Author |
Message |
|
This isn't a complaint but just wanting to understand what I'm missing here. Assuming various rates hold as noted below it means I have about 22 days to go to PSA gold. That's fine since the 5oB double check has 100+ days remaining.
In another thread someone estimated that a PSA silver badge at 200,000 BOINC credits is approx 3,600,000 PRPNet credits.
I received my bronze PSA badge at 20,000 BOINC credits on 2011-04-21 and my silver on 2011-04-27. My systems on PRPNet have remained fixed almost since the start and I didn't even pull them off for the Cinco de Mayo challenge.
I'm confused by the conversion of my credits (I check daily so the PRPNet number should be fairly close):
PSA 393,948 BOINC 8,412,698 PRPNet
PSA 506,156 BOINC 10,787,620 PRPNet
PSA 597,564.27 BOINC 19,905,159 PRPNet
So my PRPNet credit (almost exclusively on the 5oB double check) nearly doubled (1.845x) between the BOINC updates but this resulted in only a 1.18x increase. I quite mistakenly thought the next update would have gotten me to a gold badge (or very near it). Is this normal? The more PRPNet credit you have, the less it counts? The rate for the 393,948 to 506,156 jump is 1.285x and on the PRPNet side it's 1.282x.
There was a remark also in another thread about having a RAC over 200,000 will decrease the credit in the conversion. From my free-dc stats it appears 3 days in the last month I exceeded 200,000. Did those 3 days translate to the conversion going from around 933,857 BOINC to 597,564? I believe only 1 of these days is actually falls in the current credit assignment
If that conversion rate holds then I need 33,310,505 PRPNet credits for PSA gold.
If I take a look at another user, kimmyl, on PRPNet. He has 43,262,292 PRPNet credits and 1,896,258 BOINC PSA credits. However, if my conversion rate stays the same I need 63,165,312 PRPNet credits to get the same BOINC PSA credits as him.
Also, if I look at the PSA rankings and look at the highest silver badge user, Neo, then it gets even more confusing. He has 980,432 BOINC PSA credit and 14,559,432 PRPNet credits.
Additionally, in other threads there is talk of the credit trickling in. Does this mean that only some of my PRPNet credit actually was converted to PSA BOINC credit? Some of my credit is held hostage so to speak until the next update run or is filtering in a bit over the next few days?
What am I doing right and wrong both from a crunching standpoint and from my numerical assumptions on conversions for credit? |
|
|
|
Boinc PSA credits come from different Project Staging Area sub-projects of which PRPNet is one.
1PRPNet credit equates to approximately 0.18-ish Boinc PSA credits.
Other PSA sub-projects (both current and long gone) have/had different ratios of PSA credit to work performed.
Hope that helps
____________
There's someone in our head but it's not us. |
|
|
|
Yeah they seem to vary vastly....
I have approx 500k PRPNet FPS which is 24k BOINC credits.
Peter
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
|
EDIT - sorry not 0.18 as originally stated - should have been 0.05-ish.
We were thinking about getting a belly rub ....
Boinc PSA credits come from different Project Staging Area sub-projects of which PRPNet is one.
1PRPNet credit equates to approximately 0.18-ish Boinc PSA credits.
Other PSA sub-projects (both current and long gone) have/had different ratios of PSA credit to work performed.
Hope that helps
EDIT - sorry not 0.18 as originally stated - should have been 0.05-ish.
We were thinking about getting a belly rub ....
____________
There's someone in our head but it's not us. |
|
|
|
....... What am I doing right and wrong both from a crunching standpoint and from my numerical assumptions on conversions for credit?
Treat the granting of PRPNet & BOINC credits as two separate processes which are only connected by a (roughly) conversion rate of 18 PRPNet:1 BOINC.
The PRPNet PSA credits accumilate according to whatever rate is set for the associated Project(s). At a point in time, John kindly does a manual conversion. That conversion is of whatever credits that has accumilated in PSA, and would usually do it every 2-4 weeks depending on a number of factors, not least time to administer it.
