I have credits that have been pending since July. When should I get credit for them?

---

> I have credits that have been pending since July. When should I get credit
> for them?
>
Currently they have been having database issues and on top of that the validator is down. They are ALL contributing to your problems.
July itself may be another itself though, those units may just have to be purged if no consensus can be achieved.

---

Just adding to the pending credits theme:

If (for example) I completed a WU on say Monday, but it was pending credit until someone else finishes their's... Wedensday rolls around and the credits are granted... What day to the credits get added to my stats for, Monday when I completed the WU or Wedensday when the credits were granted??

Cheers

---

Cheers

Gav



Nothing is 'fool-proof', someone will always invent a better fool!

... What day to the credits get added to my stats for, Monday when
> I completed the WU or Wedensday when the credits were granted??
>
> Cheers
>
Wednesday, or maybe even Thursday or Friday if the validator is backlogged. LOL

Gheesh, one would have thought for acurate stats that the day of completion would be the day added to your stats... Mmmm

---

Cheers

Gav



Nothing is 'fool-proof', someone will always invent a better fool!

> Gheesh, one would have thought for acurate stats that the day of completion
> would be the day added to your stats... Mmmm
>
It is the day of completion...The day the validator sees three comparable WU that have been returned, it marks it as "Complete", calculates, and issues credit.

tony

[edit] if it were the day a user completed it, then the user could return a junk file and get issued credit. [end edit]

[second edit]I have done a study of my results since Jan 8th. The average number of days it takes to get a validated result is 9.30. This count starts the day the first 4 are sent and ends the last day a successful WU is returned. [end edit]

Ah not if the system was clever enough to hold the credits until validated (as is it is currently) then once the 3 pc' have returned the results, credit units but backdate to when the pc finished it. Otherwise the daily returns stats are basicly just rubbish stats because they arnt 'actually' what my pc is doing on a daily basis.

---

Cheers

Gav



Nothing is 'fool-proof', someone will always invent a better fool!

> Ah not if the system was clever enough to hold the credits until validated (as
> is it is currently) then once the 3 pc' have returned the results, credit
> units but backdate to when the pc finished it.

Paul Buck has documented this credit process. Look for one of his posts and check out what he's says about the credit process. It would take me hours to try to type it out here. It's pretty complicated. Suffice it to say that each Wu is different. Since, each is different, then the credit granted is different. The validator has to wait for three comparable results to even assign a credit value. Based upon Claimed Credit, the validator, discards the High request, the Low Request, and averages out the remaining claimed credits, then grants this amount to all who have successfully crunched that WU.

>Otherwise the daily returns
> stats are basicly just rubbish stats because they arnt 'actually' what my pc
> is doing on a daily basis.
>

Are you discussing RAC (recent average credit), or the amount of credit shown in your account?

hope this helps

tony

Here's a link to Pauls' Website

do a search for Claimed Credit, or look through the FAQ, Enjoy his 700+ pages of fascinating things related to Boinc.

Not the RAC I've had a read up about that, seems very complicated. Was just meaning the actual day of completion for a WU is not the day you get credit for it, so the daily stats are not indicative of what your pc is actually doing.

---

Cheers

Gav



Nothing is 'fool-proof', someone will always invent a better fool!

> Not the RAC I've had a read up about that, seems very complicated. Was just
> meaning the actual day of completion for a WU is not the day you get credit
> for it, so the daily stats are not indicative of what your pc is actually
> doing.
>
You are correct, sort of. It's more a representation of what you did 9.30 days ago, Roughly. If you stop doing any work, then you'd still be getting credit for roughly the next 9.3 days (or that's what my data shows, yours may vary slightly). Does this make any sense? The Classic Seti Cheaters caused this new "Fairer" credit system to come into existance. You will get all the credit you're due, it just takes time.

tony

Yeah thanks i understand how its done now. And I do realise that the credit will arive at some stage, just wish that when it did arrive it would be backdated to when you actually completed it. As I said then the daily stats would become acurate(after 9.3 days of course)

---

Cheers

Gav



Nothing is 'fool-proof', someone will always invent a better fool!

