This has happened in the past, BOINC will process quickly through WUs and then it will take at least 2 times as long. Nothing else has changed. BOINC still gets 100% cpu, but processes take twice as long. My one computer has gone from .24 duration to .5 something.

Does anyone know why this happens?

---

I suspect, from your wording, you are referring to
the Duration Correction Factor data.

This factor has nothing to do with how long units actually
take to process - it only affects BOINC's *estimate* of how
long it will take. This is used to determine how much work
you have on hand (and thus, how much more work you should
request). If so, this is not a real problem, other than
how much work is being maintained in your cache.

Unless you mean that units are actually running for longer
runtimes, and if so, you have some other problem.

Dave

Sorry to be unclear. I meant duration of the actual task. And now the correction factor is twice as high too. Meaning that my computer is actually computing tasks slower approx. by a factor of 2.

However, it seems like there were some low value tasks that took a lot of ocmputing power. I'm not sure how that value is calculated, but things look like they are on track again.

---

Sorry to be unclear. I meant duration of the actual task. And now the correction factor is twice as high too. Meaning that my computer is actually computing tasks slower approx. by a factor of 2.

However, it seems like there were some low value tasks that took a lot of ocmputing power. I'm not sure how that value is calculated, but things look like they are on track again.

I think that you are crunching very noisy WU's... That makes increase your correction factor and apparently your processes are more slowly... Is only 'apparently'. Noisy WU's can spend more cpu seconds than the normals... (a lot of...) :)

Be patient...

Best regards.

---

Logan.

BOINC FAQ Service (Ahora, también disponible en Español/Now available in Spanish)

Sorry to be unclear. I meant duration of the actual task. And now the correction factor is twice as high too. Meaning that my computer is actually computing tasks slower approx. by a factor of 2.

However, it seems like there were some low value tasks that took a lot of ocmputing power. I'm not sure how that value is calculated, but things look like they are on track again.

Every task or workunit (WU) has a different Angle Range (AR), according to what was happening at the telescope when the data was recorded.

SETI@home does different work on the WU according to AR. You can see the AR in your results e.g. "WU True angle range: 0.444545"

Different AR WU's get different credit and different deadlines too according to how long they are estimated to run.

If you look at your recent results you will see different ammounts of credit according to AR

WU true angle range is :  0.404012   Claimed credit 54.33
WU true angle range is :  1.386794   Claimed credit 19.41
WU true angle range is :  0.403866   Claimed credit 7.99   SETI@Home Informational message -9 result_overflow
WU true angle range is :  0.403816   Claimed credit 32.14   SETI@Home Informational message -9 result_overflow
WU true angle range is :  0.444076   Claimed credit 50.73
WU true angle range is :  0.348418   Claimed credit 81.12
WU true angle range is :  0.010435   Claimed credit 63.98
WU true angle range is :  0.981818   Claimed credit 29.02
WU true angle range is :  2.402946   Claimed credit 17.37

These are just an example from some of your recent results. The -9 result_overflow results are noisey WU's where SAH stopped processing when it found >30 pulses.

I think other more expert people can explain this better or refer to a web page with more information.

Keith.

Angle range is how far the telescope dragged through the stars during the 107-second data-collection duration of the workunit. If the scope followed the stars during data collection (small angle range) different calculation studies are required from when the scope swept several degrees of sky (large angle range). This results in different crunchtimes for units of different angle ranges. Credit is supposed to be awarded in accordance with crunchtimes. Unfortunately the crunchtime/credit factor, which should be as constant as possible, varies quite a bit. So the result is workunits of varying credits and of varying crunchtimes. It averages out pretty well over a month or so. The duration correction factor has something to do with estimated crunchtimes and how many units the cache will load. It may be funny temporarily but corrects out in the end.

---

At onen point I experienced my one computer slowing down to about 1/2 speed. I don't know what caused it, but things took much longer than in the past. That appeared to happen again. However, this time it was just a series of WUs that took long in relation to credit. None had been crunched by another computer at first post, so I couldn't tell if I was crunching them in proportion. Rather it was easier to jump to the conclusion that my system was somehow processing slower. :)

Thanks for the info, it helps to understand this a little better. I'm sure many others are curious what's going on. Of course if I was curious enough I'd grab a copy of the code and look.