However there is a big leveling factor, no matter how many PSA credits have accumilated he makes sure the resultant PSA credit is not out of wack with a "normal" (define normal ....) set of BOINC credits for a CPU. So on a long run of work he is careful to make sure (for example, and exagerating to illustrate) that tens of thousands are not added for a core no matter how many PSA have accumilated, if he didnt, lots of moaning would ensue re over paying BOINC credits.
So he looks at the work done in the PSA world, and converts as much as he can for that day(s) comensurate with work for those days only - that stops short term over credit on the BOINC side.
Example .... when I was recently doing 5oB full tilt, before stopping for a couple of weeks for Pentathelon, I was getting a number of PSA credits converted and credited every couple of days. It all came out in the wash in the end, and the times when there was some still left over was fine, couple days later they got converted - all at the rate of 18:1
I've just restarted 5oB, and therefor took a couple of days to rebuild the old rate of work, and for a few days, no conversions, thats fine as daily work was still building. I am back up to full rate of work now, and the PSA credits are converted every two days. Working fine.
The way to think about it is accumilate PSA according to the credit rules of that application. Once the PSA credits flow, then they are converted to give a normal flow of BOINC credits. If there is a feeling more PSA credits are needed, fine, but thats nothing to do with the conversion - the latter just converts the final outcome at a rate of 18:1, comensurate with not giving too much for Cores (ie not a rate that would be impossible for a Core for a day if solely crunched a BOINC project).
Geez that was wordy wordy, sorry, hope I have not talked myself into a corner rofl :)
Regards
Zy |
|
|
|
Just wondering why PRPNet and BOINC credits are not the same? Surely it would make porting the credits from one to the other easier!
Peter
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
pschoefer Volunteer developer Volunteer tester
 Send message
Joined: 20 Sep 05 Posts: 674 ID: 845 Credit: 2,561,995,841 RAC: 372,168
                           
|
Just wondering why PRPNet and BOINC credits are not the same? Surely it would make porting the credits from one to the other easier!
BOINC credits are based on the BOINC benchmark, PRPNet doesn't know anything about that stuff. If anyone found a formula to calculate fair LLR credits, it could be used for both.
____________
|
|
|
rogueVolunteer developer
 Send message
Joined: 8 Sep 07 Posts: 1249 ID: 12001 Credit: 18,565,548 RAC: 0
 
|
The way PRPNet computes the score is both fairly simple and pretty fair at the same time. The score is based upon the number of decimal digits of each number tested which can be used to determine how much longer one PRP test will take than another. For PRP tests the rule is fairly close to this: double the number of digits, then quadruple the time for a PRP test. It isn't exact, but it close enough.
By doing this I eliminate the need to know about the hardware that the software is running on. I also eliminate looking at the time it takes to do an individual test as that can be influenced by many factors. Other projects have done something akin to this which typically introduces a lot of unfairness. It tends to punish people with faster computers.
Keep in mind that BOINC tries to level out credits across all projects, even though they do very different types of computations. PRPNet is not (and will never be) integrated with BOINC. It has a different means of keeping score. The important thing is that the conversion is accurate. Since BOINC has a PPS LLR project and PRPNet has a PPS LLR project I have to believe that the conversion is based upon comparisons to how these projects compute for a score for each completed test. |
|
|
|
It seems I'm in the trickle category:
PSA 506,156 BOINC 10,787,620 PRPNet
2-3 weeks later
PSA 597,564.27 BOINC 19,905,159 PRPNet
24 hours later
PSA 706,530.70 BOINC 20,455,295 PRPNet
The trickle system makes a lot of sense but it's hard to know that's what is happening. If all 2+ weeks of PSA credit were granted in 1 day it may raise the RAC to a rather high level. I wonder if the trickle credit could go into PSA pending or something similar. (I know you can't just easily throw credits that don't need validation and are WUs from a totally different system into the pending state for BOINC.) |
|
|
|
BOINC credits are based on the BOINC benchmark, PRPNet doesn't know anything about that stuff. If anyone found a formula to calculate fair LLR credits, it could be used for both.