> Yeah thanks i understand how its done now. And I do realise that the credit
> will arive at some stage, just wish that when it did arrive it would be
> backdated to when you actually completed it. As I said then the daily stats
> would become acurate(after 9.3 days of course)
>
Hey, You're another New Zealander, my cousin in Wellington just sent my wife a "New Zealand" calendar. She says the mountains are OK, but the Beaches look Pretty. I wonder what she means??? LOL I guess she wants to visit my cousin.

Have you ever looked at your results? you can learn from them. For example, if you click on my name (or yours), you can look up your or my data. Note: your URL and other sensitive data is automatically hidden from everyone but you.

tony

Yeah, I'm a Kiwi, well actually Scottish but have lived here long enough to be considered a New Zealander. The whole scenery looks fantastic from the air! As I do alot of flying I reckon I have the best office in the world!

I use http://www.boincstats.com/stats/boinc_user_graph.php?pr=bo&id=xxxx for my stats

---

Cheers

Gav



Nothing is 'fool-proof', someone will always invent a better fool!

> Yeah thanks i understand how its done now. And I do realise that the credit
> will arive at some stage, just wish that when it did arrive it would be
> backdated to when you actually completed it. As I said then the daily stats
> would become acurate(after 9.3 days of course)

Tony and I have slightly different data and mine indicates that the "sweet-spot" is closer to 5 days, but what the heck ...

I suspect that the difference lies in the the rate of return. One of the things that the newer schedulers are attempting to do is to schedule work to "compatible" hosts so that if you are returning work within 2 days, it tries to put you in a "group" of hosts that also will return the work within that turn-around.

The point is to try to maximize the Project's ability to "retire" work from the on-line database to keep its size small. Einstein@Home tries to do this because of the limits they have on hardware. CPDN, Predictor@Home, LHC@Home, and SETI@Home seem to have the largest amount of work on-line, with SETI@Home by WU count being the largest.

---

> Tony and I have slightly different <a> href="http://boinc-doc.net/site-misc/redundency-analysis.php">data[/url] and
> mine indicates that the "sweet-spot" is closer to 5 days, but what the heck
> ...
>
> I suspect that the difference lies in the the rate of return.

Hi paul, I came up with this figure by adding the "Number of Days" and dividing by the sample quantity. I added more samples this morning, here's the figures.

1525 days / 165 samples= 9.24 days

I think the difference lies in two things:
1) I'm on Dial Up, and have slow puters, and my data contains, well.... my data. I may be one of the slow returners.

2) I counted the number of days including the first day. So Subtract 1 from each of my WU (1525-165=1360). recalculate 1360/165=8.24 days.


tony

> > Tony and I have slightly different <a>
> href="http://boinc-doc.net/site-misc/redundency-analysis.php">data[/url] and
> > mine indicates that the "sweet-spot" is closer to 5 days, but what the
> heck
> > ...
> >
> > I suspect that the difference lies in the the rate of return.
>
> Hi paul, I came up with this figure by adding the "Number of Days" and
> dividing by the sample quantity. I added more samples this morning, here's
> the figures.
>
> 1525 days / 165 samples= 9.24 days
>
> I think the difference lies in two things:
> 1) I'm on Dial Up, and have slow puters, and my data contains, well.... my
> data. I may be one of the slow returners.
>
> 2) I counted the number of days including the first day. So Subtract 1 from
> each of my WU (1525-165=1360). recalculate 1360/165=8.24 days.

Tony,

I was not faulting you or your analysis, I was commenting that I had slightly different data ... in mine, the peak is centered over 5 days ... and youo can see the fgraf

ok, bedtinme when i cant spell

---

> I was not faulting you or your analysis, I was commenting that I had slightly
> different data ... in mine, the peak is centered over 5 days ... and youo can
> see the fgraf
>
> ok, bedtinme when i cant spell
>
>
I know that Paul, I was getting the facts out, so that others would have a better understanding, and to help explain the disparity.

take care of yourself

tony

Paul I grabbed the data from your chart. Since the bottom is cut off at 21 days I had to exclude 5 results with the highest count in days, so the number I came up with is low. I shifted your "number of days" by one, so that it corresponds with the way I counted the number of days. So, 6763 days/ 991 samples = 6.82 days or 5.82 the way you count.

