Author |
Message |
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Welcome to the “Drive for Five” Challenge
(part of PrimeGrid’s Challenge series)
PrimeGrid is attempting to reach n=5 million for 321 Search (currently at n~4.4M).
321 Search began in February 2003 by Paul Underwood and grew into a distributed prime search effort for the form 3*2^n-1. The initial goal was to build upon the completed work at Proth Search and extend the list of known primes to an exponent of 1 million (n=1M). That was quickly achieved so they advanced their goal to finding a mega prime for which they sieved up to n=5M.
As seen on PrimeGrid’s front page, that goal was achieved on 23 Mar 2008, 7:57:28 UTC, when Dylan Bennett of Canada returned a positive result for n=4235414 (3*2^4235414-1). official announcement | decimal representation
Now all that’s left is to complete the sieved work up to n=5M. Thus the name “Drive for Five”. There’s still a chance that a mega prime exists between now and n=5M.
The Challenge will begin 15 May 2008 00:00 UTC and end 1 June 2008 00:00 UTC or until all WU’s have been completed, whichever comes first. Application builds are available for Linux 32 bit and Windows 32 bit. These applications will be sent to 64 bit clients. Unlike the “Ides of March” Challenge, there is no advantage of 64 bit over 32 bit this time.
ATTENTION: This Challenge is not for the weak of heart. WU’s will take a minimum of 6.5 hours on fast computers (C2Q 2.4GHz) and upwards of 15 hours on slower computers (Athlon64 2.1GHz). It is HIGHLY RECOMMENDED that your systems be stable as the primality program LLR does not tolerate "even the slightest of errors." Please see this post for more details on how you can "stress test" your computer: http://www.primegrid.com/forum_thread.php?id=857&nowrap=true#8317
Scores will be kept for individuals and teams. Only WU’s that are issued and returned with a valid result during the Challenge dates will be counted. We will use the “prime score” method which is based on the n value (3*2^n-1) to score the challenge. The only difference is that the primary and double checker of a WU will receive the same score.
Therefore, each completed WU will earn a unique score based on its n value. The higher the n, the higher the score. For more information on scoring, see this thread: http://www.primegrid.com/forum_thread.php?id=895&nowrap=true#8728.
Please come join the challenge and help 321 Search complete a GREAT journey.
Best of Luck in wrapping up the initial stage of the 321 Search AND to hopefully finding another MEGA PRIME!!! :)
Additional Information
Once 321 Prime Search has been completed to n=5M, we will switch over to the +1 form (3*2^n+1). Currently, that prime search is in the Project Staging Area and is ready to enter BOINC as soon as -1 is complete to 5M. WU's are availabe now in LLRNet if you are interested in testing.
We will then take +1 to n=5M. Afterwards, we'll combine 32-1 and 32+1 together and take both up to n=10M. In that range, we expect another 4-6 Mega Primes! :)
____________
|
|
|
|
Very nice event. I will take part in it. even if I had no chance for points for the Challenge Series. ;)
But I´ve one or two questions. Is the app for 321 searc checkpointing and what´s up with the progressbar? Doesn´t it function?
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Very nice event. I will take part in it. even if I had no chance for points for the Challenge Series. ;)
But I´ve one or two questions. Is the app for 321 searc checkpointing and what´s up with the progressbar? Doesn´t it function?
Yes, all PrimeGrid's applications have checkpointing. Also, progress bar may be slow to show as a WU for 321 will take from 6 hours up to 15+ hours depending on your CPU.
Good Luck!
____________
|
|
|
|
Thanks a lot John. I will warm up my 4 little cores. ;)
____________
|
|
|
|
You can count me in on this one!!
____________
John M. Johnson "Novex" |
|
|
|
Greetings from Manchester U.K.
I will be doing my bit along with other team members.
b.
____________
Member of The UK BOINC Team.
Join us here
http://www.tiny.cc/UBT
http://bobtonson.page-visit.com
|
|
|
|
Can we see the progress somewhere, like at rieselsieve here: http://boinc.rieselsieve.com/orig/llr_range_grid.php
?
How many tests are remaining etc. ?
Cheers, and thumbs up, Yours H. |
|
|
|
You can have a look here: http://www.primegrid.com/stats_321_llr.php
Atm are around 6200 checks left.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
You can have a look here: http://www.primegrid.com/stats_321_llr.php
Atm are around 6200 checks left.
Please note that we work in collaboration with 321 Search. We reserve work on an as needed basis. Therefore, the ~6200 WU's left only go up to 4.6M. There's another 23129 WU's left from 4.6M-5M. We will reserve 0.1M ranges (~5780 WU's per 0.1M) as needed for the Challenge...with the goal to reach 5M by the end. :)
Best of Luck to everyone. Hopefully we'll find another Mega Prime before 5M.
____________
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
Any chance of having a page that shows how everyone (individuals and teams) is/are doing in this challenge?
Which could easily mutate into a final standings page after it is complete...
