| Author |
Message |
|---|
Lexs Volunteer developer
Send message
Joined: 16 Mar 08
Posts: 61
ID: 20289
Credit: 49,033,000
RAC: 0
               
|
|
Time to update the ratio, as the first AP26 has been found !
AP26
43142746595714191+23681770*23#*n for n=0..25 found by [AF>HFR>RR] Jim PROFIT
____________
 |
|
|
|
|
|
Wow. Way to be on top of things paying attention. I see it's up on the twitter feed as well, but way to be on top of things.
So what's next??? Look for the 2nd AP26? Look for the AP27? Sideline this project to give power to other projects?
What happens to the PG Birthday Challenge?
So many questions!!! |
|
|
 mfl0pProject administrator
Volunteer developer
Send message
Joined: 5 Apr 09
Posts: 226
ID: 38042
Credit: 891,705,380
RAC: 288,828
                         
|
Time to update the ratio, as the first AP26 has been found !
AP26
43142746595714191+23681770*23#*n for n=0..25 found by [AF>HFR>RR] Jim PROFIT
I just saw this, wow! Great work to everyone!
Biggest thanks to Jarek and Geoff, without them it wouldn't be possible!
____________
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13570
ID: 53948
Credit: 249,538,791
RAC: 131,114
                    
|
Wow. Way to be on top of things paying attention. I see it's up on the twitter feed as well, but way to be on top of things.
So what's next??? Look for the 2nd AP26? Look for the AP27? Sideline this project to give power to other projects?
What happens to the PG Birthday Challenge?
So many questions!!!
Well, since it's currently the only CUDA project at PrimeGrid, there's not much point in making way for something else.
Congrats to whomever found the AP26, as well as everyone involved in organizing as all as everyone who participated!
Mike
EDIT: I should say the only released CUDA project.
____________
My lucky number is 75898524288+1 |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
               
|
|
Please see this post for details: AP26 Found
The project will be suspended once the current queue is empty. Discussion began at the end of last year of where to go after the AP26 is found. Once we have time to discuss this further, we'll post additional information.
The AP26 Search was definitely PrimeGrid's most accessible project (GPU, CPU, Cell/Blade). Currently the only other project with the potential of GPU support is PPS Sieve. The ppsieve program has a CUDA port. We will review what's needed to bring this into production.
As for the PrimeGrid's Birthday/Summer Solstice Challenge, another project will be selected to replace it.
____________
|
|
|
mfl0pProject administrator
Volunteer developer
Send message
Joined: 5 Apr 09
Posts: 226
ID: 38042
Credit: 891,705,380
RAC: 288,828
|
|
John, I think it would be interesting to know the full range of K and shift PrimeGrid had to test to find the first AP26 known. Is it possible to post this after the current queue is exhausted?
____________
|
|
|
|
|
|
Big congratulations!
I have updated Primes in Arithmetic Progression Records.
As the only known AP26 it is in every record table and a few other places including the opening line. |
|
|
|
|
Please see this post for details: AP26 Found
The project will be suspended once the current queue is empty. Discussion began at the end of last year of where to go after the AP26 is found. Once we have time to discuss this further, we'll post additional information.
The AP26 Search was definitely PrimeGrid's most accessible project (GPU, CPU, Cell/Blade). Currently the only other project with the potential of GPU support is PPS Sieve. The ppsieve program has a CUDA port. We will review what's needed to bring this into production.
As for the PrimeGrid's Birthday/Summer Solstice Challenge, another project will be selected to replace it.
Please, please, please include x86 linux in CUDA ports.
Congratulations to the one who found the AP26!
____________
6r39 7ri99
Beware the dual headed Gentoo with Wine! |
|
|
|
|
|
WHat??? suspend the project?? I've been hitting it hard to get the barney badge and only need 360K more. Using 8 vid card shouldn't be very long. How big is the current queue? I could dump my PSP sieve and SOB wus in progress and load up to get the barney badge. Don't tell me I wasted 1.169 Meg of computation without the chance to get the badge. I'd have stopped at 1M if I thought this was a possibility.
What other project can end before I get my badge levels? I.E which should I not bother running?
I'm serious about getting the barney badge, so serious, in fact did abort every non AP wu in progress and 12 were SOB with 35 hours elapsed each. Set cache to 10 days.