so, Fast computer and High speed connection is roughly 7ish, slow computer and dial up is 9.24. or, 6ish and 8.24 your way of counting.

tony

> Yeah thanks i understand how its done now. And I do realise that the credit
> will arive at some stage, just wish that when it did arrive it would be
> backdated to when you actually completed it. As I said then the daily stats
> would become acurate(after 9.3 days of course)
>
But if it were backdated how would you ever even know it happened?
You would have to be using a secondary site to even see the trend but you are asking them or whoever to backdate their stats everyday. That seems counter intutive. If the credit process is finally completed today then why should you or anyone get credit 3 days ago? You finished your part of the process but are waiting for someone else to do their work. This is why there has talk of getting computers that finish within parameters to be issued the same unit. For instance all of my Boinc computers can finish a unit in less than 24 hours, I have my cache set at 1 day. That means that everyday I return all work that was sent to me the previous day. If your computers do the same then we would be paired up and get the same units to crunch, meaning both of us would get our credits faster, assuming everything goes as it is supposed to. If your cache is set to 2 days you would be paired up with similar people etc.
NO decision has been made to implement this, that I am aware of.
Their is also talk of giving units to people of similar memory sizes, if you have 1 gig of memory you and others with that same amount would get the same units, while those of us with less would not get those units. That has not been implemented either, that I am aware of.
I think Boinc will evolve as time goes mkaing it more and more interesting.

---

Tony,

> I know that Paul, I was getting the facts out, so that others would have a
> better understanding, and to help explain the disparity.

I read that as you being defensive... or that I was critical or something... anyway, i misunderstood ...

> take care of yourself

I better, Nancy will kill me if I don't ...

> Paul I grabbed the data from your chart. Since the bottom is cut off at 21
> days I had to exclude 5 results with the highest count in days, so the number
> I came up with is low. I shifted your "number of days" by one, so that it
> corresponds with the way I counted the number of days. So, 6763 days/ 991
> samples = 6.82 days or 5.82 the way you count.
>
> so, Fast computer and High speed connection is roughly 7ish, slow computer
> and dial up is 9.24. or, 6ish and 8.24 your way of counting.

If you look at the chart beside the small table you can see the actual numbers.

There is a difference in the display of data because of different versions of MySQL, I am using close to the latest version and my provisioner is using an older version. SO, the last rows are supposed to be 15-20, 20-30, and >30. If I had not done that the table goes on forever with little improvement in understanding.

if you use the link embedded into the text you can see the actual 996 records, if you use this link you can get the actual datafile I used.

But you are right, these are minor details ...

---

> Tony,
>
> > I know that Paul, I was getting the facts out, so that others would have
> a
> > better understanding, and to help explain the disparity.
>
> I read that as you being defensive... or that I was critical or something...
> anyway, i misunderstood ...
>
> > take care of yourself
>
> I better, Nancy will kill me if I don't ...
>
> > Paul I grabbed the data from your chart. Since the bottom is cut off at
> 21
> > days I had to exclude 5 results with the highest count in days, so the
> number
> > I came up with is low. I shifted your "number of days" by one, so that
> it
> > corresponds with the way I counted the number of days. So, 6763 days/
> 991
> > samples = 6.82 days or 5.82 the way you count.
> >
> > so, Fast computer and High speed connection is roughly 7ish, slow
> computer
> > and dial up is 9.24. or, 6ish and 8.24 your way of counting.
>
> If you look at the chart beside the small table you can see the actual
> numbers.
>
> There is a difference in the display of data because of different versions of
> MySQL, I am using close to the latest version and my provisioner is using an
> older version. SO, the last rows are supposed to be 15-20, 20-30, and >30.
> If I had not done that the table goes on forever with little improvement in
> understanding.
>
> if you use the link embedded into the text you can see the actual 996 records,
> if you use this <a> href="http://boinc-doc.net/site-misc/redundency-data.txt">link[/url] you can get
> the actual datafile I used.
>
> But you are right, these are minor details ...
>
>
Off topic. I am glad there are people out there like paul that has compiled all this information. It seems like alot of work.

---