/me has got 3 days of WUs queued and rolling now, not that I expect to win or even come close, but one has to join in, don't you know? :-) |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Any chance of having a page that shows how everyone (individuals and teams) is/are doing in this challenge?
Which could easily mutate into a final standings page after it is complete...
/me has got 3 days of WUs queued and rolling now, not that I expect to win or even come close, but one has to join in, don't you know? :-)
Yes, same as last Challenge. Stats for teams and participants will be available on the front page in the top menu under "The Drive for Five Challenge". It will be updated every 6 hours...so 4 times a day. :)
EDIT: Please remember that score is based on the actual number tested and not BOINC credit. Here are some scores at the following n:
n=4.5M score=3010.49
n=4.6M score=3220.41
n=4.7M score=3439.97
n=4.8M score=3669.39
n=4.9M score=3908.90
n=5.0M score=4158.72
See this post for more information on scoring: http://www.primegrid.com/forum_thread.php?id=895&nowrap=true#8728
We'll also update daily how many WU's remain till n=5M. At our current rate, we only have 8 days of work left. However, we expect the rate to slow a little as everyone's cache is filled. Not to mention, as n increases, the WU completion time will increase.
____________
|
|
|
|
My 2ghz dual-core laptop sounds like it's vacuuming the tabletop, but seeing as how it takes only 8 hours to crunch a WU, six per day for sixteen days equals ninety-six possibly prime numbers.
(Sixteen days without playing video games ... Almost as bad as Shutdown Day!)
____________
|
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 94,577,455 RAC: 190,790
                     
|
First results are flowing in, I added stats links to the frontpage.
____________
|
|
|
|
Thanks Rytis. And could you please stop the challenge at this time? I don´t think that I could stay on such a good position. :)
____________
|
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 685 ID: 845 Credit: 2,886,413,454 RAC: 240,030
                              
|
OK, since I'm "numerophil", I'll put some resources on this challenge. ;)
Not everything, but at least a little bit. The LLR app is a good test for my new quadcore. :)
____________
|
|
|
|
I've got my P4 working on a 321 task too, but I'll keep my AMD GCW-sieving.
____________
Life is short and meaningless, unless you make the best of it.
http://server.by014.net/~weezepoe
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
24 hours into the Challenge and all is going well. Stats can be found here: Participants | Teams
EDIT: These stats can also be accessed in the top menu on the front page.
2100 WU's have been sent out. For those interested in how many WU's are left, here's the breakdown:
4.4M-4.5M - 259 complete (5354 complete before Challenge)
4.5M-4.6M - 1841 complete; 3827 remaining
4.6M-4.7M - 0 complete; 5795 remaining
4.7M-4.8M - 0 complete; 5715 remaining
4.8M-4.9M - 0 complete; 5794 remaining
4.9M-5.0M - 0 complete; 5825 remaining
Total 2100 complete; 26956 remaining
____________
|
|
|
|
I'm a little bit confused. Which wu's must i cruch for the contest?
I have activated the "3*2^n-1 prime search" in my account. On my client i have get 3 sorts of wu's:
PSP Sieve 1.05
GCW Sieve 1.09
LLR (3*2^n-1) 5.07
??? confused !!! |
|
|
|
You must crunch the LLR (3*2^n-1) 5.07 WU's. You can disable the other ones to get only the 321 units. Or perhaps these were already in your queue.
____________
Life is short and meaningless, unless you make the best of it.
http://server.by014.net/~weezepoe
|
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 94,577,455 RAC: 190,790
                     
|
I'm a little bit confused. Which wu's must i cruch for the contest?
I have activated the "3*2^n-1 prime search" in my account. On my client i have get 3 sorts of wu's:
PSP Sieve 1.05
GCW Sieve 1.09
LLR (3*2^n-1) 5.07
??? confused !!!
You must have got PSP/GCW work before you changed settings.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
I am reposting this from the first thread just as a reminder:
ATTENTION: This Challenge is not for the weak of heart. WU’s will take a minimum of 6.5 hours on fast computers (C2Q 2.4GHz) and upwards of 15 hours on slower computers (Athlon64 2.1GHz). It is HIGHLY RECOMMENDED that your systems be stable as the primality program LLR does not tolerate "even the slightest of errors." Please see this post for more details on how you can "stress test" your computer: http://www.primegrid.com/forum_thread.php?id=857&nowrap=true#8317
____________
|
|
|
Scott Brown Volunteer moderator Project administrator Volunteer tester Project scientist
 Send message
Joined: 17 Oct 05 Posts: 2379 ID: 1178 Credit: 17,805,338,706 RAC: 5,362,567
                                                
|
I am reposting this from the first thread just as a reminder:
ATTENTION: This Challenge is not for the weak of heart. WU’s will take a minimum of 6.5 hours on fast computers (C2Q 2.4GHz) and upwards of 15 hours on slower computers (Athlon64 2.1GHz). It is HIGHLY RECOMMENDED that your systems be stable as the primality program LLR does not tolerate "even the slightest of errors." Please see this post for more details on how you can "stress test" your computer: http://www.primegrid.com/forum_thread.php?id=857&nowrap=true#8317
Just to add to the time ranges:
C2Duo E6750 - 6.0 hours
Pentium M 1.73Ghz - 24.5 hours
Also, a quick question. There appears to be a penalty for AMD machines. For example, my Athlon x2's (a 4200+ and a 3800+) are getting pounded by my old Pentium D 830. The former are in the 15 hour range while the latter is around 10 hours. These are normally fairly comparable machines? Or is it maybe a memory issue (1.5 and 1.0 GB on the Athlons vs. 2GB on the Pent. D)?