How fair the BOINC scoring system is a debate unto itself. For instance the way to really get your RAC up in any project that supports GPUs is to have even 1 of them run full tilt and no one really with any number of CPU cores can touch you.
The scoring/credit system isn't hard to find for BOINC. Simply grab the source, visit the sched directory and look at the sample_validator.cpp which will lead you to calls in credit.cpp for assign_credit_set and grant_credit.
I don't know what custom work was done to the validator or credit granting in this project but it would have to be fairly extensive to take into account things like FFT size. |
|
|
|
It seems I'm in the trickle category .....
You are definitely more than that with the WUs you have running now. Its been reflected in your frequency of conversion already. If you revisit your PSA credits you'll see the frequency of your PSA conversions increasing fairly rapidly now. Its likely to be every day/every two days now you have the steady work rate that you have, and the conversions step up to meet your higher work rate. With large contributions over a period of time like yours, it will not be a case of waiting 2 weeks or so, as you will now be seeing starting to happen.
EDIT:
This stats table may help your visibility on PRPNet movement:
http://u-g-f.de/PRPNet/user_search?uid=quel
http://u-g-f.de/PRPNet/
Also I find multiplying the PSA BOINC credits given already (mouse over avatar) by 18, then looking at Stats, you can see whats left over still to come at any time. Click the blue links as you go along and follow your nose - lots of PRPNet stats tables that will help you.
Regards
Zy |
|
|
|
Thanks for the tips on the 18 multiplier. I just got another update today on the BOINC side from 706,530.70 to 833,549.24. If you are correct then I already have enough PRPNet credit for gold and I just have to be patient. At least I'm not at the mercy of wingman for the various LLR projects that I'm working on badges for. I started my first Seventeen or Bust WU and have seen users who through timeouts, aborts, etc. have been waiting over 2 months for validation. I guess 2 weeks isn't so bad :) |
|
|
|
Well, it all worked out and I must say these badges are addictive. Gold PSA! |
|
|
|
Well, it all worked out and I must say these badges are addictive. Gold PSA!
Congratulations =D
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
|
Well, it all worked out and I must say these badges are addictive. Gold PSA!
Great Stuff :)
With the ever increasing hardware abilities, it cant be that long now before the 10m Badge is introduced - that'll start a mass feeding frenzy rofl :)
Regards
Zy |
|
|
|
With the ever increasing hardware abilities, it cant be that long now before the 10m Badge is introduced - that'll start a mass feeding frenzy rofl :)
Yes for the two CUDA enabled sieves (and the one ATI as well) going from nothing to ruby with a recent GPU is fairly short work. I have only a single GPU (GTX 470) which is 1 generation behind and not the top model in that line. No problem cranking out 150k credit a day. Though, GPU v CPU is in general a problem for BOINC projects and there is a general lack of balance. That card cost me about $250USD almost a year ago and compared to the number of systems and procs to advance without CUDA is much more costly for hardware, power, cooling, time, etc.
Though, for Sieve projects: 20 / 200 / 1000 / 2000 / 4000 thousand credits for badges. So, *10, *5, *2, and *2. Surely the next jump would need to be 8m ;) going to 10m is a *2.5 jump and that just won't do since we much prefer 2, 3, and 5.
If I remember correctly then the last badge addition also had a bit of a re-balancing of the credits on the existing badges as well. |
|
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 17,239,060 RAC: 37,915
                     
|
Could someone please check my PRPNet to PSA credit conversion.
http://u-g-f.de/PRPNet/user_search?uid=mfbabb2
I have not received any credit since the Challenge, and by my calculations, I should be over 200,000 Boinc credits PSA by now (/18 = 209,949).
Is it possible that I have that much PSA credit still pending? Or did something get lost when the DB gliched?
Thanks to anyone who has any insight.
WAG'ers may also comment or put in their 2 cents worth... ;-)
____________
Murphy (AtP)
|
|
|
|
As always with PRPNet --> BOINC conversion, its just a case of being patient and waiting.
According to PRPNet stats I should be sitting on 36k BOINC credits, so it is definitely behind a little as expected.
Peter
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
|
When PSA first got going there was a delay of more than a month. Lately the credits have been within a day or so, which is actually amazing. Maybe we have become spoilt?
