Well, I have a question about "granted credit". I have, out of my last 20 files processed, 4 that are in "pending" status for granted credit, with one of them completed 5 days ago. What is the BOINC process for moving credits from "pending" to "granted". I can understand a delay of hours, but 5 days?

It can be a lot longer than 5 days. Credit is not granted until at least 3 validated results are returned. If the WU you have pending is still awaiting return from a slower computer, or if it has errored out from another user and had to be reissued, it could take a while. You will not lose your credit, it just won't move from pending to granted until the quorum of 3 is reached.
Patience, my friend. I currently have about 3000 credits pending, it's all good.

Thanks. Right after I posted the question, I actually bothered to search the Wiki and arrived at the same answer. I'm glad that they finally got BOINC's act together as far as the installation: I originally tried to install BOINC almost a year ago, but after 6 or 7 attempts with different versions, I gave up (it would never connect to localhost). I finally tried again 2 weeks ago after receiving an e-mail asking about my account, and presto, no problems. Now, this may have been self-inflicted: I had the Cox Cable Security Suite installed. I found about 4-5 months ago that it was preventing me from using my company's new remote access software, so I de-installed it. I wonder if that was also preventing BOINC form working. Anyway, all is well with the world, but I lost a year of processing (or UC Berkeley did).

---

Well, I have a question about "granted credit". I have, out of my last 20 files processed, 4 that are in "pending" status for granted credit, with one of them completed 5 days ago. What is the BOINC process for moving credits from "pending" to "granted". I can understand a delay of hours, but 5 days?

It can be a lot longer than 5 days. Credit is not granted until at least 3 validated results are returned. If the WU you have pending is still awaiting return from a slower computer, or if it has errored out from another user and had to be reissued, it could take a while. You will not lose your credit, it just won't move from pending to granted until the quorum of 3 is reached.
Patience, my friend. I currently have about 3000 credits pending, it's all good.

---

"Learn from yesterday. Live for today. Hope for tomorrow." Albert Einstein
"With cats." kittyman

Well, I have a question about "granted credit". I have, out of my last 20 files processed, 4 that are in "pending" status for granted credit, with one of them completed 5 days ago. What is the BOINC process for moving credits from "pending" to "granted". I can understand a delay of hours, but 5 days?

---

Guys ... thank you ....
What I take from this is that same work same credit ... applies to those who crunch the same WU ...
That is fine by me.

Thank you for your time ...

It's the way workunits are calculated as explained earlier in this thread. There are several types of workunits. The two most common are the 60-credit units and the 30-credit units. The latter yield fewer credits per hour. The 50-credit units have almost the same yield as the 60-credit ones. The 12-to-20-credit units (the shorter-to-calculate units) give variable, but, on average, slightly better yields. The high-50s units are often the VLARs that give slightly poorer yields. In my experience thay all seem to average out pretty well day after day, but that could, of course, change. Credit assignment varies with the angle range of the unit, which is the distance, in degrees, that the Seti receiver moves through the stars, during the course of a workunit's collection of stellar data, about 107 seconds. Differing angle ranges require different studies by the Seticruncher application program because of the way the receiver looks at the stars: The receiver stares at one tiny area of them for 107 seconds (VLAR), drags across several degrees of them (VHAR), or in-between.

---

you did more cpu flops with one, than the other, ie you has something running like windows media player, using cpu time up.

thats what i remember from the other threads, it's all about flops, i think something just flops over in the cpu more times when it's busy, so you dont get as many with your WU

see aint i a smart cookie ;p

Uh, no.

CPU seconds are kept by task. They have very little to do with "wall time" or other operations running on the same machine.

When we went to counting FLOPs, every floating point operation counts as 1. A floating point ADD counts the same as a floating point SIN(), even though the ADD happens a whole lot faster.

---

Nowhere can I get either a logical nor yet believable answer to my question. Please would someone explain to me in simple terms before I just shut off my computers yet again.
For 2 hours work (7,300cpus) I get a credit of 27.8, for a per hour rate of 13.5 credits.
For 3 hours work (10,841 cpus) I get a credit of 60.11, for a per hour rate of 20 credits.
I thought that there was a same work/same credit policy in force.

Why this discrepancy?

"Same work/Same Credit" means that everyone who successfully crunches the same work unit gets the same credit -- and nothing more.

---

Very simplified example, let's say you've got 2 types of calculations, type-1 is +- and type-2 is */.

Now, not unexpectedly, type-2-calculations is slower to do, so let's say a computer can do 3 type-1-calculations in the time it takes to do 1 type-2-calculation.

2 wu, A and B:
A needs 1000 calculations, thereof 90% type-1.
B needs 500 calculations, thereof 50% type-1.

If also says 1 type-1-calculation takes 1 second, you get:
A => 900 calculations * 1 second/calculation + 100 calculations * 3 seconds/calculation => 1200 seconds for 1000 calculations.
B => 250 * 1 + 250 * 3 => 1000 seconds for 500 calculations.

If gets 1 "credit" per 50 calculations:
A = 20 "credits" => 60 credits/hour.
B = 10 "credits" => 36 credits/hour.


Now, in this simplified example, someone will immediately say, "Why not give each type-2-calculation 3x more credit than each type-1-calculation?"

But, the problem is, while cpu-1 uses 3x as long per type-2-calculation, cpu-2 can use 2x as long, while cpu-3 can use 4x as long...