____________
141941*2^4299438-1 is prime!
|
|
|
|
I think, there is really a penalty for AMD. My C2D@3.0 GHz needs around 6 hours to finish a WU. My brandnew laptop with AMD Athlon X2@1.9 GHz needs over 15 hours for completing one task...
____________
|
|
|
|
I am reposting this from the first thread just as a reminder:
ATTENTION: This Challenge is not for the weak of heart. WU’s will take a minimum of 6.5 hours on fast computers (C2Q 2.4GHz) and upwards of 15 hours on slower computers (Athlon64 2.1GHz). It is HIGHLY RECOMMENDED that your systems be stable as the primality program LLR does not tolerate "even the slightest of errors." Please see this post for more details on how you can "stress test" your computer: http://www.primegrid.com/forum_thread.php?id=857&nowrap=true#8317
A word of caution for those downloading this app for the first time. BOINC does a really bad job at estimating the expect time to complete these WUs so if you have a large cache you may wind up with more WUs than you can possible hope to finish before the deadline.
My Q6700 (stock) still estimates about 5 hours even though it's done a bunch of these WUs at about 6.5 hours per. BOINC estimates my AMD 3700+ at around 3 hours per WU. My laptop's T7500 is estimated to take around 1 hour per WU. |
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 94,577,455 RAC: 190,790
                     
|
I increased the estimates by a factor of 1.5, should help.
As for favoring Intel over AMD, it generally depends on CPU features like L2 cache. Older AMD models are a little slower for sure, I'm not exactly sure about newer models.
____________
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
Just posting this question for clarification...
I've submitted about 8 321 WUs since the drive began, and I see points awarded in the standings page. All good.
Some/most/all of these are still in pending status as per normal project behaviour.
What would happen to drive points if doublechecking later showed a bad residue? I assume they'd be deducted.
I guess that WUs submitted just before the end of the drive would count for points even whilst in pending state.
So - and this is the evil question - how long after the end of the drive do we wait for doublechecking before finalising the results? :-)
(I suspect the leaders all know they have good machines, since it's the usual suspects who score well the rest of the time, so I doubt it would actually alter the top N places anyway...) |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Also, a quick question. There appears to be a penalty for AMD machines. For example, my Athlon x2's (a 4200+ and a 3800+) are getting pounded by my old Pentium D 830. The former are in the 15 hour range while the latter is around 10 hours. These are normally fairly comparable machines? Or is it maybe a memory issue (1.5 and 1.0 GB on the Athlons vs. 2GB on the Pent. D)?
Historically speaking, Intel has always done better with primality testing and AMD has done better with sieving. However, with new technology and new programming, that doesn't always hold true these days. With certain sieve software, Intel has caught up with AMD. With AMD's Phenom, it comes close to Intel in primality testing.
The hardware/software combinations involved are far beyond the scope of my experience. However, the following details I have received from people much more knowledgeable than me:
The main issue is Double Precision Floating Point (DPFP) multiply/adds issued and completed per clock cycle. LLR uses George Woltman's GWNUM library. In recent versions, there was a move towards SSE2 and that had the side effect of giving AMD a better performance. Prior to this, it heavily relied on x87 FP and this is where the P4 really shined.
The new Phenom is much better than the Athlon64, but Penryn has the better of raw execution timings. Phenom still has the better memory access and management. But wait until Nehalem. That is really going to change things.
As for sieving, older sieves were Integer based and there is where AMD shined and still does. As for the new sieve programs of which sr2sieve & gcwsieve are among (which PrimeGrid uses), the code is mostly DPFP based so Intel does not fare as badly in sieving as it used to.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
So - and this is the evil question - how long after the end of the drive do we wait for doublechecking before finalising the results? :-)
(I suspect the leaders all know they have good machines, since it's the usual suspects who score well the rest of the time, so I doubt it would actually alter the top N places anyway...)
We don't expect this to alter the outcome of the Challenge (well, at least not the top 30 places). However, we will wait until the end is closer so we can review error rates and outstanding WU's to determine how much of an impact they will have. At that time, we'll make a decision on how long to wait for double checks, if necessary.
____________
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
Thanks John - that makes perfect sense. |
|
|
|
could the Primegrid challenge series page be updated so it links to the participants and team results of the drive 4 5 challenge?
____________
Life is short and meaningless, unless you make the best of it.
http://server.by014.net/~weezepoe
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
could the Primegrid challenge series page be updated so it links to the participants and team results of the drive 4 5 challenge?