____________
 |
|
|
|
|
|
I wanted the badge, too...this is the last AP, or there are lots of AP24s and AP25s, which haven't been found yet?
____________
 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13570
ID: 53948
Credit: 249,538,791
RAC: 131,114
|
What other project can end before I get my badge levels? I.E which should I not bother running?
Hmmmm. The general answer is "Probably none of them". A number of the tests DO have specific goals, so they could, in theory, end. However, unlike AP26 those projects are looking for multiple goals to 'solve' the problem, so it's unlikely all of those numbers will be found. I think the general consensus is that none of these will be completed in less than a few years.
The Sierpinski problems and the Riesel problem all have specific goals. There are three associated LLR projects: the Sierpinski problem (Seventeen or Bust) is searching for 6 primes, the Prime Sierpinski problem is searching for another 9 primes, and the Riesel problem is searching for 64 primes. There's also two sieve projects associated with these problems, the PSP/SOB sieve and the Riesel sieve.
So there's 5 projects in total that can, in theory, end.
However, unlike AP26 (which wasn't expected to run for years and years), it's not expected that these problems will be solved quickly. SoB, although down to only 6 k's, is way up there in the n range and, as you know, the WUs take a really long time. It's not unreasonable to guess that the other two problems will also need very large n's to knock of their last remaining K's. So although these are projects which do have the potential for ending, it might take a very long time. It might not happen in our lifetimes.
Another thing to consider is that these projects are each attempting to prove a conjecture. If the conjecture is true, they will eventually prove the conjecture to be true and be done. On the other hand, these projects are not capable of proving the conjecture to be false. If the conjecture is indeed false, the projects will run forever searching for a solution that does not exist.
Bottom line: those five projects might end, but it's probably not likely it will happen any time soon. We'll probably be three or four badge colors beyond amethyst before these things finish! One of our great grandchildren might find the last remaining K-prime on a 10-PHz optical quantum computer with 50K CPU cores, cranking out a trillion credits per day.
As for the other projects, I think they are all open ended, but my understanding of those is still kind of fuzzy. I'm sure someone much more knowledgeable than I will chime in with a definitive answer.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
I'm just going to make the bronze badge. On the bright side, there are over 100,000 workunits left. That should be enough for a few people to get a badge. |
|
|
|
|
|
I'm glad I was able to help out with this effort with the Mac & Mac/CUDA ports.
Do we have any statistics that can be pulled out of the DB about how many WUs were crunched by each client type? I suspect from the number of AP24/25/26 found by PS3 and CUDA hosts that when AP26 closes down we may be losing the crunching power of a large community of users.
We should work on getting a GPU sieve (at least) ASAP so we can keep these users crunching for PrimeGrid...
- Iain |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13570
ID: 53948
Credit: 249,538,791
RAC: 131,114
|
|
From the home page:
Note: there will be no more work once the queue is empty, so we would like to ask people not running for the badges to stop running AP26 subproject and give a chance for others to reach their final levels.
I've aborted all the remaining AP26 tasks in my queue (about 50ish) and set the CUDA beast back to crunching at GPUGrid. (I've got GPUGrid set as a backup project, so once there's a new CUDA application here it will switch back automatically.)
Good luck to everyone trying to get a badge!
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
I guess I'm a little confused as to why this project is ending so abruptly...
I understand that the AP26 project had a very specific goal in mind, one that we have now met. But if I understand the theory behind the project correctly, there should be more AP26s out there, as well as AP27s, AP28s, etc. Why not simply rename this project and call it the AP27 subproject and continue on?
It seems to me that it would be a shame to not continue utilizing all the hard work that the developers have put in. You can run AP26 on a Cell Blade, any CPU with any OS, CUDA capable cards, and there is an ATI app on the way. Why not keep pushing this project? The theory, code, hardware, and infrastructure all seem to be in place... |
|
|
|
|
|
I'm of the opinion they should give us two weeks worth of work to finish getting our goals met. Everything about the project is for the project. All we contributors get is cobblestones and badges.
I had reached gold and switched everything to other projects to get them to silver and gold. Days later they came out with the barney badge, and so, it was back to AP26. Since I could figure no way to configure the project or boinc options to only run AP26 on GPU while running PSP and SOB on the CPU except to set up two different preferences, and due to limits on the number of wus, I had to switch every the location of each computers every 12 hours or so. Been doing it for weeks. In other words, I've been investing lots of my personal time to do this.