____________
Warped
|
|
|
|
Curious .... I've only ever been in one PRPNet Project, the current 5oB marathon, which is PSA I'm clear.
What constitutes a PSA credit Project, all PRPNet (which would seem strange) or more likely the ones John declares as a PSA credit Project?
I'm still relatively new to PG no doubt as my statement above shows to more seasoned members .... I have a strong feeling the PSA credit bit is when John declares it so, but my definitive ignorance is bugging me ....
Anyone care to get rid of the annoying itch I have about it :)
Regards
Zy |
|
|
|
As far as I am aware, all PRPNet projects contribute to PSA BOINC credits.
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
Scott Brown Volunteer moderator Project administrator Volunteer tester Project scientist
 Send message
Joined: 17 Oct 05 Posts: 2329 ID: 1178 Credit: 15,588,524,304 RAC: 13,396,289
                                           
|
All PRPNet projects at PrimeGrid receive BOINC credits (eventually) that are part of the PSA subproject for the purpose of badges. No BOINC subprojects are part of PSA. Additionally, there are (and have been) several other non-BOINC and non-PRPNet endeavors (mostly manual sieves) that also count for PSA credit/badges.
____________
141941*2^4299438-1 is prime!
|
|
|
|
All PRPNet Projects receive PSA credits.:
PRPNet
PRPNet Projects
So do all PSA Sieve Projects:
PSA
PSA Sieve Projects
Hope that helps.
____________
Warped
|
|
|
|
Thanks guys, appreciated - itch duley disposed of :)
Regards
Zy |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
I have not received any credit since the Challenge, and by my calculations, I should be over 200,000 Boinc credits PSA by now (/18 = 209,949).
It's not possible to apply a conversion number to PSA credit. Between sieve, PRPNet, CPU and GPU work, there's too many variables. However, if all you're doing is PRPNet work, then the number is ~20.
____________
|
|
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 17,239,060 RAC: 37,915
                     
|
I have not received any credit since the Challenge, and by my calculations, I should be over 200,000 Boinc credits PSA by now (/18 = 209,949).
It's not possible to apply a conversion number to PSA credit. Between sieve, PRPNet, CPU and GPU work, there's too many variables. However, if all you're doing is PRPNet work, then the number is ~20.
Thanks.
In that case, I am caught up and need to run additional PRPNet stuff to get PSA silver.
____________
Murphy (AtP)
|
|
|
|
Warped
All PRPNet Projects receive PSA credits.:
PRPNet
PRPNet Projects
So do all PSA Sieve Projects:
PSA
PSA Sieve Projects
Hope that helps.
How do you add PSA to your account?
____________
The art of flying is throwing yourself at the ground and missing.
|
|
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 17,239,060 RAC: 37,915
                     
|
Warped
All PRPNet Projects receive PSA credits.:
PRPNet
PRPNet Projects
So do all PSA Sieve Projects:
PSA
PSA Sieve Projects
Hope that helps.
How do you add PSA to your account?
Read this thread: http://www.primegrid.com/forum_thread.php?id=1215&nowrap=true#13329 PRPNet.
____________
Murphy (AtP)
|
|
|
|
Brilliant. Exactly what was needed.
Thanks for the help mfbabb2
____________
The art of flying is throwing yourself at the ground and missing.
|
|
|
|
Since about 10 days I am not getting any PSA points anymore...
Do others have that problem as well or is it just me?
In the latter case, might there be a problem with my account settings?
(Which would be strange since -to my knowledge- I didn't change anything..) |
|
|
Honza Volunteer moderator Volunteer tester Project scientist Send message
Joined: 15 Aug 05 Posts: 1931 ID: 352 Credit: 5,704,569,559 RAC: 1,050,473
                                   
|
Credit is added in batches.
(every couple of weeks score from PRPNet is transformed into credit, which is added to your account in couple of days to prevent RAC going totally crazy)
____________
My stats
Badge score: 1*1 + 5*1 + 8*3 + 9*11 + 10*1 + 11*1 + 12*3 = 186 |
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
Since about 10 days I am not getting any PSA points anymore...
