## Other

drummers-lowrise
 Advanced search

Message boards : Proth Prime Search : PPSE occasional WU has ~75% of expected runtime & credit

 Subscribe SortOldest firstNewest firstHighest rated posts first
Author Message
River~~

Joined: 17 Mar 07
Posts: 342
ID: 6533
Credit: 15,792,075
RAC: 0

Message 105042 - Posted: 15 Feb 2017 | 12:05:36 UTC
Last modified: 15 Feb 2017 | 12:13:08 UTC

I have noticed that there is a trend of slowly rising runtimes and credits on PPSE. For WU issued earlier today credits are 43.44 as shown on website, and these rounded values have been notching up by 0.01 roughly every 36 hours with the rate of progress during TdP. Other than this steady persistent rise there is rarely any jitter in the points awarded, nor in the runtimes on a given computer at a given config.

I understand that is due to the gradually increasing n values, or to put it another way the longer the candidate the more time it takes to check. That's not my question.

What I do not understand is that against this steady gradual rise there is an occasional task that completes in about 75% of that time and with its credit pro rata. For example this WU was awarded 32.79

Is this an occasional fortuitous set of values that fit better than usual into a smaller FFT? Or is it an occasional reprise of a much older WU? Or if not, is it possibler to say briefly what is going on?

I am merely curious. Please don't go off and do any research to answer this, but if you know why off the top of your head, I would be interested in a short answer, or a link to a longer one.

R~~
____________
My computers found:

9831*21441403+1 is a quadhectokilo prime prime, ie >400,000 digits ;)

252031090528237591 + 65521*149*23*19*17*13*11*7*5*3*2*n is prime for every n in { 0..20 } (an arithemtic progression of 21 primes)

Crun-chi
Volunteer tester

Joined: 25 Nov 09
Posts: 2980
ID: 50683
Credit: 58,097,404
RAC: 107,986

Message 105044 - Posted: 15 Feb 2017 | 12:39:06 UTC - in response to Message 105042.
Last modified: 15 Feb 2017 | 12:40:26 UTC

What I do not understand is that against this steady gradual rise there is an occasional task that completes in about 75% of that time and with its credit pro rata. For example this WU was awarded 32.79

Is this an occasional fortuitous set of values that fit better than usual into a smaller FFT?
R~~

32.79 credits Using all-complex FMA3 FFT length 96K,
43.44 credits Using all-complex FMA3 FFT length 120K

Short enough?

If you take and look at few results , you will get answer before ask on forum. But it looks like it is easier to answer here, and wait for answer.
____________
314187728^131072+1 GENERALIZED FERMAT :)
93*10^1029523-1 REPDIGIT PRIME
31*332^367560+1 CRUS PRIME
Proud member of team Aggie The Pew. Go Aggie!

River~~

Joined: 17 Mar 07
Posts: 342
ID: 6533
Credit: 15,792,075
RAC: 0

Message 105055 - Posted: 15 Feb 2017 | 17:55:07 UTC - in response to Message 105044.
Last modified: 15 Feb 2017 | 18:19:18 UTC

Thanks Crun-Chi

I hadn't realised that info was available in the task page, thanks for the heads-up.

I will certainly look there first next time.

But it looks like it is easier to answer here, and wait for answer.

Well! you expressed that in a rather pointed manner, and it is not the first time.

I would like to respond to that.

I would like you to know that I spent over an hour looking at threads in these forums to see if I could find the answer before I asked. Had I known where to look it would have been easier than what I actually did.

Strictly speaking I am not a newbie, as I did a little TPS work back in 2007. When I returned in 2016 I had forgotten everything, and some of it had changed anyway, so I am effectively a newcomer again.

Please don't expect people to magically know where to look for stuff. I ask questions when I know I am ignorant and want to reduce my ignorance.

It is not all one way. I try to give back by answering other people's questions, when I think I know the answers myself.

I also make suggestions about how to improve things on projects I am crunching.

Sometimes those suggestions are useful, sometimes they turn out to be far more work that I imagine and therefore impractical. I have made a few specific requests here, about half of which have been implemented. When I am told "that is impractical" I accept that the project maintainers know what they are saying, and don't make a fuss about it. When something is accepted, and implemented, I accept that as an indication that it was an idea worth mentioning.