Done :)
____________
|
|
|
|
That's strange: Team Sweden@boinc@Home dropped not only in rank, but in credit as well. Were they disqualified or bought out or what?
H. |
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 94,577,455 RAC: 190,790
                     
|
That's strange: Team Sweden@boinc@Home dropped not only in rank, but in credit as well. Were they disqualified or bought out or what?
H.
The top cruncher (Lennart) left the team and joined another one; challenge scores do not stay with the team but instead move with the participant.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
A discrepancy has been brought to our attention in the Challenge points system. Basically, our current system awards the same points for a 24 hour Challenge as it does for a 30 day Challenge.
It is appropriate that the time commitment of a longer Challenge deserves a larger reward. Therefore, we have modified the Challenge points to better represent the time commitment of the Challenge.
While most of the Challenges will be 24 hours, the “points modification” is necessary to allow us to run Challenges for PrimeGrid’s more difficult and time intensive projects: Cullen LLR, Woodall LLR, PSP LLR, and 321 LLR which need more than 24 hours to run an adequate Challenge.
The changes can be viewed here: PrimeGrid's Challenge Series
We greatly appreciate the time commitment individuals and teams make to participate in these Challenges. Thank you very much.
____________
|
|
|
|
Have any primes been found yet? |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Have any primes been found yet?
None have been reported. :( At most, we are hoping for 1 Mega Prime.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Congratulations to L'Alliance Francophone as they are the first team to surpass 10M Challenge points.
Also, a big thank you to everyone who is helping PrimeGrid and 321 Search tackle this monumental task. The progress achieved so far has been excellent. However, at this time, those huge 321 numbers are moving ahead of us in the race. They are proving to be very tough and are not going down easily. At our current rate, we will fall just short of the 5M goal.
Therefore, we would like to propose a "sprint" this weekend to see if we can get back even or maybe ahead of where we need to be in order to reach 5M by the end of 31 May. We humbly ask all participants and teams to please put a little extra on 321 for the 24th & 25th.
This is truly an uphill Challenge...the closer we get to 5M, the longer the tests take...which means in order to keep the same completion rate, we actually have to increase processing power as the Challenge progresses. It's a difficult task, but one that we are very close to accomplishing. Just over a 1/3 into the Challenge and 9937 WU's have been completed with 19309 remaining. Our daily totals have been:
5/15/2008 - 2100
5/16/2008 - 1166
5/17/2008 - 1847
5/18/2008 - 2002
5/19/2008 - 1347
5/20/2008 - 1285
5/21/2008 - 190 (as of this post)
Thank you again for participating in The Drive for Five Challenge and for all your help and support at PrimeGrid. :)
____________
|
|
|
|
Congratulations to L'Alliance Francophone as they are the first team to surpass 10M Challenge points.
We have to thanks our big cruncher, Banshee.
As there is a fight L'Alliance Francophone against SETI.USA on protein@home, without Banshee, we coud'nt be very present on this challenge |
|
|
|
Congratulations to L'Alliance Francophone as they are the first team to surpass 10M Challenge points.
Also, a big thank you to everyone who is helping PrimeGrid and 321 Search tackle this monumental task. The progress achieved so far has been excellent. However, at this time, those huge 321 numbers are moving ahead of us in the race. They are proving to be very tough and are not going down easily. At our current rate, we will fall just short of the 5M goal.
Therefore, we would like to propose a "sprint" this weekend to see if we can get back even or maybe ahead of where we need to be in order to reach 5M by the end of 31 May. We humbly ask all participants and teams to please put a little extra on 321 for the 24th & 25th.
This is truly an uphill Challenge...the closer we get to 5M, the longer the tests take...which means in order to keep the same completion rate, we actually have to increase processing power as the Challenge progresses. It's a difficult task, but one that we are very close to accomplishing. Just over a 1/3 into the Challenge and 9937 WU's have been completed with 19309 remaining. Our daily totals have been:
5/15/2008 - 2100
5/16/2008 - 1166
5/17/2008 - 1847
5/18/2008 - 2002
5/19/2008 - 1347
5/20/2008 - 1285
5/21/2008 - 190 (as of this post)
Thank you again for participating in The Drive for Five Challenge and for all your help and support at PrimeGrid. :)
Set most of my PG crunchers to devote most, if not all of their cycles to PG for this weekend. They should run out of work for other projects by then.
Hopefully this will move me up in the ranks for this challenge!
Mark
____________
|
|
|
|
Just over a 1/3 into the Challenge and 9937 WU's have been completed with 19309 remaining.
Just out of curiosity, in seeing the 19,309 WUs remaining, how often is that updated on the home page? That's the number that was on the page yesterday, and still is right now.
Also, thanks for updating the Primegen item on the home page.
____________
. |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Just out of curiosity, in seeing the 19,309 WUs remaining, how often is that updated on the home page? That's the number that was on the page yesterday, and still is right now.
The remaining WU's number is updated once a day...as close to 00:00 UTC as possible. As of this post, 18333 WU's remain...looks like we'll be close to 18000 in another 3 1/2 hours.
