Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummerslowrise

Message boards :
Proth Prime Search :
help how do i know when i have found prime
Author 
Message 

Hi prime grid
I am really new to this and wanted to donate to the cause using my graphics cards
Proth Prime Search (sieve) tasks
Completed tasks 180
Credit 405,180.00
Factors found 1076
Average factors per workunit 5.9778
How do I know when i have found a prime number, if so how do i claim it, also does factors found 1076 mean i have found a prime number
many thanks
Data  


Only LLR applications can find prime.
You are doing sieveing, so you cannot find prime number.
____________
92*10^^{1439761}1 REPDIGIT PRIME :) :) :)
314187728^^{131072}+1 GENERALIZED FERMAT
31*332^^{367560}+1 CRUS PRIME
Proud member of team Aggie The Pew. Go Aggie!  


Please have a look here to see what sieving is doing:
http://www.primegrid.com/forum_thread.php?id=630
It is an important part of finding prime numbers but as Crunchi mentioned already you cannot find prime numbers with sieving.
With the BOINC application you can use your graphic card only for sieving tasks. But on PRPNet you can also search for primes. See here:
http://www.primegrid.com/forum_thread.php?id=1215
http://www.primegrid.com/forum_thread.php?id=3215
____________
ESP: Eliminated k=94373
SR5: Eliminated k=97366 and k=325918  

Michael GoetzVolunteer moderator Project administrator
Send message
Joined: 21 Jan 10 Posts: 13513 ID: 53948 Credit: 237,712,514 RAC: 903

Please have a look here to see what sieving is doing:
http://www.primegrid.com/forum_thread.php?id=630
It is an important part of finding prime numbers but as Crunchi mentioned already you cannot find prime numbers with sieving.
With the BOINC application you can use your graphic card only for sieving tasks. But on PRPNet you can also search for primes. See here:
http://www.primegrid.com/forum_thread.php?id=1215
http://www.primegrid.com/forum_thread.php?id=3215
Another option, if you would prefer to use your graphics card to directly search for prime numbers, as opposed to merely helping the overall effort by sieving, would be to use GeneferCUDA with the GFN262144 and GFN524288 projects. See this PRPNet thread for details. The number in my signature was found a few weeks ago running GeneferCUDA on my graphics card.
A lot of BOINC users have some trepidation about using PRPNet. It's actually very simple to set up, and offers more control of what your computer is doing than does BOINC. It's kind of like driving a car with a manual transmission instead of an automatic. Not necessarily harder, just different.
____________
My lucky number is 75898^{524288}+1  

JohnHonorary cruncher
Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0

I still think I may be cheated on this stuff. Still I have not been provided with a satisfactorily explanation.
Using an online prime number calculator, you may be able to see that the number used for p in this example, 7266111922899983, is a prime number on its own, though it is a factor of the assumedly much higher number 79817*2^45327351+1 .
Are all factors supposed to be prime numbers on their own when it comes to the large numbers then?
The very nature of a sieve uses ever increasing prime numbers to see if they divide a determined search field. In the case of the current PPS (Sieve), that search field is k*2^n+1 for k<10000 and 3M<n<6M.
Yes, the factors are prime but they are not "found" primes. They are generated by the sieve. The sieve uses these "small" generated primes to see if they divide any of the candidates in the sieve file.
Anyway, checking the tpfactors.txt.0 for one of my PPS tasks, I get the following line (among several above each other):
10386770072114297  6423*2^3764378+1
10386771725936107  6141*2^5276803+1
Is it then just a coincidence that p is starting with some number at the start of this file, then increases for different numbers for the formula made by k*b^n+1 and ends with the larger number for p than in the first line together with some other number for k*b^n+1 ?
Can anyone please explain. Maybe p only are the last digits of a possible even larger number used in this calculation?
No, it is not a coincidence. As you have determined:
p  k*b^n+1
In the above example, p=10386770072114297 is the factor that divides the candidate: 6423*2^3764378+1
Candidate 6423*2^3764378+1 no longer needs to be primality tested with LLR. We now know it is a composite. It will be removed from the master candidate file.
The current sieve depth is p=81P (81000T). So far, the sieve has used primes up to this number to see if they can divide any of the candidates k*2^n+1 for k<10000 and 3M<n<6M. The last sieve for n<3M over the same k range went up to p=100P. It is currently unknown how high the 3M<n<6M file will need to go.
____________
 


The current sieve depth is p=81P (81000T). So far, the sieve has used primes up to this number to see if they can divide any of the candidates k*2^n+1 for k<10000 and 3M<n<6M. The last sieve for n<3M over the same k range went up to p=100P. It is currently unknown how high the 3M<n<6M file will need to go.
John, how do you decide how deep should sieves go?
Can you please provide information what sieves depth and new ranges should we expect for PPS Sieve?
Can you also give the same information on Cullen/Woodall sieve please? I wonder what range are we sieving now on Cullen/Woodall?
This information is valuable to know how long will both sieves last around. Knowing that helps with decisions on video card upgrades )))
Thank you in advance!
 