Now, I find out they project is abruptly ending. I'm 300K from the barney badge. I figure at 60K/day It'll take 5 days. OK, I try to get a measely 5 days work. Only to be met by "max wus in progress" messages. It allows 100GPU wus and at 9 min per that's only 15 hours work. 100 cpu wus/core at 30 min each is 50 hours or just over two days. I've tried everything I can think of and can't get that 5 days worth loaded up. Looks like I'll end at 2,400,000+ on AP 26. and that'll be a waste of 1.4 Meg of wus that could have gotten me gold on two other projects(when added to my existing totals). It'll just continue to wear and eat at me.
How about giving us a set deadline weeks in the future that give some consideration to your contributors. Asking others not to get work so we can get more work is an effort, but not one on the projects part. It would be so easy to just kick out a couple hundred K more wus for us to use for our own goals.
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1219
ID: 12001
Credit: 18,565,548
RAC: 0
|
I understand that the AP26 project had a very specific goal in mind, one that we have now met. But if I understand the theory behind the project correctly, there should be more AP26s out there, as well as AP27s, AP28s, etc. Why not simply rename this project and call it the AP27 subproject and continue on?
Because the app isn't tuned to find AP27s and because finding an AP27 will take many years (barring luck). |
|
|
|
|
I'm of the opinion they should give us two weeks worth of work to finish getting our goals met. Everything about the project is for the project. All we contributors get is cobblestones and badges.
+1
I would also like some additional time to meet my goal. And it's not like it's going to waste. There may be other AP26s found, right?
____________
Reno, NV
 |
|
|
geoffVolunteer developer
Send message
Joined: 3 Aug 07
Posts: 99
ID: 10427
Credit: 343,437
RAC: 0
|
|
Congratulations to Benoãt and everyone else involved in this project!
I think it was great how so many people got involved in improving, porting, and testing the application. I couldn't keep up with all the new developments, but with the porting to cell/GPU architectures especially I think a lot of things have been learned that could be a big help to future PrimeGrid projects.
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
I think it was great how so many people got involved in improving, porting, and testing the application.
Yes, AP26 was a perfect example of the success that can be achieved through open source and collaboration. Well done everyone!
____________
|
|
|
|
|
|
I noticed that an article published in 'The Observer' http://www.guardian.co.uk/culture/2010/apr/11/the-10-best-mathematicians the day before the AP26 was found mentions Tao & Green's proof of the existence of arbitrarily large AP-k's, and states that the largest known is an AP25. I have emailed them with an update, and hopefully a little bit of nice publicity for PrimeGrid.
- Iain |
|
|
|
|
|
Times running out...
There's only 15K wus left...
Anyone gonna tell us how to get a full cache of work?
Even by altering the duration correction factor down to 0.500000, I hit the max cache limit of 100/core GPU and 96/core CPU or 15 hours on the GPU and 2 days on the CPU.
I need 4.5 days worth and it looks like we'll be out of work within two days.
help.
____________
|
|
|
|
|
|
I think we're not going to make it. I'm in just about the same boat as you. Now I need to figure out something else for my PS3s to do. |
|
|
|
|
|
AP26 is one of my favourite subprojects...and finally, just today, I got my first AP23!
It would be awesome if it were available at least until another CUDA app developed. Personally, I say we go for the AP27. Yeah it might take a while, but there is a lot of GPU computing power available for such a search (especially if ATI GPUs can be utilized) and it's not like it would take any longer than some of the other subprojects (SoB, PSP).
Cheers!
Alan
____________
  |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13570
ID: 53948
Credit: 249,538,791
RAC: 131,114
|
|
AP26 is also one of my favorite projects, but I'm not in favor of keeping it running just to do crunching just for crunching's sake.
Yes, I'm having fun collecting badges and I suppose I keep an eye on credit as much as the next guy, but both credit and badges are almost 100% meaningless and arbitrary. That's not the reason I'm here.
IMHO, if we continue with AP26, there needs to be a reason to do so. If not, all we're doing is consuming electricity for no good reason.
And, yes, I too was looking forward to collecting another badge in AP26 -- another month or two and I'd have gold. But in the end, that's nothing more than a hundred or so yellow pixels on the screen. I'll survive without it.