____________
|
|
|
|
That's fine. Yesterday when I looked, it was after midnight UTC, so that would make sense.
____________
. |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Congratulations to L'Alliance Francophone as they are the first team to surpass 10M Challenge points.
We have to thanks our big cruncher, Banshee.
And 1 day later, Congratulations to Banshee...the first individual to pass 10M Challenge points. That equates to over 3100 tests!!! :)
Also, thank you to everyone participating in the sprint this weekend. It will go a long ways towards helping us reach our 5M goal.
____________
|
|
|
|
Could be a little bit off topic but this is stopping me from fully running this challenge:
Why there is so much dispersion on the boinc credit for two quite similar WU (see bellow)?
I thought credit (Boinc) was co-related to the n being tested (in differential increments: 4.4M to 4.5M x, 4.5M to 4.6M y... and so on) , and not to the quickness of the co-checker of the WU (kind of "average")... if the attribution system is the second one I won´t run any more 321... sorry.
It takes too much of my resources to gain 2.25 credit/1000 seconds (compared to 3.40 in the second) - I´m 25012 on these WU´s:
http://www.primegrid.com/workunit.php?wuid=30020984
http://www.primegrid.com/workunit.php?wuid=29889756
(Incredibly my P4 is stronger at sieving, with 4.04 cr/thousand seconds!... ?) |
|
|
|
If I'm not mistaken the credit is based on a fixed scale for the n being tested.
____________
|
|
|
|
If I'm not mistaken the credit is based on a fixed scale for the n being tested.
I thought the same... now I doubt it. See the two WU. There _cannot_ be a 50% difference for just 0.1M n variation! |
|
|
|
If I'm not mistaken the credit is based on a fixed scale for the n being tested.
I thought the same... now I doubt it. See the two WU. There _cannot_ be a 50% difference for just 0.1M n variation!
Where do you see the n value? I sure don't see it listed anywhere.
~BoB
____________
|
|
|
|
If I'm not mistaken the credit is based on a fixed scale for the n being tested.
I thought the same... now I doubt it. See the two WU. There _cannot_ be a 50% difference for just 0.1M n variation!
Where do you see the n value? I sure don't see it listed anywhere.
~BoB
That´s easy: deconnect boinc access to internet before the WU finish... then copy the file of the result when finished, you reconnet and it goes but you get the copy in your directory (it is the same txt as the LLR result of "k*2^n-1 is not prime re.....").
Yes, I´m too curious. |
|
|
|
According to John in the first post of this thread:
"Scores will be kept for individuals and teams. Only WU’s that are issued and returned with a valid result during the Challenge dates will be counted. We will use the “prime score” method which is based on the n value (3*2^n-1) to score the challenge. The only difference is that the primary and double checker of a WU will receive the same score.
Therefore, each completed WU will earn a unique score based on its n value. The higher the n, the higher the score. For more information on scoring, see this thread: http://www.primegrid.com/forum_thread.php?id=895&nowrap=true#8728."
____________
|
|
|
|
Well at least we are clear on using the right value for n.....
@aardvark1917
the current discussion is on credit rather than points
~BoB
____________
|
|
|
|
Well the easy answer is, boinc doesn't use that formula based on n. |
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
Indeed it doesn't.
I do think it sucks that you do a WU that takes X amount of CPU and wallclock time, and the amount of BOINC credit you get depends on if the other WU done as doublecheck was done on a fast or slow computer. If the other WU was done on a fast machine, you get less credit - on a slow machine, you get more.
Is one of the reasons I tend to work on fixed credit subprojects, cos I like stats that reflect work done, not how it was done.
(Not that I expect this to change or anything, and other projects have issues with this sort of thing) |
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 94,577,455 RAC: 190,790
                     
|
This method of scoring was chosen because we had no way to correctly estimate credit based on n value. However, we now have scoring formula, which is very simple to adapt to credit, and I promise that we'll do that after the challenge is over (it's probably not a very good idea to start fiddling with credit specifically for the app that is in the challenge).
____________
|
|
|
|
This method of scoring was chosen because we had no way to correctly estimate credit based on n value. However, we now have scoring formula, which is very simple to adapt to credit, and I promise that we'll do that after the challenge is over (it's probably not a very good idea to start fiddling with credit specifically for the app that is in the challenge).
retroactively? (ok, does that word exist in English?)....
and could you, please, publish the boinc credit per n (in 0.1M margins from 4.5 for 321 proyect)? (just like you did with the challenge scores, I mean...)
thx |
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
That's great news - will this apply to all LLR projects going forward?
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
FYI...BOINCstats now offers stats for Challenges. This current Challenge and all future Challenges will be available there. :)
It's located on the front page, just right of the news posts.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The sprint is officially over. Well, it was more like a nice stroll in the park...at least the scenery was nice. :)
We didn't get the push that we needed but we were able to sustain throughput even with the increasing test times. Good news is that we are over half way completed (16193 WU's sent out, 12863 WU's remaining).