Do others have that problem as well or is it just me?
In the latter case, might there be a problem with my account settings?
(Which would be strange since -to my knowledge- I didn't change anything..)
Please see Summer hours and The Project Staging Area.
____________
|
|
|
|
Not sure if everyone is in the same boat but I received credit last night.
____________
@AggieThePew
|
|
|
|
I recived some credit too, not mutch though. |
|
|
|
I also got credit last night. Another update and I should finally reach Amethyst. |
|
|
|
Woo! New shiny PSA badge :) |
|
|
|
Congrats there quel
____________
@AggieThePew
|
|
|
|
When was the last time when PSA credits was addet to BOINC?
____________
|
|
|
|
When was the last time when PSA credits was addet to BOINC?
last time I received PSA credit was a little over 10 days ago.
As I understand it these gaps are to be expected as credit is added manually. |
|
|
|
It would help if somewhere on the site there is a mention of the last date on which PSA credits were transferred. |
|
|
|
I have been running PSA WUs since about the middle of July but have not received any BOINC credit for them. I know its summer and people are busy with lots of different things, but I just wanted to be sure that I have everything set up right and that eventually i will see credits on my boinc account from it. |
|
|
|
It is summertime and all are not at home.
You will get credit but it might take som more time then it use to do.
Lennart
|
|
|
|
Alright that's what i was thinking but i wanted to be sure. Thanks Lennart |
|
|
|
Credits in a day ..... lol yeah right.
I am running it for about a week or so and nothing showed up on the free-dc psa primegrid stats
Not even my name shows up.
I even ran it some 20 hours or so on my cuda cards but took them off quickly cause 10k for 2 cuda cards for more then a day is kinda .... nothing
I think one of the cards did not do any work even though i had started the 2 cards start file.
If i would calculate the 18/1 ratio i would have made in 5 or 6 days about 24k on credits... i make more in less then 30 minutes with the cuda devices on other projects
To compare something i get after about 22 hours run time psa/boinc 10,000/555 credits on a gts450 card with 2 core cpu active as well.
So i turned them quickly back to better giving projects, then i tried running 3 machines with in total 9 cores, that did speed up a little but after some hours running i saw even this does not really make it worth while.
So i turned back to now 2 cores running on seperate machines but these psa points seems slow as hell and no points showing up in a for me understandable stats page makes it not my favorite
In boinc most good projects make a calculation for credits based on processing power done in the projects.
Ofcourse there are many issues with credit awarding since we have lots if people who are for a fair system based on the M/G flops spend on the projects, and other seem favor more to put it into some weird low credits system.
These people tend to say well its nice you use a more powerfull tool but that should not give more points .....
At some points people even was saying gpu projects do the same work so they should get the same points.
They simply disregard that a 100 till 350 watt device is pumping calculations like mad through it, and pumps up electra bills
Hell some persons on boinc spend enough electricity on this to feed an entire village for a full day on their electra usage.
So yes its hard to say what is a fair credit system but for me collatz is the one which seems to have the most fair rewarding system. |
|
|
|
uBronan, while it's true that it appears the credits given to PSA from prpnet are not up to what you might think is fair or right, there are lots of primes to be found. There are several projects including the gpu ones you ran as well as some really fun ones on the lower end that run fast and can be addictive to watch as well.
You can actually accumulate quite a few PSA credits in a short period of time if you want.
Just my 2 cents worth.
____________
@AggieThePew
|
|
|
|
lol
well with my luck, i think i have more change on finding a 5 euro bill on the street then to find a prime.
I see people having found prime in a few units but for me it probably won't happen in the next months or years and if i do find one it probably will be just to verify another.
Please do explain how to get the best psa credits because i have no idea what i am doing :D
Ofcourse i am somewhat limited since i try to get other badges from boinc cpu projects as well |
|
|
|
I run normal boinc at the same time as Project Staging, the same as you are trying to...
I have found that while the credits for the normal boinc projects stay the same, or slow down just a tad, the Project Staging is a bit slow, BUT, as far as I'm concerned, anything they give is a bonus, as it's credits you would not have got otherwise. Surely you don't expect huge extra credits with no spare processing power?