I think some people may have lost sight of the fact that the end of AP26 is not a bad thing. It's an incredible, fantastic, event! We succeeded in finding something that had never been found before. That is why I'm here.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
I'm all in favor of not wasting electricity.
I'm all in favor of not needlessly wasting crunching resources (the contributors resources).
What I have trouble with is the sudden ending, the "We've decided to turn off the tap without the slightest notice or concern about the wants/presumed needs of the contributor.
IF they'd said "The minute we find an AP26, we'll be turning off the tap", then It'd be oK.
IF in absence of that, they said "we've reached the goal, thank you, the project will shut down in X weeks, please attempt to concentrate your work to achieve any goals you may have". Then I'd be OK.
In my opinion, I care little about primes or sieve files. I'm crunching for cobblestones and badges. Cutting it off this abruptly is to my mind "wasting the electricity required to get 1.2 M cobblestones (that above the amount needed for gold)". It represents an "opportunity cost", in that I could have been using that electricity to get gold in a couple other sub projects.
Can I trade in that 1.25M cobblestones and switch it to other projects? NO. It's a waste, as gold is all it'll ever be.
____________
|
|
|
HAmstyVolunteer tester
Send message
Joined: 26 Dec 08
Posts: 132
ID: 33421
Credit: 12,510,712
RAC: 0
          
|
|
when the goal of a subproject is reached, it will be switched off immediately. that was the case with tps and now with ap26, that wasn't unforeseeable imho.
and of course the credit is not wasted, it stands for the work you have done. without it, the goal would be achieved some days/weeks later!
____________
 |
|
|
|
|
|
I don't consider it a waste to continue to find APs any more than it is a waste to find the next proth prime in the PPS subproject. Since AP26 is the only CUDA enabled search, it is currently the only way I can use my GPUs to contribute to Primegrid. I can understand stopping the project if the resources can be allocated to another subproject, but in this case they can't (yet). This is why if the project is to be axed I think it should wait until the CUDA app for the PPS sieve is available (but I really want to continue on to the AP27!).
Cheers!
Alan
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
After further consideration and in the absence of an established project termination policy, but in the spirit of good faith, work has been extended. This is solely for those users who are "on the cusp" of their next badge level. It is not intended as a continuation of the project. The work queue will be kept full until 19 Apr 2010 20:00 UTC ("find" date +7 days).
Although contrary to conventional wisdom, in the era of badges, extra consideration should be given beyond simply the project's end goal. In just over 15 months, the AP26 was found and thus, the end goal has been met.
Other projects that have specific end goals are:
- Prime Sierpinski Problem (LLR)
- Seventeen or Bust (LLR)
- Sophie Germain Prime Search (LLR)
- The Riesel Problem (LLR)
- PSP/SoB (Sieve)
- The Riesel Problem (Sieve)
Therefore, as PrimeGrid's project termination policy, when an end goal has been met, the work queue will be kept full for no more than 1 week beyond the "find" date. The project will then be closed when all outstanding work has been returned and credited.
NOTE: A good reference for "on the cusp" is within 1 week of the next level.
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
AP26 is one of my favourite subprojects...and finally, just today, I got my first AP23!
It would be awesome if it were available at least until another CUDA app developed. Personally, I say we go for the AP27. Yeah it might take a while, but there is a lot of GPU computing power available for such a search (especially if ATI GPUs can be utilized) and it's not like it would take any longer than some of the other subprojects (SoB, PSP).
Congratulations on your first AP23.
Another CUDA app has already been developed by Ken for the ppsieve. It's just not in production yet. It is unknown at this time when this will go live. However, it has similar GPU vs. CPU timing ratios as AP26 did. Hopefully ppsieve can be rolled out sooner rather than later. :)
An AP27 search would be anywhere from 50 to 100 times more difficult than the AP26 search. While that's no reason not to do it, it is reason enough to pause. An AP27 search is not off the table for future consideration, but it is off for at least the remainder of this year.
It is disappointing to loose the computing power that supported the AP26 search. Hopefully other BOINC projects will benefit from this success. However, as Michael says, "It's an incredible, fantastic, event!" This is a time for celebration.
Again, congratulations to all who participated in this project.