Thank you to everyone for putting forth that extra effort this weekend as well as those who continue to participate in the Challenge. We have 6 days remaining...let's do as much as possible. At our current pace, we are expected to fall about 3700 WU's short. Not bad considering how difficult and time intensive this Challenge is. Below is our daily throughput:
5/15/2008 - 2100
5/16/2008 - 1166
5/17/2008 - 1847
5/18/2008 - 2002
5/19/2008 - 1347
5/20/2008 - 1285
5/21/2008 - 1189
5/22/2008 - 1339
5/23/2008 - 1307
5/24/2008 - 1277
5/25/2008 - 1334
To see the impact we've had on 321 Search, please see this graph. It'll put a smile on your face. :)
Thanks again for all the hard work! It's very much appreciated.
____________
|
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 685 ID: 845 Credit: 2,886,413,454 RAC: 240,030
                              
|
The sprint is officially over. Well, it was more like a nice stroll in the park...at least the scenery was nice. :)
Unfortunately, I had trouble with my quad (a defect RAM module caused many outages and hard resets, sometimes even killing workunits). But now it's up and running again and with some overclocking I compensated the longer runtimes. ;)
At least I'm back in Top30 participants for now. :)
____________
|
|
|
|
Wow that graph really does show us making a nice dent :) I know I've been able to hold my own since we started the challenge and it's been great so far! I will stay with it until the WU's are done or when we hit the 5M mark. I won't quit just because the month's over thats for sure :)
Everyone keep up the good work!!!
____________
John M. Johnson "Novex" |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Wow that graph really does show us making a nice dent :) I know I've been able to hold my own since we started the challenge and it's been great so far! I will stay with it until the WU's are done or when we hit the 5M mark. I won't quit just because the month's over thats for sure :)
Everyone keep up the good work!!!
Thanks John! That's GREAT to hear. :)
For anyone else interested in staying beyond the challenge end date, right now we're approximately 3 1/2 days over...but that continues to grow. By 31 May we may be 5 days over. That means 5 June would be the n=5M date. We would appreciate any extra help to reach that goal. :)
Just a reminder, the next Challenge is on 12 June and will only last 24 hours. More details to come after the completion of this Challenge.
____________
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
My AMD machine can go back to the sieve, but the rest will stay on this until it's complete (subject to joining in the next challenge - if need be, they can go back to 321 afterwards).
So please keep us up to date on the countdown to 5M. |
|
|
|
I have a quick question. What is best for speed, set the cpu affinity to one core, or multiple cores? |
|
|
|
Can we keep the stats on this going on an unofficial (ie not scoring for your challenge series) basis until we get to 5M or until I've whopped novex's ass?
John M Johnson: :) Could go either way old man (I tip you to win by a hair) but real fun watching it!
|
|
|
|
Can we keep the stats on this going on an unofficial (ie not scoring for your challenge series) basis until we get to 5M or until I've whopped novex's ass?
John M Johnson: :) Could go either way old man (I tip you to win by a hair) but real fun watching it!
Lol oh your on bro! I'm still trying to catch up to you in points and man were both close as hell all the time in terms of credits hehe, it's good for a little competition /evil grin
John M. Johnson "Novex"
____________
John M. Johnson "Novex" |
|
|
|
Can we keep the stats on this going on an unofficial (ie not scoring for your challenge series) basis until we get to 5M or until I've whopped novex's ass?
John M Johnson: :) Could go either way old man (I tip you to win by a hair) but real fun watching it!
Lol oh your on bro! I'm still trying to catch up to you in points and man were both close as hell all the time in terms of credits hehe, it's good for a little competition /evil grin
John M. Johnson "Novex"
John M. Johnson "Novex" .....
What is the reason you switched teams form SETI.USA to PrimeSearchTeam
Francophone is over 17 M points ahead ????? If you have a good reason I will switch too I would love to see L'Alliance Francophone lose their lead.
____________
From the High Desert in New Mexico
|
|
|
|
I would love to see L'Alliance Francophone lose their lead.
I'm sure Banshee has his price. ;)
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
For those switching from 321 Search at the conclusion of the Challenge on 1 June, please consider modifying your cache to a day or less. This will help minimize the expired/aborted/not completed WU's after the end of the Challenge and give us a better idea of the remaining WU's. We are currently experiencing a "heavy" expired rate from WU's that were sent out at the beginning of the Challenge. I'll have more details at 00:00 UTC tomorrow.
For those continuing 321 Search afterwards, as it looks right now, there will be plenty to crunch for at least a week and probably longer.
EDIT: For those continuing, it would be a good idea to lower cache at this time to at most 3 days. In the week following the Challenge, we'll ask for everyone to eventually go down to 2 days, 1 day, and finally 1/2 day. You can do that now if you like, it won't hurt anything. :)
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The expired WU's from the beginning of the Challenge have been excessive. 321 WU's have a 12 day deadline. On the 27th we started receiving expired WU's from the first day of the Challenge. We expected a moderate level of expired WU's from the new attaching hosts as BOINC adjusted to the longer 321 WU lengths. However, as you can see below, we used a lot of our current effort to process those expired WU's on the 27th.