Meaning, regardless of which cpu you uses to "calibrate" credit vs. time, cpu-1, cpu-2 or cpu-3, one of them will get "paid" more credit/second than the other cpu's.


In SETI@home, there's 4 different signal-types, each with their own calculation-routines, and the Angle Range of a wu decides how many calculations of each type needs to be done. For some angle-ranges, not all types of signals is even looked-for at all. Some cpu's is better on one type of the calculations than the other types, and for SETI@home even memory-bandwidth influences crunch-times.

So, just like my simplified example, for some Angle Ranges a particular computer will be slower doing the calculations, and therefore gets "paid" less "credit" per hour.


Bottom line is, for 90%+ of wu the credit/hour is roughly equal, but for Angle Ranges outside the most common range there is some variations. While there are plans to make these variations smaller, it's impossible to fully remove them, since the fact is, some computers is better on certain calculations than another computer is.

i think mine was the really simple way to explain it ;p

---

I AM NOT FAT! I AM BIG-BONED!

Very simplified example, let's say you've got 2 types of calculations, type-1 is +- and type-2 is */.

Now, not unexpectedly, type-2-calculations is slower to do, so let's say a computer can do 3 type-1-calculations in the time it takes to do 1 type-2-calculation.

2 wu, A and B:
A needs 1000 calculations, thereof 90% type-1.
B needs 500 calculations, thereof 50% type-1.

If also says 1 type-1-calculation takes 1 second, you get:
A => 900 calculations * 1 second/calculation + 100 calculations * 3 seconds/calculation => 1200 seconds for 1000 calculations.
B => 250 * 1 + 250 * 3 => 1000 seconds for 500 calculations.

If gets 1 "credit" per 50 calculations:
A = 20 "credits" => 60 credits/hour.
B = 10 "credits" => 36 credits/hour.


Now, in this simplified example, someone will immediately say, "Why not give each type-2-calculation 3x more credit than each type-1-calculation?"

But, the problem is, while cpu-1 uses 3x as long per type-2-calculation, cpu-2 can use 2x as long, while cpu-3 can use 4x as long...

Meaning, regardless of which cpu you uses to "calibrate" credit vs. time, cpu-1, cpu-2 or cpu-3, one of them will get "paid" more credit/second than the other cpu's.


In SETI@home, there's 4 different signal-types, each with their own calculation-routines, and the Angle Range of a wu decides how many calculations of each type needs to be done. For some angle-ranges, not all types of signals is even looked-for at all. Some cpu's is better on one type of the calculations than the other types, and for SETI@home even memory-bandwidth influences crunch-times.

So, just like my simplified example, for some Angle Ranges a particular computer will be slower doing the calculations, and therefore gets "paid" less "credit" per hour.


Bottom line is, for 90%+ of wu the credit/hour is roughly equal, but for Angle Ranges outside the most common range there is some variations. While there are plans to make these variations smaller, it's impossible to fully remove them, since the fact is, some computers is better on certain calculations than another computer is.

---

"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

Simple terms.......OK.
All work units are not created equal. Although the size of each WU is the same, the data they contain is not. Some WUs require more and/or more complex computations to complete than others. The yield in terms of credit/hour can differ widely from one WU to the next, even ones from the same tape of raw data. It has to do with something called AR (angle range), which in simple terms, is whether the radio telescope dish was looking straight up into the sky when the data was recorded, or aimed to the left or right.
This is the same for all computers crunching Seti, not just yours. All cpus, all platforms, all OSs are affected by this.
Same work/same credit policy holds true here. If you check your results, you should see that, except in a few odd cases, when you are getting less credit/hour for a particular WU you have processed, all other users returning the WU are getting the same amount of credit for that WU.
Your results are normal, so please don't give up on the project, you are contributing valuable science to Seti. As I have said before, the points are just a game, not the goal.

And the kitties say..."Keep on Crunchin'..Meow"

---

"Learn from yesterday. Live for today. Hope for tomorrow." Albert Einstein
"With cats." kittyman

Nowhere can I get either a logical nor yet believable answer to my question. Please would someone explain to me in simple terms before I just shut off my computers yet again.

Wow, after 15 days with Seti you have already (according to you) turned your computers off once (for some reason) and are thinking of doing it again because you think you are not getting enough credit.

What would you ever do if you lost a WU?

That’s a frightening thought! Please don’t tell me, I was scared enough this Halloween.

Credits are meaningless, it’s supposed to be about the science.

The only person that will ever get the “most” credit is the person(s) who find

“THE SIGNAL”

It's no longer about time. It's about how many FLoating point Operations Per Second (FLOPS). Some angle ranges can be calculated faster per second than others, because there are not as many difficult math routines. It's like most people having a math problem that is just adding, or one that has complex algebra. Which one do you think you can do faster?

---

you did more cpu flops with one, than the other, ie you has something running like windows media player, using cpu time up.

thats what i remember from the other threads, it's all about flops, i think something just flops over in the cpu more times when it's busy, so you dont get as many with your WU

see aint i a smart cookie ;p

---

I AM NOT FAT! I AM BIG-BONED!

Nowhere can I get either a logical nor yet believable answer to my question. Please would someone explain to me in simple terms before I just shut off my computers yet again.
For 2 hours work (7,300cpus) I get a credit of 27.8, for a per hour rate of 13.5 credits.
For 3 hours work (10,841 cpus) I get a credit of 60.11, for a per hour rate of 20 credits.
I thought that there was a same work/same credit policy in force.

Why this discrepancy?