____________
|
|
|
|
|
|
I think the AP26 found knew that Sony killed Linux on the PS3 and thereforeit came out in the open... ;)
Could the end-terms of all searched ranges be published for those that want to search for another AP26 outside of primegrid after the queue has run out? |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
I think the AP26 found knew that Sony killed Linux on the PS3 and thereforeit came out in the open... ;)
Yes, Sony's move is very disappointing...ironic now that the PS3 made such a significant discovery. Tell that to the Board of Directors. :D
Could the end-terms of all searched ranges be published for those that want to search for another AP26 outside of primegrid after the queue has run out?
Yes, all details will be released once the project is complete.
____________
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13570
ID: 53948
Credit: 249,538,791
RAC: 131,114
|
I think the AP26 found knew that Sony killed Linux on the PS3 and thereforeit came out in the open... ;)
Yes, Sony's move is very disappointing...ironic now that the PS3 made such a significant discovery. Tell that to the Board of Directors. :D
The truly ironic thing is that the AP26 find could have been great publicity for Sony. Now they can't mention it at all.
On the other hand -- there's also nothing stopping them from revising the firmware again. Maybe if this find was sufficiently publicized Sony might reconsider its position.
____________
My lucky number is 75898524288+1 |
|
|
|
|
Another CUDA app has already been developed by Ken for the ppsieve. It's just not in production yet. It is unknown at this time when this will go live. However, it has similar GPU vs. CPU timing ratios as AP26 did. Hopefully ppsieve can be rolled out sooner rather than later. :)
An AP27 search would be anywhere from 50 to 100 times more difficult than the AP26 search. While that's no reason not to do it, it is reason enough to pause. An AP27 search is not off the table for future consideration, but it is off for at least the remainder of this year.
It is disappointing to loose the computing power that supported the AP26 search. Hopefully other BOINC projects will benefit from this success. However, as Michael says, "It's an incredible, fantastic, event!" This is a time for celebration.
Again, congratulations to all who participated in this project.
Cheers John!
I will be looking forward to the ppsieve app. I'm not extremely interested in the other BOINC CUDA projects. I've run Collatz and GPUgrid in the past, but have stopped them on my personal computers since they do not intrigue me enough to pay for the huge increase in power consumption I see when using the GPUs. I will likely give my GPUs a break until that ppsieve app becomes available.
I am indeed super excited that an AP26 was found! Sad to see the project go though:( Will keep my fingers crossed for an AP27 project in the future. Perhaps when the average computer has multiple 3rd generation Fermi cards or better and an octacore CPU, finding an AP27 will start to look like a feasible 2-3 year project.
____________
|
|
|
|
|
After further consideration and in the absence of an established project termination policy, but in the spirit of good faith, work has been extended. This is solely for those users who are "on the cusp" of their next badge level. It is not intended as a continuation of the project. The work queue will be kept full until 19 Apr 2010 20:00 UTC ("find" date +7 days).
Although contrary to conventional wisdom, in the era of badges, extra consideration should be given beyond simply the project's end goal. In just over 15 months, the AP26 was found and thus, the end goal has been met.
Other projects that have specific end goals are:
- Prime Sierpinski Problem (LLR)
- Seventeen or Bust (LLR)
- Sophie Germain Prime Search (LLR)
- The Riesel Problem (LLR)
- PSP/SoB (Sieve)
- The Riesel Problem (Sieve)
Therefore, as PrimeGrid's project termination policy, when an end goal has been met, the work queue will be kept full for no more than 1 week beyond the "find" date. The project will then be closed when all outstanding work has been returned and credited.
NOTE: A good reference for "on the cusp" is within 1 week of the next level.
PrimeGrid Admins,
Thanks for the reconsideration on AP26. The 1 week termination policy sounds totally reasonable to me.
Having the policy published will encourage me (and maybe some others) that have set credit/badge goals to work more actively on them since the above projects will end 1 week after their goal has been achieved. I know for me having the estimated end of project dates over at WCG allows me to allocate CPUs there to meet my goals; but there is no way to estimate end dates here since the next WU may contain that elusive number we are hunting.
____________
 |
|
|
|
|
|
Since everyone who worked on AP26 contributed toward the discovery, why don't you just bump everyone's badge up one level for the discovery.