5/15/2008 2100
5/16/2008 1166
5/17/2008 1847
5/18/2008 2002
5/19/2008 1347
5/20/2008 1285
5/21/2008 1189
5/22/2008 1339
5/23/2008 1307
5/24/2008 1277
5/25/2008 1334
5/26/2008 1264
5/27/2008 938
5/28/2008 1023
This will have an impact on our goal of n=5M. Good news is the expired WU's decreased on the 28th. Hopefully this trend will continue. Current projection for n=5M is 7 June. Expect this to change as completion rates change.
Please remember to start lowering your cache. A day or less is ideal.
____________
|
|
|
|
Lol oh your on bro! I'm still trying to catch up to you in points and man were both close as hell all the time in terms of credits hehe, it's good for a little competition /evil grin
John M. Johnson "Novex"
I am coming for you, boy.
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
5/29/2008 1007 WU's went out.
8631 WU's remain until n=5M. 7 June still remains the n=5M date.
Please remember to start lowering your cache as soon as possible. Half day is best right now. The reason why this is important is that once we get down to our last 5000 WU's, we will start lowering the deadline. Currently it's 12 days. For each 1000 WU's after 5000, we'll transition to 10 days, 7 days, 5 days and 3 days. This will expedite the successful conclusion of all WU's up to 5M.
Once we get down to the last 500 WU's, it is recommended that you select an additional project so that when we finally do reach 0 WU's remaining, your machines won't go idle. However, please do not deselect 321, as sporadic work will still be available when expired/error WU's are processed. BTW, TPS is the featured project in the next Challenge...PrimeGrid's Birthday Challenge. :)
Good Luck to all the participants and teams as y'all battle it out for those last points. The white flag is almost upon us.
Thanks again for all your help and support. :)
____________
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
I know that wrapping up 321 to 5M is the goal here, but I believe that reducing the deadline by too much will result in work being lost unnecessarily. Please don't be too harsh! :-)
Also, how soon after the conclusion of this will the new 32+1 project be live?
(and will this be under the same 321 banner, or a seperate subproject?)
|
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 685 ID: 845 Credit: 2,886,413,454 RAC: 240,030
                              
|
It looks like the stats are not updated any more. No one got challenge points in the last 8 hours: http://www.boincstats.com/stats/challenge_stats.php?pr=pg&challenge=1&object=user
Only some points were subtracted (maybe invalid results?).
____________
|
|
|
RytisVolunteer moderator Project administrator
 Send message
Joined: 22 Jun 05 Posts: 2653 ID: 1 Credit: 94,577,455 RAC: 190,790
                     
|
It looks like the stats are not updated any more. No one got challenge points in the last 8 hours: http://www.boincstats.com/stats/challenge_stats.php?pr=pg&challenge=1&object=user
Only some points were subtracted (maybe invalid results?).
The challenge script ended challenge prematurely, I restored the functionality. The challenge was limited to 15 days, whereas it should have been limited to 17. Human error, no points lost :)
____________
|
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 685 ID: 845 Credit: 2,886,413,454 RAC: 240,030
                              
|
It looks like the stats are not updated any more. No one got challenge points in the last 8 hours: http://www.boincstats.com/stats/challenge_stats.php?pr=pg&challenge=1&object=user
Only some points were subtracted (maybe invalid results?).
The challenge script ended challenge prematurely, I restored the functionality. The challenge was limited to 15 days, whereas it should have been limited to 17. Human error, no points lost :)
Thanks for fixing. So the battle for p28 continues (getting even more exciting with Sid catching up). :)
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The white flag has been waved…we are down to our last 24 hours. Good Luck to everyone.
5/30/2008 1054 WU's went out. 7 June still remains the n=5M date.
I know that wrapping up 321 to 5M is the goal here, but I believe that reducing the deadline by too much will result in work being lost unnecessarily. Please don't be too harsh! :-)
Every CPU cycle is needed for 321 Search to conclude efficiently. Therefore, we don't plan on being "too harsh". :) However, that's why it's so very important for everyone to start lowering their cache. My Computing preferences Network usage is already set to the following:
- Computer is connected to the Internet about every 0 days
- Maintain enough work for an additional 0 days
My cache is empty and I get 1 WU at a time now.
Also, to help avoid any "wasted" cycles, at the conclusion of the Challenge, all "slower" computers (those that take over 72 hours per WU) should consider deselecting 321 Search from their project list. This will insure that no deadlines are missed. :) The sieves are always a good place for slower machines...OR...TPS is the featured project in the next Challenge (12 June) so why not get a head start. :) BTW, the 3 day deadline probably won't come till 5-6 June, so there's still plenty of time to crunch.
Additionally, if your computer is crunching several projects (internal to PG and/or external), the above advice would be good to follow. Or give PG a higher percentage of resources until n=5M is reached.
Also, how soon after the conclusion of this will the new 32+1 project be live?