If you want Project Staging to go much quicker and give higher credits, you should not run boinc on those cores.
Oh, and if you want to find a Prime, you'll need to run LLR projects rather than Sieve ones!
____________
|
|
|
|
Well i just do not want to spend much power on this project.
Even with 9 cores on 3 machines and 2 cuda cards i saw such low rewards after running 28 hours, that the choice was not hard to switch back to only 1 core each.
To give an example i make easily 600,000 cobblestones on boinc in the same time, and probably even more if i was not trying to get some cpu projects to nr.1 like leiden.
After more then a week i received in boinc psa 13,000 cobblestones this includes the 28 hours running on all cores and gpu's and this without running boinc. And the rest of the time between 2 and 3 cores full on prpnet client.
Another reason is also because there are a lot of other projects where i am still not country number 1 or on signup date ;)
I do however set my goal to get some more badges, so when i reach them i probably need more power to stay on top on my other projects.
Ofcourse i have not much knowledge about this search for primes, i know many projects are trying to find the biggest including prime95 from mersenne which i ran some years ago (and ran some hours last month but i fear the $50,000 is not gonna be payed anytime soon :) )
Btw on prime95 i also never found a first prime in about a year or so.
At present its also not possible to run some prpnet projects cause they simply seem to be out of work because of some challenge.
But no i have several issues if i run the prpnet client along with normal boinc, especially when together with for instance einstein or some other gpu projects which also need a free core to run well
That makes it almost a day time job to toggle between projects and the ones who get into problems when something else runs along.
Some projects slow down almost to double/triple the normal time needed or in worst case scenario crash or stall.
I try to avoid running those projects together with those which do not like each other much.
Sadly that makes you have to choose what you should run instead of run them all together and let boinc choose. |
|
|
|
PSA credit conversion is a manual process that happens about once every 3-6 weeks. It's tied in partially with your BOINC RAC so they space the credit out such that your RAC isn't overly inflated by 1 large dump. (Yes I've learned a bit on this since my post in May.) Here is an example from my PSA credit conversions going on now:
2011-07-17: PSA 2,296,195.00 BOINC 46,584,083 PRPNet
2011-09-03: PSA 2,340,123.91 BOINC 70,333,508 PRPNet
2011-09-04: PSA 2,398,574.27 BOINC
2011-09-05: PSA 2,475,784.65 BOINC
2011-09-06: PSA 2,573,584.46 BOINC
2011-09-07: PSA 2,687,464.22 BOINC
2011-09-08: PSA 2,819,248.07 BOINC
2011-09-09: PSA 2,956,826.76 BOINC
2011-09-10: PSA 3,126,374.23 BOINC
2011-09-11: PSA 3,332,624.04 BOINC
2011-09-12: PSA 3,510,597.00 BOINC
I don't include the PRPNet values because I am still getting a back log of credits for the 70mil PRPNet I had on the 3rd.
PRPNet shows you with 397,394 credits which is approximately 71,530 BOINC credits and will net you a bronze badge when the credit transfer happens.
If you want instant gratification in the form of credits then PSA is not the place to get it for your BOINC credits. As someone pointed out the 2 sieves you are running here have no chance of finding primes, merely factors. (If you want to count factors then I have 43935 - almost entirely from the PPS Sieve with a single GPU.) In fact GPUs won't help you find primes in most cases and the majority of prime searching is still CPU only.
Almost all of my PSA work is on 5oB which is a double check project, at this stage, with almost no likelihood of finding primes but such work is essential. There are several PSA projects that in fact have a high chance for prime finding. If you take a look at this stats page then it appears that currently PPSElow is a good place to go for finding primes as it seems almost daily primes are found.
