| Author |
Message |
|---|
|
|
|
Forgive me if this has come up before. I searched, I swear! :)
How common is it for PPS to rediscover old primes? I am getting no shortage of results from my processing, but they are both primes found one, two years ago. Mr. Harvey seems to have covered the field pretty well. Or is it that when people cite their prime hit statistics, they strictly mean original hits? |
|
|
 warddr Volunteer tester
Send message
Joined: 7 Feb 08
Posts: 254
ID: 18735
Credit: 10,383,352
RAC: 0
          
|
|
It isn't really common..
The ranges we (primegrid) search are unsearched. There are some primes found in these range, but noone here knows why, good guesses?
____________
 |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
 
|
It isn't really common..
The ranges we (primegrid) search are unsearched. There are some primes found in these range, but noone here knows why, good guesses?
Steven Harvey had done a number of ranges that PrimeGrid is now getting to. There will be some primes found that he had already discovered. For these ranges PrimeGrid is really doing a double-check. |
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 785
ID: 18447
Credit: 262,879,636
RAC: 0
                    
|
|
AFAIK He usually submits his residues to PG, so we only have to do a single pass, rather than two.
____________
 |
|
|
|
|
|
How about running PPS on something that Primegrid has already submitted to the top 500 primes db?
Supposedly, I was the initial finder on a "new" prime, found using PPS on a WU that was sent to one of my machines in January or February. (http://www.primegrid.com/primes/?section=primelist&userid=51267)
However, that same prime is in the prime db as having been submitted in October, before I even joined Primegrid. You can find it at http://primes.utm.edu/primes/page.php?id=90659
If I'm reading the entry correctly, Primegrid was involved in the submission of that particular prime. I'm still pretty new here, so I don't completely understand how all of this is set up, but it strikes me as odd that if PG found a prime and submitted it in October, that it would then have me check the same prime in January -- and credit me as the initial finder...
-JMP |
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 785
ID: 18447
Credit: 262,879,636
RAC: 0
|
|
Steven Harvey credits PrimeGrid on his finds even though his search is manual - maybe he's had some factors supplied on those ranges...? Puzzle-Peter found the one you list - maybe it was found via PRPnet?
____________
|
|
|
|
|
|
I thought that PRPnet used the same underlying database as the BOINC side of Primegrid, so that the two should not duplicate efforts.
I'm just kind of curious how this all works...
-JMP |
|
|
VatoVolunteer tester
Send message
Joined: 2 Feb 08
Posts: 785
ID: 18447
Credit: 262,879,636
RAC: 0
|
|
No, they do not share the same database.
____________
|
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
They do not share the same database. Lennart has loaded the two databases such that there is no overlap between the two of them. |
|
|
|
|
They do not share the same database. Lennart has loaded the two databases such that there is no overlap between the two of them.
In which case it's highly unlikely that the prime I found using BOINC was previously found using PRPnet.
I suppose it's only a matter of time before I'm actually the first to get to a newly discovered prime...
-JMP |
|
|
vasmVolunteer tester
Send message
Joined: 6 Dec 08
Posts: 47
ID: 32604
Credit: 990,892
RAC: 0
             