____________
 |
|
|
|
|
|
thank you. |
|
|
|
|
|
Thank you for extending the deadline - that's very helpful to many.
Are there any thoughts to editing existing AP26 code to search for an AP27 and creating a new subproject? I ask because I'm very interested in PrimeGrid, and AP26 was the subproject where I understood the research goals best, and I'd like to continue to see the APk proof examined... |
|
|
|
|
|
I'm too voting for AP27 ))
____________
wbr, Me. Dead J. Dona
 |
|
|
|
|
Thank you for extending the deadline - that's very helpful to many.
Are there any thoughts to editing existing AP26 code to search for an AP27 and creating a new subproject? I ask because I'm very interested in PrimeGrid, and AP26 was the subproject where I understood the research goals best, and I'd like to continue to see the APk proof examined...
An AP27 search would be anywhere from 50 to 100 times more difficult than the AP26 search. While that's no reason not to do it, it is reason enough to pause. An AP27 search is not off the table for future consideration, but it is off for at least the remainder of this year.
As far as i could understand the difficulty is not in the code itself, somewhere i read the code used for ap26 can be used for ap27 with a minimum of changes. The Reason why there will be no ap27 search in near future is the huge amount of crunching time it would take to find a ap27.
There is a prediction in the AP26 search Thread, that says the chance a AP26 is part of a AP27 is 11% meaning up to 9 AP26 have to be found to get an AP27
To find the first AP26 it took 15 month, to find 9 AP26 would take 10+ years.
I absolutely understand why an AP27 search is out of interest to primegrid ...
I hope i got the things right, please correct me if i messed something up.
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
Of course, if something spectacular comes along which significantly improves the computing efficiency of the search or should PG's computing power grow by an order of 10 by the end of the year, AP27 could be implemented earlier. :)
____________
|
|
|
JarekVolunteer developer
Send message
Joined: 28 Dec 08
Posts: 57
ID: 33488
Credit: 10
RAC: 0
|
There is a prediction in the AP26 search Thread, that says the chance a AP26 is part of a AP27 is 11% meaning up to 9 AP26 have to be found to get an AP27
To find the first AP26 it took 15 month, to find 9 AP26 would take 10+ years.
Unfortunately things are even worse than that. The problem is that expected time for each next AP26 would be longer and longer. The expected time for 10th AP26 could be 10-20 as much as the expected time for the 1st one.
That's true, the 1st AP26 was overdue, we should have found 2 AP26 by now, but that doesn't influence much the order of magnitude of computations needed to expect AP27. Also AP27 search program could run a bit faster that AP26 search, but still, that doesn't influence the order of magnitude.
As for the above quoted 11% figure:
* this could be slightly optimistic like most ratios in my predictions
* even my predictions assumed search for AP26. As the search goes on, all the ratios grow, meaning that on average it would take more than 9 AP26 to honestly expect AP27.
Anyway, AP27 search should be ready to invest 10-20 times the computer power than all the power used by the whole PrimeGrid up to now (approaching 2.8G cobblestones at the moment). As you know it is extremely hard to predict how long the actual search would take, but it is reasonable to assume that AP27 search would most likely consume something in the range 10G-100G cobblestones.
|
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2222
ID: 1178
Credit: 9,225,654,939
RAC: 3,298,978
                                       
|
|
The new Fermi cards from NVidia appear to be around 3 to 4 times faster than the fastest current GTX 2xx cards (single GPU...not the dual GTX 295). The current HD 5800 series ATI cards are already that fast (or faster!) with a new updated series from ATI due out in the late summer or fall.
While a factor of 10 might not be possible, a factor of 4 or 5 might. :)
____________
141941*2^4299438-1 is prime!
 |
|
|
|
|
|
First of all, congratulations to everyone who contributed to AP26.
Then, just want to ask something: someone has verified if this AP26 could be also an AP27? |
|
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 920
ID: 3110
Credit: 192,430,496
RAC: 144,358
            
|
|
That's easy enough to do. The 27th term would be 43142746595714191+23681770*223092870*26 = 180506831531191591.
$ factor 180506831531191591
180506831531191591: 113 49789 32083504163
So the 27th term has three prime factors.
____________
|
|
|
|
|
That's easy enough to do. The 27th term would be 43142746595714191+23681770*223092870*26 = 180506831531191591.
