| Author |
Message |
|---|
|
|
|
Hi, folks...
I´ve bouht a new Phenom II X6 just a week ago... It is really a powerful CPU, I can crunch about 1000-1200 PPS LLR WUs/day (in case of running 24/7, of course).
My question is simple - how big is probability of finding a prime? Is 1 number of 5000 tested a prime? Or 1/6000, 1/10000, 1/20000?
I think you must have statistics like that ;-)
____________
 |
|
|
rogueVolunteer developer
Send message
Joined: 8 Sep 07
Posts: 1218
ID: 12001
Credit: 18,565,548
RAC: 0
 
|
Hi, folks...
I´ve bouht a new Phenom II X6 just a week ago... It is really a powerful CPU, I can crunch about 1000-1200 PPS LLR WUs/day (in case of running 24/7, of course).
My question is simple - how big is probability of finding a prime? Is 1 number of 5000 tested a prime? Or 1/6000, 1/10000, 1/20000?
I think you must have statistics like that ;-)
It depends upon the project you are computing against. BOINC PPS will have a lower ratio of primes than PRPNet port 10000. |
|
|
|
|
|
I see very few people find primes on PPS before they've done tens or hundreds of thousands of WUs. I've done close to 4000 and no primes.
PRPNet boasts better chances of finding primes with PPS extended, which I've yet to try.
At any rate, you've convinced me of my next CPU...
____________
 |
|
|
|
|
|
Hi everyone. Do not despair keep on crunching . I joined Primegrid and had a top 5000 Prime within 9 days after joining and on my Birthday and I am using an old CPU. Luck obviously has something to do with my find but one in 10000 seems to be a good target to use. So I am expecting one again in 10000 . |
|
|
|
|
Hi, folks...
I´ve bouht a new Phenom II X6 just a week ago... It is really a powerful CPU, I can crunch about 1000-1200 PPS LLR WUs/day (in case of running 24/7, of course).
My question is simple - how big is probability of finding a prime? Is 1 number of 5000 tested a prime? Or 1/6000, 1/10000, 1/20000?
I think you must have statistics like that ;-)
I think it would be hard to have good stats on that due to the sporadic nature of finding the prime and how "lucky" you are. However, as many wu's as you are cranking out, it would seem to me that you should be getting hits quite frequently (again, this ain't helping answer your question). On average I've been getting hits about every 2 weeks and I'm not running that many cores.
However, and this is truly selfish of me, keep crunching those pps's cause the more you get out of the way without a prime the better chance I have of getting lucky with my slow(er) pc's :) (sorry just had to throw that in)
____________
@AggieThePew
 |
|
|
|
|
|
An overclocked X6 certainly eats the short WUs for breakfast and will be starving by lunch :D
If it's any help, my all time stats on PPS LLR are about 1 prime in 6000 inclusive of double checks. While there is certainly a probability aspect to the rate, if you crunch enough it will start to average out. |
|
|
|
|
|
On the PPS LLR subproject I don't think it's much higher than about 1 prime every 4-5,000 WU's. Look at the stats of people with hundreds of thousands of credits and it seems to average around that. I seem to be above average with 5 primes in 12,000 WU's. |
|
|
Honza Volunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 0
                             
|
|
Done 677,998 tests and 176 primes (both prime and doublechecker) over the years, ratio is close to 1:4000.
Note it's "long average", includes both very old primes and recent ones that are just into Top5000.
PPS LLR is best way to find a reasonable prime (Top5000) and this subproject needs a push anyway.
btw, Ondra, merge your X6 hosts into single host.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
|
Thank you for all the answers, I hope I´ll have my first "top 5000" soon...
btw, Ondra, merge your X6 hosts into single host.
?! Damn... it says I can´t merge it becouse of some differencies...well, I can´t see any difference :D ...probably a kind of mistake. The other "host" has credit 0.
Edit : I did some OC - can it make some difference in BOINC system? I have still the same configuration.
____________
|
|
|
|
|
|
At n=500k I did several hundred thousands of tests. Average was 1 prime in about 7000 tests. Now it's probably closer to one in 10000.
____________
There are only 10 kinds of people - those who understand binary and those who don't
 |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 0
|
?! Damn... it says I can´t merge it becouse of some differencies...well, I can´t see any difference :D ...probably a kind of mistake. The other "host" has credit 0.
Edit : I did some OC - can it make some difference in BOINC system? I have still the same configuration.
Once there are no tasks under host, you can delete the host.
(can't be merge when their scheduler times are overlapping, even if they are HW/SW identical).
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 11,549,356
RAC: 0
                    