|
|
The issue with ranges that have been searched outside Primegrid and we're now double-checking has been discussed before. It's the workunits that have a fixed 600sec runtime and unknown (---) computer host.
What is discussed by Jeremy Posner above happened to me as well. Found a prime as an initial finder, another Primegrid user was the double-checker and apparently a third user has already submitted it crediting Primegrid.
Judging by the name in my case it also looks like it was Puzzle-Peter (PRPnet?). If it was PRPnet and a range was searched earlier and primes were found and reported then fair enough. A BOINC double-check could be done. But as it stands for 6813*2^485125+1 Primegrid has triple checked. And if that happens for a whole range it seems like wasted effort. |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
|
|
I don't see any reason why PrimeGrid can't remove known primes from the database so that this doesn't continue to happen. Let's see if John and Lennart have any ideas. |
|
|
|
|
Judging by the name in my case it also looks like it was Puzzle-Peter (PRPnet?). If it was PRPnet and a range was searched earlier and primes were found and reported then fair enough. A BOINC double-check could be done. But as it stands for 6813*2^485125+1 Primegrid has triple checked. And if that happens for a whole range it seems like wasted effort.
sorry, I haven't seen this for some time it seems. I have indeed done that range offline and reported the results. I was told my results were going to be imported into the database and only a double check would be done in BOINC. As far as I know it was not intended to send out two instances of these tests via BOINC. As to why this did not work I have no idea.
____________
There are only 10 kinds of people - those who understand binary and those who don't
 |
|
|
|
|
|
PrimeGrid tells me that I was the initial discoverer of a new prime today. However, it looks like it was submitted to the Top 5000 Primes database back in June, crediting PrimeGrid as a discoverer. I thought known primes weren't being sent out as WUs. Yet I somehow "discovered" this prime today.
This is the second time this has happened to me. PrimeGrid credits me as the initial discoverer of two primes, both of which were already in the db with PrimeGrid credited as a discoverer. It's no wonder that PPS is falling out of the top 5000, as we're clearly working on numbers that have already been prime tested. Can we please purge all of the known primes from future WUs?
-JMP |
|
|
|
|
|
Now it's just getting silly. Primegrid today added another prime to my tally as an initial discoverer. So I look in the Top Primes database, and find that it was submitted on Saturday, with PrimeGrid listed as one of the discoverers, but not in my name.
Is there some reason that I'm only finding primes that PrimeGrid is already credited with having discovered? Why are these being sent out as WUs? Is there some way to opt out of testing already-discovered primes?
-JMP |
|
|
|
|
|
Well, the project seems to be taking a break, spending its time double-checking a large range submitted by someone else. That's why all the WUs have names ending with _1, minimum, and no matter how fast you are, '---' manages to submit before you.
Despite that, I agree. I have not actually discovered a Proth yet in 38,000 tests; they all get submitted to the Top 5000 by independent checkers without getting filtered from PPS. Once we get back into claiming original primes, if I get one more false positive for 'initial finder,' I'm not checking any more Proth numbers. |
|
|
|
|
Well, the project seems to be taking a break, spending its time double-checking a large range submitted by someone else. That's why all the WUs have names ending with _1, minimum, and no matter how fast you are, '---' manages to submit before you.
Despite that, I agree. I have not actually discovered a Proth yet in 38,000 tests; they all get submitted to the Top 5000 by independent checkers without getting filtered from PPS. Once we get back into claiming original primes, if I get one more false positive for 'initial finder,' I'm not checking any more Proth numbers.
But that's the rub. Word is put out that PPS needs attention so that the new primes found at the leading edge will continue to make the top 5000 list, so people start running more PPS WUs. However, when we start getting primes from that newfound focus, we discover that they've all been submitted to the top 5000 list already. This then discourages exactly the people who are needed to devote resources to PPS if it's going to start generating top 5000 primes again.
The powers that be should figure out how they want to address this issue. If people are being asked to devote resources to PPS so that the leading edge stays within the top 5000 primes list, the confirmed primes that have already been found in the range should be excluded. If they can't be excluded, warn people that even if the primes they find are within the range of the top 5000 list, they likely won't be new to that list.
So is the goal for PrimeGrid to find new primes, or are we just triple checking primes that have already been confirmed? |
|
|
|
|
Is PPS LLR still the LLR Top Priority?
Yes, it will remain a priority until n=500K is reached. Currently, the leading edge is just under n=497K.
Can we get primes that are already on the top 5000 list removed from the work queue? See this thread about the issue. I've found two primes this week for which I was the initial discoverer according to PrimeGrid, yet they had both already been submitted to the top 5000 list with PrimeGrid credited for the effort. If the goal is to find new primes, there's no reason to be wasting resources on numbers that are already known to be prime...
-JMP |
|
|
|
|
|
Once again, I've found a prime that was submitted to the top 5000 list before I found it, with PrimeGrid credited. This is the fourth prime for which PrimeGrid tells me I'm the initial finder, and the fourth that was already in the top primes database before I "discovered" it.
This has got to be fixed.
I'm currently crunching about 45-50 PPS WUs per hour. I understand that the purpose of PrimeGrid is to find new primes, yet PrimeGrid is sending out primes that are already known to be prime. Can someone give a compelling reason why I should bother with PPS if it isn't serving PrimeGrid's nominal function, or explain how checking primes that have already been proven furthers the goal of discovering new primes? In the absence of a compelling answer (which I've been asking for here for several days), I'm going to disable PPS and avoid that project in the future, and will encourage others to do so, as it does not fulfill the goals of PrimeGrid. |
|
|
|
|
|
Once again, I've found a prime that was submitted to the top 5000 list before I found it, with PrimeGrid credited. This is the fourth prime for which PrimeGrid tells me I'm the initial finder (all from PPS), and the fourth that was already in the top primes database before I "discovered" it.
This has got to be fixed.
I'm currently crunching about 45-50 PPS WUs per hour. I understand that the purpose of PrimeGrid is to find new primes, yet PrimeGrid is sending out primes that are already known to be prime. Can someone give a compelling reason why I should bother with PPS if it isn't serving PrimeGrid's nominal function, or explain how checking primes that have already been proven furthers the goal of discovering new primes? In the absence of a compelling answer (which I've been asking for here for several days), I'm going to disable PPS and avoid that project in the future, and will encourage others to do so, as it does not fulfill the goals of PrimeGrid. |
|
|
|
|
|
Obviously PrimeGrid Boinc is having us 'double-check' the some ranges (presumably from PRPNet) so any discoveries are new to PrimeGrid Boinc but old to the world. See John's message earlier in this thread:
Two sections have previously been tested: 497K-498K and 499K-500K. Therefore, that leaves 495K-497K and 498K-499K.
Just hang on as in a couple of days (hopefully) we will finish all the WUs with n<500K - my current WUs have n=499147. Then we should jump to a new range where PrimeGrid should be first. |
|
|
|
|
Obviously PrimeGrid Boinc is having us 'double-check' the some ranges (presumably from PRPNet) so any discoveries are new to PrimeGrid Boinc but old to the world. See John's message earlier in this thread:
Two sections have previously been tested: 497K-498K and 499K-500K. Therefore, that leaves 495K-497K and 498K-499K.
Just hang on as in a couple of days (hopefully) we will finish all the WUs with n<500K - my current WUs have n=499147. Then we should jump to a new range where PrimeGrid should be first.
But I had understood John's message to mean that those ranges had already been tested, therefore we would skip over those ranges. That is, because 497k-498k and 499k-500k had been tested, we would stop at 497k, test the range between those two ranges, then resume testing above the completed ranges. At least that's what I understand the section you quoted to mean. Can someone confirm that things were done that way, or have we been double-checking all of the previously completed work? |
|
|
|
|
|
PrimeGrid tells me I was the initial discoverer of a fifth prime today. The top 5000 Primes database tells me that PrimeGrid submitted that very same prime in May 2009.
I'll ask again, if the goal is to find new primes, why is PrimeGrid having users retest primes that PrimeGrid has already participated in discovering?
Why should anyone who wants to help discover new primes run PPS on their machines if these ranges have all been tested already? |
|
|
|
|
|
PrimeGrid tells me I was the initial discoverer of a fifth prime today. The top 5000 Primes database tells me that PrimeGrid submitted that very same prime in May 2009.
I'll ask again, if the goal is to find new primes, why is PrimeGrid having users retest primes that PrimeGrid has already participated in discovering?
Why should anyone who wants to help discover new primes run PPS on their machines if these ranges have all been tested already? |
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2649
ID: 1
Credit: 26,363,112
RAC: 0
               
