the units were so short?
____________
My lucky numbers are 121*2^4553899-1 and 3756801695685*2^666669±1 |
|
|