$ factor 180506831531191591
180506831531191591: 113 49789 32083504163
So the 27th term has three prime factors.
Just for completeness sake, how about for n= -1?
(My silly unix computer has a limit of 1E14 on the factor command) |
|
|
pschoeferVolunteer developer
Volunteer tester
Send message
Joined: 20 Sep 05
Posts: 664
ID: 845
Credit: 2,274,533,360
RAC: 793,103
                
|
That's easy enough to do. The 27th term would be 43142746595714191+23681770*223092870*26 = 180506831531191591.
$ factor 180506831531191591
180506831531191591: 113 49789 32083504163
So the 27th term has three prime factors.
Just for completeness sake, how about for n= -1?
43142746595714191+23681770*223092870*(-1) = 37859512559734291
$ factor 37859512559734291
37859512559734291: 43 972277 905558581
____________
  |
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 791
ID: 18447
Credit: 305,284,678
RAC: 370,047
   
|
|
Just to stick my oar in (I've been away, stuck behind the volcanic ash no-fly zone)....
Congrats to all!
I do think it's a pity that we aren't continuing AP26 for a while due to it being available on the widest range of platforms, and also has the smallest memory footprint combined with not being very "CPU hot" like LLR. I say keep it going until some of these features are matched by other subprojects.
Stopping it certainly means I'll be stopping a few machines from running at PG as I need a thermally and memory unobtrusive app (using cpu is fine).
____________
 |
|
|
|
|
|
50K left to go, have plenty of work on hand to complete the barney badge. Should have it by tomorrow morning.
thanks
____________
|
|
|
|
|
50K left to go, have plenty of work on hand to complete the barney badge. Should have it by tomorrow morning.
thanks
Congrats! It looks like it'll take me until tuesday to get it, which leads to my question. How is the shutdown going to be handled?
Option A: No more work added to the que as of 2010-04-19 2000UTC, but all work in the que at that point will be distributed, users will be allowed to finish what they have and stats will be finalized once all work is submitted.
Option B: No more work added to the que as of 2010-04-19 2010UTC, all work in que abandoned, users allowed to finish what they have and stats will be finalized once all work is submitted.
Option C: 2010-04-19 2000UTC stats are frozen, project is done.
I really hope it's A or B, because I should get my 2.5 mil on 2010-04-20 sometime. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Congrats! It looks like it'll take me until tuesday to get it, which leads to my question. How is the shutdown going to be handled?
Option A: No more work added to the que as of 2010-04-19 2000UTC, but all work in the que at that point will be distributed, users will be allowed to finish what they have and stats will be finalized once all work is submitted.
Option A, and the queue will be quite small by the time 2010-04-19 20:00 UTC comes around.
____________
|
|
|
|
|
|
Keep pouring it on "endless mike". If you PM me, I'll give you a method to get the max wus allowed by law 100 GPU and 96 cpu/core if you're not already getting that message in the message tab.
17K to go. |
|
|
|
|
|
Hello Barney badge. PSP sieve barney badge is next
____________
|
|
|
|
|
|
YAY!
Congrats!
Damn, now I need to crunch another application......
____________
 |
|
|
|
|
|
Not long time left !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Fill your caches now.
Lennart
|
|
|
|
|
|
I think I made it, started out a 3 days ago with 760K short for the Blarney I did not think i would make it.
But now, with just 80K left to go, I am sure to be there with what I have in cache.
I will do a NNW and abort WU's so anybody else trying to make the Blarney will have a better chance.
Good luck everyone! |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
|
2010-04-19 20:00 UTC has been reached. No more work will be inserted into the buffer. The buffer will be allowed to empty. Once all work has been returned and credited, AP26 will come to a very successful close. :)
____________
|
|
|
|
|
|
So, it looks like I made it to purple.
If I abort the rest of my tasks, will they be availale to someone else who might need just a few more, or will they be lost to the ether?
I will continue to work on them if they'll just be lost, but if they would be available to help someone else, I'd rather do that. |
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2651
ID: 1
Credit: 36,106,205
RAC: 192,851
          
|
So, it looks like I made it to purple.
If I abort the rest of my tasks, will they be availale to someone else who might need just a few more, or will they be lost to the ether?
I will continue to work on them if they'll just be lost, but if they would be available to help someone else, I'd rather do that.
They will be reissued to others.