|
|
I have 2 PPS primes out of 10498.
ALL of the rest of my [BOINC] LLRs together have 0 primes (not including PRPNet PSA stuff of course).
____________
Murphy (AtP)
 |
|
|
|
|
|
Hehe, after two days of crunching, I can see first results :-)
2 hours ago, I had only one prime (double check) - now, there is shining a number "2"...so, I´m going to wait, if it´s my first prime, or just another double check.
____________
|
|
|
|
|
|
After ~12500 PPS-WU's without a prime i think someone is stealing my luck. Give it back please ;( luckily i still have patience, so i will wait a little longer.
Nitpicking: in this topic:
under "Other LLR Projects"
Prime Sierpinski Problem LLR (PPS)
Shouldn't that be PSP? Its confusing haven same names for different projects ;)
____________
 |
|
|
 NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
              
|
|
I got lucky; I discovered my first Proth prime after 87 tests. Go figure.
Neo
____________
 |
|
|
|
|
|
OK I must be lucky. I found another prime within 280 tasks! |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 11,549,356
RAC: 0
|
|
I apparently found two (2) PPSE primes last night. No info yet on them, and I will NOT search through the task Results! Hoping that I am Initial Finder, but I will settle for Double Checker.
;-)
____________
Murphy (AtP)
|
|
|
|
|
|
Murphy -
You are having quite the run of finds lately ... The luck 'o the Irish - you got it it - I don't!
Congratulations!
J. M. Murphy (TheDawgz)
(We'll cross our paws that you're the initial finder on them both)
____________
There's someone in our head but it's not us. |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 11,549,356
RAC: 0
|
|
Turned out to be the Double-Checker for both. Still, it is nice to find a fairly large prime occasionally. Until late last year, I had not found ANY primes at all!
____________
Murphy (AtP)
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 0
|
Turned out to be the Double-Checker for both.
Turned out that last 11 primes on my list was double-checker. Hmm...
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
|
Most of my early primes were double-checks until I found out that the credit goes to the first prime result returned. For short computing time units, it means you really need to turn them around as quickly as possible by running a very short cache and turn on report immediately in one of the boinc config files. By doing that, I was then mostly the 1st finder.
Personally I'd prefer the system be changed so that all valid assignees for the WU in boinc get equal share in the prime credit as co-finders. Then there would be no time rush and people can optimise longer caches and reduce load on the server by not hitting it so often. |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 0
|
|
Is 0.25 days of cache too much?
Did last 11 primes reported quite slow two quads that are on BOINC only during nights and weekends and other faster quad 24/7 missed all of them?
I'm not complaining...just curious.
Well, I would rather prefer that once a prime is found by first results returned, notification would be send right away to primes finder. Perhaps we can reconsider how long would it take before double-checker is contacted if prime-finder doesn't respond.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
|
You need to be faster than others out there, and I don't know how fast that is for what level of chance. But 0.25 day is a pretty long time compared to the unit length. I was running practically zero cache when seriously doing PPS or SGS to maximise my chances even if it meant I had idle crunchers during maintenance or other outages. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
            