John, how do you decide how deep should sieves go?
Can you please provide information what sieves depth and new ranges should we expect for PPS Sieve?
Can you also give the same information on Cullen/Woodall sieve please? I wonder what range are we sieving now on Cullen/Woodall?
This information is valuable to know how long will both sieves last around. Knowing that helps with decisions on video card upgrades )))
Thank you in advance!
Some information you will find here.  

Michael GoetzVolunteer moderator Project administrator
Send message
Joined: 21 Jan 10 Posts: 13513 ID: 53948 Credit: 237,712,514 RAC: 903

This information is valuable to know how long will both sieves last around. Knowing that helps with decisions on video card upgrades )))
Thank you in advance!
At the present time there's at least a few years worth of crunching on the two sieves, an immense (but finite) amount of crunching to be done by GeneferCUDA looking for GFN PRPs, and in the hopefully very near future a perhaps almost limitless amount of crunching to be done with llrCUDA. The bottom line is that even if you went and bought some $2000 Tesla GPUs and crunched with those until they're hopelessly obsolete, we will most likely still be working on some of the same projects we're crunching today. That assumes you're interested in using your GPU for more than just sieving; if that assumption is incorrect then refer to the thread linked to in the post prior to this one.
____________
My lucky number is 75898^{524288}+1  

JohnHonorary cruncher
Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0

John, how do you decide how deep should sieves go?
Can you please provide information what sieves depth and new ranges should we expect for PPS Sieve?
Can you also give the same information on Cullen/Woodall sieve please? I wonder what range are we sieving now on Cullen/Woodall?
There are many factors that determine sieve depth...i.e. search goals, search space range, hardware/software used, technology advances, etc.
However, in general, we sieve to a depth where factor removal is the same time as a primality test of a candidate at a certain n level. For the current PPS (Sieve) of 3M<n<6M, that n level is roughly 4.95M. For Cullen/Woodall (Sieve) 10M<n<25M, that n level is roughly 20M. As you can see, there's plenty of sieving left.
PPS (Sieve) will switch to 6M<n<9M after the current range is done. It is unknown when that might happen but years is a good estimate.
NOTE: This presume you plan to test the entire sieve file. For conjectured searches, k are removed once a prime is found, therefore shrinking the sieve file a bit. For PSP, SoB, and TRP we expect there to be k's remaining at n=50M.
____________
 


At the present time there's at least a few years worth of crunching on the two sieves, an immense (but finite) amount of crunching to be done by GeneferCUDA looking for GFN PRPs, and in the hopefully very near future a perhaps almost limitless amount of crunching to be done with llrCUDA.
As far as I see llrCUDA isn't doing well at the moment. GTX460 shows the same time as say i5 (judging by posts in the forum). Assuming that i5 has 4 threads and consumes twice less energy the application isn't efficient. When GTX570 starts to outperform i7 (around the same price) only then it will be worth using in llrCUDA.
Please correct me if i'm wrong about it.  


Thank you for the answer John!
For the current PPS (Sieve) of 3M<n<6M, that n level is roughly 4.95M.
We're currently at the range 81P  82P and I read it would finish at 100P. Does it mean that PPS Sieve is coming to an end of 3M<n<6M range? I see that we finish 1P in about 24 days (up to 3 moths to go I assume). Or we're just finishing 5M range? Sorry about the questions that might sound silly to you )))
 

Michael GoetzVolunteer moderator Project administrator
Send message
Joined: 21 Jan 10 Posts: 13513 ID: 53948 Credit: 237,712,514 RAC: 903

At the present time there's at least a few years worth of crunching on the two sieves, an immense (but finite) amount of crunching to be done by GeneferCUDA looking for GFN PRPs, and in the hopefully very near future a perhaps almost limitless amount of crunching to be done with llrCUDA.
As far as I see llrCUDA isn't doing well at the moment. GTX460 shows the same time as say i5 (judging by posts in the forum). Assuming that i5 has 4 threads and consumes twice less energy the application isn't efficient. When GTX570 starts to outperform i7 (around the same price) only then it will be worth using in llrCUDA.
Please correct me if i'm wrong about it.
You're not wrong, hence the "hopefully very near future" part of my statement. It's not there yet, obviously.
My point was there's plenty of GPU work available in the short term. Just considering software that actually IS production quality right now, you have GeneferCUDA crunching at n=18 and n=19, and we also have n=20, n=21, and n=22 being sieved prior to starting PRP testing.
A rough backofthenapkin calculation shows that n=22 alone would produce about 20 million hours worth of work for a GTX 460. That's roughly 100,000 WUs at 200 hours per WU. The WUs are huge, but so is the potential payoff: any PRP found will be around 13 million digits long, and except for the first 500 or so lowest b values, all would set a new world record for the largest known prime number if the PRP is proven prime.
____________
My lucky number is 75898^{524288}+1  

JohnHonorary cruncher
Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0

We're currently at the range 81P  82P and I read it would finish at 100P. Does it mean that PPS Sieve is coming to an end of 3M<n<6M range? I see that we finish 1P in about 24 days (up to 3 moths to go I assume). Or we're just finishing 5M range? Sorry about the questions that might sound silly to you )))
The previous sieve ended at p=100P. The currently one, however, will go much higher. The end depth is unknown at this time.
____________
 

Message boards :
Proth Prime Search :
help how do i know when i have found prime 