When I was previously active in Boinc, ten years ago, there was something about the variation of runtimes of Rosetta WU that was really annoying a lot of crunchers there. I posted a pseudo-code workaround, which the project team felt was worth implementing then, and are still using that stategy now. . It is still listed on their website as one of the ten reasons to choose Rosetta: (here (my idea is at no 2 in the left hand list, and no 6 on the right -- and no, my strategy would not be relevant to PG WU).

That is by far my best Boinc contribution yet, above and beyond the machine time.

To get to a place where I knew enough to spot the solution to the issue, I had to ask a lot of questions first. It is fair to say I have come up with a huge number of less valuable ideas as well....

Perhaps I will never have a good idea here. Perhaps one of my ideas will again turn out to be useful, who knows?

Please think about how you word what you say. You may believe I (or someone else) is being lazy, but it is not often helpful to say so -- not most of the time.

There are usually ways to impart the information that are more gracious --- look at Mike's posts for a consistent example of saying "No" graciously to many people. In this instance, maybe you could have said something like "I wonder if you are aware that that info is at the foot of the page of every task?". That would pass the same factual message but without the dig.

And if you do not feel like phrasing your answer graciously, in my opinion it might be wiser not to answer at all.

R~~
____________
My computers found:

9831*21441403+1 is a quadhectokilo prime prime, ie >400,000 digits ;)

252031090528237591 + 65521*149*23*19*17*13*11*7*5*3*2*n is prime for every n in { 0..20 } (an arithemtic progression of 21 primes)

Crun-chi
Volunteer tester

Joined: 25 Nov 09
Posts: 2980
ID: 50683
Credit: 58,097,404
RAC: 107,986

Message 105060 - Posted: 15 Feb 2017 | 23:36:54 UTC - in response to Message 105055.

And if you do not feel like phrasing your answer graciously, in my opinion it might be wiser not to answer at all.

Mr River.
It is OK to have idea, and you, like I did, and many others do have idea. Some will be good, some will not be. I know about you ,since I read many forum threads before and noticed your name.

I am just one od many users of Primegrid, I am not admin, not moderator, not anything. Small , simple person that search for primes on and outside Primegrid. I dont care about badges, about how many credits I have, what color will be next badge. As you can see in my signature, all types of primes I found are not even searched on Primegrid. So I am here just for fun.

To have one, two, three, five idea(s) it is OK, but you open 23 threads ( I stop counting after that). Suppose that any of us open 23 threads in short period of time. No one before do that except you, and I think that you are not valuable member of the community, at least not more then any other. And since nobody do that it must be reason why?
Reason is: all forum is on the back a couple of administrators and moderators. They are humans like us, and they have limit. And they do extremely hard work extremely good. You are saying that my answer is brisk and their decent. And you give me advice to not answer at all to (yours question)

And I will write this now and never more:
1. if you have something against me, then use PM,or you send message to admin.
I accept every measure they got against me,because I respect them.
But since you are same as me, I will continue to answer to your question in way I know.

Thread ( this or any other) is not for private conversation , for that use PM
Best regards Mr River
____________
314187728^131072+1 GENERALIZED FERMAT :)
93*10^1029523-1 REPDIGIT PRIME
31*332^367560+1 CRUS PRIME
Proud member of team Aggie The Pew. Go Aggie!

River~~

Joined: 17 Mar 07
Posts: 342
ID: 6533
Credit: 15,792,075
RAC: 0

Message 105144 - Posted: 18 Feb 2017 | 22:05:56 UTC - in response to Message 105060.
Last modified: 18 Feb 2017 | 22:18:09 UTC

I wasn't aware I had invited a personal conversation. Certainly I did not mean to.

Just to be clear: don't want a personal conversation with you, not here and not on PM.
____________
My computers found:

9831*21441403+1 is a quadhectokilo prime prime, ie >400,000 digits ;)

252031090528237591 + 65521*149*23*19*17*13*11*7*5*3*2*n is prime for every n in { 0..20 } (an arithemtic progression of 21 primes)

Message boards : Proth Prime Search : PPSE occasional WU has ~75% of expected runtime & credit

[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2020 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.97, 4.21, 3.82
Generated 27 Oct 2020 | 14:43:52 UTC