|
|
mackerel wrote: I was running practically zero cache when seriously doing PPS or SGS to maximise my chances even if it meant I had idle crunchers during maintenance or other outages.
PPS (LLR) and SGS (LLR) are both using Adaptive Replication. Therefore, WU's are being generated with initial replications of 1 or 2. For single replications, you will always be the first to return it. :) If the results indicate that the WU is prime, then a second task is sent out to confirm the prime. However, for dual replications, it is true that the first host to return the result is credited as prime finder and the second as double checker.
The only solution that makes sense is to delay the sending of all dual replications until after the first result has been returned. But I'm not aware of that logic being offered in BOINC.
Honza wrote: Well, I would rather prefer that once a prime is found by first results returned, notification would be send right away to primes finder.
A WU is not declared prime until a second result confirms it. If an email could go out, all it would state is that you may have found a prime and that we're waiting on the DC to confirm it. But I'm not aware if an email being generated at this point is even possible. It all results in you being notified about the prime the same time you're being notified now.
____________
|
|
|
|
|
|
How long has adaptive replication been in use here? Without using short caches I seem to get far more double checks. |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
How long has adaptive replication been in use here? Without using short caches I seem to get far more double checks.
Since the beginning of the SGS (LLR) project and since June 2010 for PPS (LLR).
____________
|
|
|
|
|
|
Thanks, looks like I was running on incomplete or outdated assumptions in my crunching strategy :) It was mostly my earlier PPS tasks I was basing this on, since I didn't do much SGS with the false prime finding lull that was there previously.
I will certainly turn off report-immediately and run bigger caches as and when I return to these small WU projects, going for big unit work at moment so doesn't make much difference.
Forgot to ask, is this also enabled for other LLR subprojects? |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Forgot to ask, is this also enabled for other LLR subprojects?
Only SGS (LLR), PPS (LLR), and TRP (LLR) use Adaptive Replication.
____________
|
|
|
|
|
|
Thanks again. I have been dabbling in a bit of TRP recently so that's good news. Hoping to get my 3rd one ever! |
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 0
|
|
This host has ~2500 tasks in DB, all of them valid (~2300) or pending (~150) or a few in progress; none invalid, errored or aborted.
I have clicked many WU, all of them are with initial replication of 2, even this host is HIGHLY reliable, running 24/7, fast turn around etc.
Is adaptive replication applied to PPS LLR Anonymous platform (CPU) as well?
This would save considerable overhead - and PPS LLR need it desparetaly.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
|
|
|
I've been doing this for less than a month (despite my join date) and I think I found my first prime!
> Dear Primefinder,
>
> Congratulations! Our records indicate that a computer registered by you has
> found a unique prime number. This computer is running BOINC, is attached to
> the PrimeGrid project, and is assigned to the Proth Prime Search. What makes
> this prime unique is that it's large enough to enter the Top 5000 List in
> Chris Caldwell's
> The Largest Known Primes Database.
>
> Since you have auto-reporting selected, the following prime was submitted on
> your
> behalf:
>
> Added 97833 : 8115*2^617718+1 (185956 digits)
>
> If you have any questions or concerns, please contact me and we will surely
> resolve any
> problem.
>
> Once again, congratulations on your find. Thank you for participating in
> PrimeGrid.
>
> John Blazek of PrimeGrid
____________
|
|
|
|
|
|
I just started for the challenge and got the same here. I am using a AMD phenom 2 X4 955
Dear Primefinder,
Congratulations! Our records indicate that a computer registered by you has
found a unique prime number. This computer is running BOINC, is attached to
the PrimeGrid project, and is assigned to the Proth Prime Search. What makes
this prime unique is that it's large enough to enter the Top 5000 List in Chris Caldwell's
The Largest Known Primes Database.
Since you have auto-reporting selected, the following prime was submitted on your
behalf:
Added 98053 : 1757*2^625647+1 (188342 digits)
If you have any questions or concerns, please contact me and we will surely resolve any
problem.
Once again, congratulations on your find. Thank you for participating in PrimeGrid.
John Blazek of PrimeGrid |
|
|
ktoSend message
Joined: 13 Jan 09
Posts: 27
ID: 34164
Credit: 41,469,365
RAC: 0
           