(and will this be under the same 321 banner, or a seperate subproject?)
As for the beginning of 32+1, we'll start that as soon as humanly possible...AND a nice clean finish to 32-1 will help facilitate that greatly. Once 321 Search reaches n=5M, it will become part of PrimeGrid's 321 Prime Search (just a small name change, 321 Search to 321 Prime Search). 321 Prime Search will bring 32+1 even with 32-1 and then continue both of them from n=5M upward. BTW, for 32+1, we are expecting 2-3 new primes below n=5M with one or more being a Mega Prime.
Thanks again for everyone's help! :) It's now a race to the checkered flag.
Stats: Participants | Teams
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Less than 12 hours remain.
If you have any WU's that will be completing between now and the finish, remember to hit project update in BOINC to make sure those WU's get submitted and earn points for the Challenge.
Good Luck!
Stats: Participants | Teams
____________
|
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 685 ID: 845 Credit: 2,886,413,454 RAC: 240,030
                              
|
10 hours to go... time to start the final spurt. ;)
____________
|
|
|
|
1 Hr to Go, Hit those update buttons
Good Luck from SETI.USA
____________
From the High Desert in New Mexico
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
Hold off those update buttons until 00:01Z - then submit :-)
|
|
|
|
Woohoo! I made it into the points this challenge (just missed on the last one). |
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 685 ID: 845 Credit: 2,886,413,454 RAC: 240,030
                              
|
It's over now. Pretty close race for me in the end. Was a great fun. :D
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The Checkered flag is out (thanks pschoefer)...and The Drive for Five Challenge has ended. Unfortunately, no Mega Prime was found. However, a good amount of work was completed. Here are some stats:
WU's
29056 Started
22512 Sent out
6544 Remain
255 teams and 1681 individuals participated.
The Challenge Points stats will remain unofficial for the time being as we investigate how long to wait for "double check" WU's to come in. Congratulations to L'Alliance Francophone as the top team and Banshee as the top individual. Remember, the top 40 earn points for this Challenge.
For those continuing until n=5M, at our current rate, 7 June is still the date. For those moving on, thank you very much for participating in The Drive for Five Challenge. PrimeGrid's next Challenge will be 12 June. It will be a 24 hour Challenge celebrating PrimeGrid's Birthday. Rest up and get ready...the 24 hour Challenge is fast and furious...especially if you compare it to this one. :D
Thank you again to everyone. We appreciate your participation!
Stats: Participants | Teams
EDIT: Those staying till n=5M, please remember to lower your cache to less than a day. It is very important. See this post for suggestions. The next 1000 WU's have already been released with a 10 day deadline. We currently only have 674 WU's remaining with a 12 day deadline. The next 1000 WU's will go out with 7 days deadline, then 6 days, 5 days, and finally 4 days.
____________
|
|
|
Vato Volunteer tester
 Send message
Joined: 2 Feb 08 Posts: 840 ID: 18447 Credit: 639,858,000 RAC: 498,429
                           
|
Well done all - I'm pretty please to get some points on this one - last time I was on holiday and it was over before I knew. |
|
|
|
That was fun.
Thanks to the great and good for organising.
I'd rather a contest like this than the shorter ones because there is more chance to see positions change. I'm also less likely to be bothered to set up for a 24 hour one than a long one and if I forget it doesn't matter so much with the long one as I can join in later.
Thanks to others for taking part. Especially to novex/John M Johnson for being a real sportsman; we had fun and were neck and neck to the end.
I must fly there is some American bloke with a baseball bat banging on the door... |
|
|
|
That was fun.
Thanks to the great and good for organising.
I'd rather a contest like this than the shorter ones because there is more chance to see positions change. I'm also less likely to be bothered to set up for a 24 hour one than a long one and if I forget it doesn't matter so much with the long one as I can join in later.
Thanks to others for taking part. Especially to novex/John M Johnson for being a real sportsman; we had fun and were neck and neck to the end.
I must fly there is some American bloke with a baseball bat banging on the door...
Lol It was a blast Tim, neck and neck with ya the whole way was fun!! You were a great friend to compete with. Thanks to the whole community as well for helping with this great Drive for Five Challenge :)
____________
John M. Johnson "Novex" |
|
|
|
I've set my PG cache to 0 day's, but i noticed that it has also set my other project's to a 0 day cache. It look's like al my project's have been set to a 0 day cache. I can change each project invitingly back manually. Is this meant to be happening?
Thank you
Speedy |
|
|
|
The cache setting affects the whole client and therefore all projects you are attached to. It is not possible to change the cache size for just one project.
____________
Synecdoche
|
|
|
|
The cache setting affects the whole client and therefore all projects you are attached to. It is not possible to change the cache size for just one project.
Ok thank you for this information
Speedy |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The results of the "Drive for Five" Challenge are now official. The Challenge Series rankings will soon be updated.
Thank you again to everyone who participated.
For those continuing with the n=5M goal, please see the this thread for updates.
____________
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
The Challenge Series rankings have been updated.
Stats: Participants | Teams
____________
|
|
|