|
|
We are working on solutions to ensure this doesn't repeat.
____________
|
|
|
|
|
We are working on solutions to ensure this doesn't repeat.
Has there been any progress on this? Now that the challenge is over, I'll rejoin the effort to get PPS moving forward, but only if I can get some assurances that if I find a prime it won't have been submitted to the top primes database months before the WU was assigned to me. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
 
|
We are working on solutions to ensure this doesn't repeat.
Has there been any progress on this? Now that the challenge is over, I'll rejoin the effort to get PPS moving forward, but only if I can get some assurances that if I find a prime it won't have been submitted to the top primes database months before the WU was assigned to me.
Measures have been established to minimize re-discoveries. However, this is not 100% as primes can be found by anyone. Nevertheless, the incidence of re-discoveries should decrease substantially.
____________
|
|
|
|
|
Measures have been established to minimize re-discoveries. However, this is not 100% as primes can be found by anyone. Nevertheless, the incidence of re-discoveries should decrease substantially.
It seems to have worked. I've been the first to find 3 reportable primes over the last week, and I've actually been cited as the first to find each of them in the Top Primes database. (Well, the third is still being verified, but it was only submitted about 20 minutes ago.)
Thanks...
-JMP |
|
|
|
|
|
My stats account info says:
Proth Prime Search tasks
Completed tasks 142579
Credit 822,599.19
Primes found 14
Percentage of tests resulting with a prime 0.01
But when I click on my prime page it only shows a total of 11. Are the other 3 dupes or pending some kind of validation?
____________
|
|
|
|
|
|
You will only see the number off primes found before they are submitted.
You will see them in primes when they are verified on TOP5000 page.
This can take 1-5 days.
Lennart
|
|
|
|
|
You will only see the number off primes found before they are submitted.
You will see them in primes when they are verified on TOP5000 page.
This can take 1-5 days.
Maybe this should be explained on the "Prime number list for participant" page, as this question appears regularly? :) |
|
|
|
|
You will only see the number off primes found before they are submitted.
You will see them in primes when they are verified on TOP5000 page.
This can take 1-5 days.
Maybe this should be explained on the "Prime number list for participant" page, as this question appears regularly? :)
Why make it so easy :d
Lennart |
|
|
|
|
My stats account info says:
Proth Prime Search tasks
Completed tasks 142579
Credit 822,599.19
Primes found 14
Percentage of tests resulting with a prime 0.01
But when I click on my prime page it only shows a total of 11. Are the other 3 dupes or pending some kind of validation?
I am working on them now. I hope you soon will see them. :)
Lennart
EDIT: This may take a day or 2 before all are submitted. |
|
|
|
|
|
yes, I see more showing up..... patience is a virtue. I guess I will have to keep finding them to keep you busy. :) Thanks for the info.
____________
|
|
|