|
|
Here's my situation.
I've crunched about 250k PPS LLR credit in the past couple months - and have found only 2 primes! That's easy math - 1 prime in 125k work units.
Prior to this, I had found 50 PPS primes in 750k credit. That's 1 prime every 15k wu.
I can see th Proth primes are slowly getting bigger - they're around 196k digits - but I didn't think the probability would shoot down so quickly?!
My PPS LLR probability has even gone below my SGS LLR probability!
Any thoughts? Am I just having a dry spell?
|
|
|
HonzaVolunteer moderator
Volunteer tester
Project scientist
Send message
Joined: 15 Aug 05
Posts: 1893
ID: 352
Credit: 3,142,312,174
RAC: 0
|
Any thoughts? Am I just having a dry spell?
If I can see corretly, you also have ~3 primes as double-checker.
Still low number of primes comparing to test done.
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
ktoSend message
Joined: 13 Jan 09
Posts: 27
ID: 34164
Credit: 41,469,365
RAC: 0
|
|
Ok, so I have 5 PPS in 250k wu - thanks.. That now works out to 1 prime in 50k wu. Again, compared to 1 prime in 15k wu - the ratios are way off.
Right?
BTW, this might be a dumb question, but how did you know I had gotten 3 double-checks in the past couple months?
thanks. |
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 11,549,356
RAC: 0
|
Ok, so I have 5 PPS in 250k wu - thanks.. That now works out to 1 prime in 50k wu. Again, compared to 1 prime in 15k wu - the ratios are way off.
Right?
BTW, this might be a dumb question, but how did you know I had gotten 3 double-checks in the past couple months?
thanks.
Click on the "Primes Found" number under the sub project in "Your account".
Or use (substitute your own id): http://www.primegrid.com/primes/?section=primelist&userid=30360
____________
Murphy (AtP)
|
|
|
ktoSend message
Joined: 13 Jan 09
Posts: 27
ID: 34164
Credit: 41,469,365
RAC: 0
|
|
Thanks for the response - I am familiar with the "Primes found" link. However, it doesn't have a date associated with my prime. That is how I was curious Honza was able to figure out how I had 3 double checks in the past couple months. Cuz if you look at my "primes found" stats, I've got a couple dozen+ double-checks on PPS LLR:
http://www.primegrid.com/primes/?section=primelist&userid=34164
So to restate, is there a way to see the date a prime was found? |
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2165
ID: 1178
Credit: 8,777,295,508
RAC: 0
                                   
|
...So to restate, is there a way to see the date a prime was found?
Copy the prime into the search box on the Top 5000 page and it will give the link to the prime as reported there (which should include the date found for the initial finder...same or later date for double checker).
____________
141941*2^4299438-1 is prime!
 |
|
|
|
|
|
not sure if I am getting lucky, but My computers have found 3 primes (plus 3 double checkers) in the last 2 weeks..
|
|
|
ktoSend message
Joined: 13 Jan 09
Posts: 27
ID: 34164
Credit: 41,469,365
RAC: 0
|
not sure if I am getting lucky, but My computers have found 3 primes (plus 3 double checkers) in the last 2 weeks..
Wow - that's incredible - how many PPS credits did you have to put in for that?
Well, after putting in about 300k credit on PPS alone and finding 5 PPS, I'm switching over to SGS. Maybe I'll have more luck there.
|
|
|
|
|
|
Finally after many moons, a prime and the notification was rather quick. A hit and an email on the same day. I suppose it's because it will fall out of the top 5000 in the next few hours.
1561*2^656304+1 - plus it looks like we are working well towards the 666666 wall.
Rick
____________
@AggieThePew
|
|
|
|
|
not sure if I am getting lucky, but My computers have found 3 primes (plus 3 double checkers) in the last 2 weeks..
Wow - that's incredible - how many PPS credits did you have to put in for that?
Well, after putting in about 300k credit on PPS alone and finding 5 PPS, I'm switching over to SGS. Maybe I'll have more luck there.
i am at 534k credit and found 13 pps primes. my percentage is sitting at .0153
my 2 computers (10 cores) is finding an average of 2 per week right now.
edit: goofed again.. |
|
|
