Could anyone tell me please how the system works for W.U. completed
You give 26,826 sec's of CPU time and get 30.88 credits
You give 17,267 sec's of CPU time and get 62.43 credits

Not sure how you clever clogs put up easily read boards. I have copied and pasted part of mine up so you can see the irregularity of the system. (If it shows up that is)

557027160 136323820 21 Jun 2007 23:28:18 UTC 23 Jun 2007 13:49:53 UTC Over Success Done 26,826.85 30.88 30.88

557027024 136323768 21 Jun 2007 23:28:18 UTC 23 Jun 2007 6:45:07 UTC Over Success Done 25,738.88 30.89 30.88

556947586 136298293 21 Jun 2007 20:22:43 UTC 22 Jun 2007 0:31:53 UTC Over Success Done 4,074.01 14.58 12.79

556784882 136245976 21 Jun 2007 14:14:59 UTC 23 Jun 2007 13:51:31 UTC Over Success Done 17,267.78 71.03 62.43

556453608 136143398 21 Jun 2007 2:32:51 UTC 22 Jun 2007 15:20:23 UTC Over Success Done 12,222.73 51.83 45.55

556192490 136060031 20 Jun 2007 16:04:12 UTC 22 Jun 2007 12:31:22 UTC Over Success Done 16,501.96 69.24 60.86

Thanks if you can make me wiser with a logical answer
Yours
Confuddled

---

Alien Pet Lover

Could anyone tell me please how the system works for W.U. completed
You give 26,826 sec's of CPU time and get 30.88 credits
You give 17,267 sec's of CPU time and get 62.43 credits

Although this might be better off posted in the number crunching forum I'll try to answer you here....

Credit is awarded based on the number of computations, also known as Floating Point Operations or FLOPS it takes to complete a given workunit. It doesn't matter how long it takes......what matters is the amount of effort that the processor had to put in to compute the result.

For example, if I ask you to perform a series of additions I can bet that you'll be able to perform them faster than a similar set of divisions on a given sample of data....

That said, different workunits have been recorded at different angle ranges by the Arecibo dish. Some angle ranges have easier computation than others....so it is often the case that the mid-range workunits take longer in time than the ones that give you 60 or more credits.

It might be worth reading the Enhanced FAQ to get an idea of the whole process.

---

Could anyone tell me please how the system works for W.U. completed
You give 26,826 sec's of CPU time and get 30.88 credits
You give 17,267 sec's of CPU time and get 62.43 credits

Although this might be better off posted in the number crunching forum I'll try to answer you here....

Credit is awarded based on the number of computations, also known as Floating Point Operations or FLOPS it takes to complete a given workunit. It doesn't matter how long it takes......what matters is the amount of effort that the processor had to put in to compute the result.

For example, if I ask you to perform a series of additions I can bet that you'll be able to perform them faster than a similar set of divisions on a given sample of data....

That said, different workunits have been recorded at different angle ranges by the Arecibo dish. Some angle ranges have easier computation than others....so it is often the case that the mid-range workunits take longer in time than the ones that give you 60 or more credits.

It might be worth reading the Enhanced FAQ to get an idea of the whole process.

Being blind causes me a time lag actually waiting for Jaws on my computer to read what is on screen and then going back to links etc. However I did follow your link and having listened to the explanation given it makes me no wiser. It states that you are paid for CPU time rather than the W.U. which I understand. I also understand that a work unit taking 4 hours (14,400 seconds) in real time may take much longer depending on how hard the CPU is working on other things as it only uses spare power.

It is the fact that it gives you an estimate of CPU time and credit when you download the W.U. and as most of the time it is roughly correct, these discrepancy's being so large are queryable.

Thanks again for the links as finding things like links unless I have my God-daughter here to help can sometimes be impossible to find.

Baz

---

Alien Pet Lover

Real time taken is always different from the clock time or time reported when the workunit is completed. As you said, in real time it may take much longer depending on how hard the cpu is working on other things....

As an example consider if you had to perform a series of math operations where the end result was the number 50.

It could be done this way starting from any arbitrary number.

10 plus 15 equals 25 plus 11 equals 36 plus 14 equals 50

I'm sure that it would be very easy to add these numbers together. However consider the following...

225 divided by 3 equals 75 divided by 6 equals 12.5 multiplied by 4 equals 50

I'm sure in this case the first division would take that much longer and so would the second operation...This is just a simple example, but I hope you understand what I mean.

The other thing that you mentioned was the initial estimate of cpu time when you download the workunit. This is dependent on lots of additional factors, that include what the recent performance of your pc has been as reported by BOINC.

Initially you wouldn't really have an idea of how much credit you would get for the workunit unless you sort of have been keeping track of various angle ranges and the like or are rather more technically minded....ha!

These variations in credit have been graphed before and it's been found that there is a sort of negative sweetspot in the orientation of the telescope when the signal was recorded. I believe that these workunits that give lesser credit per hour because they take longer are generally fewer and occasionally we all have the bad luck to be assigned these.

But don't worry about, your pc should be working fine and this situation is more or less normal.

---