I've done a bit of badge chasing and have 2 primes to my name, both from PPS, and both from early this month, though only 1 from the challenge:
7897*2^691994+1
2525*2^693055+1
Oh and I ran prime95 for many years as well as some other prime projects well before any BOINC prime projects and those 2 I listed above are it for me. Of course finding primes is neat but |
|
|
mfbabb2 Volunteer tester
 Send message
Joined: 10 Oct 08 Posts: 510 ID: 30360 Credit: 17,239,060 RAC: 37,915
                     
|
In order to keep from slowing CPU down (switching back and forth between too many tasks), you should decrease the # (or percent) of cores to use in Boinc when running PRPNet (one for one).
Example: 4 core CPU = 50% use Boinc + 2 PRPNet tasks running = 100% CPU used.
____________
Murphy (AtP)
|
|
|
John Honorary cruncher
 Send message
Joined: 21 Feb 06 Posts: 2875 ID: 2449 Credit: 2,681,934 RAC: 0
                 
|
In fact GPUs won't help you find primes in most cases and the majority of prime searching is still CPU only.
While true, here is some info on where the GPU can be used to find primes:
- Generalized Fermat Number Prime Search - using geneferCUDA, the GPU is far superior to the CPU for both N = 262144 and N = 524288.
- Almost all other projects - using llrCUDA, the GPU is best at significantly higher n's. For example, in the PSP and SoB projects, the GPU outperforms the CPU. llrCUDA can be used in PRPNet or in BOINC. However, it is not yet a standard application in BOINC so you'll have to have an app_info file to use it.
Additionally, GIMPS is using the GPU for trial factoring (mfaktc) and Lucas Lehmer testing (CUDALucas).
____________
|
|
|
|
Yes, I'm aware of some progress on the GPU front but my last look at it indicated a very mixed bag: specific ll/llr and at specific values and generally not speeds of 25x+ that is common in a CPU -> GPU move. ll is essentially rounds of multiply, subtract, and mod - in x86/GPU architecture the mod is effectively a divide and the most expensive part of the computation. Given that the results of the previous round are a dependency to the next round that limits the vectorization you can do.
It seems the trial factoring app you mention follows a similar pattern and Montgomery multiplication is applied but they are showing 5-6x performance compared to a single fast cpu core. I don't have time to do the math but wouldn't be surprised if in terms of power and cooling it is still more expensive to use the cuda side than the cpu side for performance gains that small. |
|
|
|
I got some usefull information from your responces
At certain point while i was expecting to see the cuda client to be working with the cuda client noticed that it was switching between the different versions.
I did not see the genefercuda work but mostly saw the genefx64 and later the genefer80 application in process explorer.
Ofcourse i am not constant watching to what it does just noticed it when i had to stop a proces or was doing some test.
Should i change the stop option to delete all previous work, since it seems to finish well but get rejected by the server.
On the conversion to boinc i have an idea of what to expect now, that it takes a while i do not mind.
Today i restarted the client on my q6600 it started the unit it previously downloaded and did finish it completely
But now i followed the log to see something i did not notice before and could have been a reason why i seemed to have a low output.
What would be the reason for the job not being accepted as valid.
Since i am also alpha/beta tester in many different projects including os/software/game/boinc which sometimes causes my system to reboot or crash.
Most of the time when i going to test i stop the current running job by using the good old ctrl+c combination. In almost all cases i wait till it finished the job and shows downloading a new job.
But ofcourse sometimes when the system freezes a reboot is inevitable. |
|
|
|
Most often that error message is due to you returning the test beyond it's deadline. You can see the deadlines for your current tasks on the port's pending tests page e.g. for SGS - http://prpnet.primegrid.com:12000/pending_tests.html - all WUs have a deadline of 24 hours, so if you are part-time crunching, you may have trouble meeting some of these deadlines. You can set in your master_prpclient.ini file the number of WUs you want to download at a time from each port - probably best to set this to 1 in your case. Don't forget to run the approriate *update* script after editing the file!
Cheers
- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
Generalized Fermat Number Prime Search - using geneferCUDA, the GPU is far superior to the CPU for both N = 262144 and N = 524288
A while ago, I tested genefercuda on 262144 and 524288 in prpnet. In my case, 262144 genefercuda fall back to Genefx64 because limit was reached. Only 524288 worked with genefercuda.
Regards Odi
____________
|
|
|