---

At onen point I experienced my one computer slowing down to about 1/2 speed. I don't know what caused it, but things took much longer than in the past. That appeared to happen again. However, this time it was just a series of WUs that took long in relation to credit. None had been crunched by another computer at first post, so I couldn't tell if I was crunching them in proportion. Rather it was easier to jump to the conclusion that my system was somehow processing slower. :)

Thanks for the info, it helps to understand this a little better. I'm sure many others are curious what's going on. Of course if I was curious enough I'd grab a copy of the code and look.

Glad to help, you can see the AR of your current tasks by looking in the slot folders while the tasks are crunching.

e.g.
C:\\Program Files\\BOINC\\slots\\0
C:\\Program Files\\BOINC\\slots\\1

There will be a file in each slot folder or directory called stderr.txt, if you open that file using Notepad, you can see the AR of your current task. Make sur that you do not modify that file while viewing it.

---

Sir Arthur C Clarke 1917-2008

I followed this thread to see whether it may explain my Intel dual-core Extreme X6800 processors slowing down dramatically after running just a few minutes. Bascially, my preferences are set to consume my machine's capabilities. It once ran very fast, but then I noted a steady decrease in speed. (Both CPU's used to run at 100% as indicated in the CPU Usage History of the Windows Task manager.)

I've read through the FAQ's that describe the preference settings, but tweaking them doesn't increase my machines usefulness. Are there any other keywords that one might use to search for CPU utilization issues that someone might recommend in order to find different message threads that discuss how to trouble shoot this kind of number crunching issue?

Finally, is there a best-practices way to troubleshoot this kind of phenomenon? I really don't find anything wrong - except for the fact that my CPU's ramp down from 100% after starting BOINC - to 15-30% - after a few minutes, but they never ramp back up like my preferences would suggest they should do.

Thanks for any help in advance!

---

I really don't find anything wrong - except for the fact that my CPU's ramp down from 100% after starting BOINC - to 15-30% - after a few minutes, but they never ramp back up like my preferences would suggest they should do.

Thanks for any help in advance!

Is your cooling OK? It may be that the CPUs are throttling back if they're getting hot?

Just a thought ...

Steve.

---

I really don't find anything wrong - except for the fact that my CPU's ramp down from 100% after starting BOINC - to 15-30% - after a few minutes, but they never ramp back up like my preferences would suggest they should do.

Thanks for any help in advance!

Is your cooling OK? It may be that the CPUs are throttling back if they're getting hot?

Just a thought ...

Steve.

You might also want to check the power savings settings. Boinc runs at the lowest priority and it has been reported that some mb/cpu/os combinations do not recognize this as using the cpu so it throttles down the cpu into 'power savings" mode. If your computer has this feature you might want to turn it off and see if it has any effect.

---

Jim

Some people plan their life out and look back at the wealth they've had.
Others live life day by day and look back at the wealth of experiences and enjoyment they've had.

I really don't find anything wrong - except for the fact that my CPU's ramp down from 100% after starting BOINC - to 15-30% - after a few minutes, but they never ramp back up like my preferences would suggest they should do.

Thanks for any help in advance!

Is your cooling OK? It may be that the CPUs are throttling back if they're getting hot?

Just a thought ...

Steve.

Hello Steve,

My cooling is okay (based on my Intel-provided monitoring software). I've tried several preference changes, and I also double-checked my power settings are okay based on another suggestion I received in response to my original post.

I'm getting suspicious of a recent bios update I made though. It was probably not needed, except that it addressed components I am using. However, I did not benchmark my computer afterwards, and the BOINC benchmark did seem to change with it. However, if I can safely roll back that change, I'll see if this brings performance back up.

Brgds.

Michael

---

I really don't find anything wrong - except for the fact that my CPU's ramp down from 100% after starting BOINC - to 15-30% - after a few minutes, but they never ramp back up like my preferences would suggest they should do.

Thanks for any help in advance!

Is your cooling OK? It may be that the CPUs are throttling back if they're getting hot?

Just a thought ...

Steve.

You might also want to check the power savings settings. Boinc runs at the lowest priority and it has been reported that some mb/cpu/os combinations do not recognize this as using the cpu so it throttles down the cpu into 'power savings" mode. If your computer has this feature you might want to turn it off and see if it has any effect.

Hello Jim,

I checked to be sure my power settings. They seem to be disabled, so they don't seem to be interfering.

I did find that my decrease in performance lagged a recent bios update I made for another issue. I didn't perform any benchmarks afterwards, except that my BOINC benchmarks seemed to not be affected. Anyway, I am suspicious of it now, and if I can roll back the BIOS update safely, I do it! Anyway, I wanted to let you know that the power settings seemed okay.

Best regards,

Michael

Best regards,

---