____________
|
|
|
|
|
|
Hi, as far as I know (and I don't know a lot...), AP26 was the only primegrid project that run on PS3 with boinc (with this release of boinc ps3-boinc.tar.gz installed).
With the sucesfull end of the AP26, I wonder what my PS3 can crunch... I'm a physicist so I'm interest mostly by physics and pure mathematics. I'not so welling to stack my 3 ps3 (buyed only to crunch on primegrid) in the closet and wait for the AP27...
Is anyone know if I can continue to crunch on primegrid with these and if so, what release/version of the boinc software I need to install???
I really hope that primegrid continue to support PS3 even with the disapointing move of sony...
Thanks in advance and congratulation for the AP26 found!!! |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13570
ID: 53948
Credit: 249,538,791
RAC: 131,114
|
Is anyone know if I can continue to crunch on primegrid with these and if so, what release/version of the boinc software I need to install???
I really hope that primegrid continue to support PS3 even with the disapointing move of sony...
Thanks in advance and congratulation for the AP26 found!!!
Short answer: No. This is/was the only project available on a PS3 (or a CUDA-GPU, for that matter).
Long answer: Well, maybe, if the stars line up and someone wants to put the work in.
As you're probably aware, some problems lend themselves to being solved on parallel processors, and some don't. From what I understand the LLR-type prime-finding programs at PrimeGrid don't parallelize easily, so they're not going to run on any type of graphics processor. The AP26 obviously does, and apparently so do the sieves, as evidenced by ongoing work on a CUDA sieve program.
Once that CUDA-based sieve program is released, it's not inconceivable that someone might go ahead and port that to run on the ATI GPUs and/or the PS3. If the latter happens, you might be able to run the PrimeGrid sieve projects on your PS3s.
That's a lot of "ifs", the biggest of which is probably whether there's someone out there willing to do the coding. With the new PS3s not being able to run Linux, and the supply of older PS3s shrinking as people upgrade their firmware, the effort to code for the PS3 will become of lesser and lesser value as time goes on.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Thanks Michael of this respond.
It's look like that my PS3 return where they come from... on ebay :( |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1219
ID: 12001
Credit: 18,565,548
RAC: 0
|
Hi, as far as I know (and I don't know a lot...), AP26 was the only primegrid project that run on PS3 with boinc (with this release of boinc ps3-boinc.tar.gz installed).
With the sucesfull end of the AP26, I wonder what my PS3 can crunch... I'm a physicist so I'm interest mostly by physics and pure mathematics. I'not so welling to stack my 3 ps3 (buyed only to crunch on primegrid) in the closet and wait for the AP27...
Is anyone know if I can continue to crunch on primegrid with these and if so, what release/version of the boinc software I need to install???
I really hope that primegrid continue to support PS3 even with the disapointing move of sony...
Thanks in advance and congratulation for the AP26 found!!!
I believe that you can participate in some of the PRPNet projects using a PS3. IIRC, someone was able to build phrot, which can be used for PPS and GCW13. |
|
|
|
|
|
Don't sell it, yoyo@home's subproject OGR uses it, too!!
Edit : Here is an example
____________
|
|
|
|
|
|
I'd be willing to port some of primegrid apps to cuda, but I cannot find the proper source code to do it. If any one knows where I can find source code to the projects on here, I would be greatfull.
____________
  |
|
|
|
|
|
I have a Slim, so I was always out of luck. I'm sort of curious what it would take to make a Sony-sanctioned app for Life with PlayStation, like Folding@Home. It would need a pretty front-end, of course. What else?
Anyone here do any Cell programming? :/ |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Is anyone know if I can continue to crunch on primegrid with these and if so, what release/version of the boinc software I need to install???
Please see this thread for continued PS3 work at PrimeGrid. It relates to running PRPNet in the Project Staging Area.
____________
|
|
|
|
|
Please see this thread for continued PS3 work at PrimeGrid. It relates to running PRPNet in the Project Staging Area.
Thanks for the link. I think I will try this. Is the PRPNet application generated credit (I guess they give some sort of credit...) compatible with boinc/boincStats? |
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 791
ID: 18447
Credit: 305,284,678
RAC: 370,047
|
|
Yes, it will count under the Project Staging Area (which will get you the PSA badge).
____________
|
|
|
