Mine when they've been coming in roughly the last month have been about 37 hours. Before that they were usually around 5.

Mine when they've been coming in roughly the last month have been about 37 hours. Before that they were usually around 5.

Maybe you should think again about using one of the optimized applications? Don't be fooled by the predicted runtimes, BOINC doesn't know how much faster another application is going to be. Just let it run and see how long it really takes. To show you what calculation times banditwolf gets with one these on a slower CPU than you have:

CPU time 18164.92
stderr out <core_client_version>5.10.45</core_client_version>
<![CDATA[
<stderr_txt>
Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 2.66GHz (1 cores/threads) 2.66624 GHz (1146ms)

WU completed. It took 18164.9 seconds CPU time and 20133.5 seconds wall clock time @ 2.66625 GHz.

</stderr_txt>
]]>

Validate state Valid
Claimed credit 43.0809464309155
Granted credit 213.759239

Looks like 5 to 5.5 hours to me on a 2.66GHz P4. With hyperthreading (very efficient here at MW, at least on Core i7) it should be roughly 2 WUs in maybe 7-8 hours for your 3.4GHz P4. Definitely not too shabby considering the credits and also quite doable within the three days deadline.

I tried one and it didn't help. Maybe I didn't put it in right - I'm not very computer literate. It predicts about 37 hours but ends up taking much longer - a lot of times it takes more than a second for the time to drop by 1 second. Before I was only aborting when it got to more time left then there was until the deadline. Eventually I just aborted if it was a more than 30 hour project.

I tried one and it didn't help. Maybe I didn't put it in right - I'm not very computer literate. It predicts about 37 hours but ends up taking much longer - a lot of times it takes more than a second for the time to drop by 1 second. Before I was only aborting when it got to more time left then there was until the deadline. Eventually I just aborted if it was a more than 30 hour project.

You can use the sse2 app for sure (since that is the one I use). Download & unzip it. Turn Boinc off. Insert the two files into your Boinc folder ...\BOINC\projects\milkyway.cs.rpi.edu_milkyway . Restart Bonic. It should load the app in the first lines: Milkyway@home|Found app_info.xml; using anonymous platform. Then get work.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

A Very interesting discussion ...

But I have not seen where anyone asked the big question.

Why is the server dishing out these large tasks to slow machines?????

I have read that the server software should not send tasks to hosts that are unable to complete them in time.

Obviously the server settings are incorrect!!!!

The poor 'set n forget' hosts will be continuously returning WU's too late.

I do not see how this helps the project - continuously sending out tasks to slow hosts that 'time out' and have to be resent.

What a waste of time and energy!!!

I got the sse2 one - I also took it out of the folder and put the contents directly in the MW folder this time. The "time left" will get longer for a bit but then drop by at least a minute and a half. Overall it seems it will get done in about 1/4 the time given (a bit over 9 hours.)

You should be under 5 hours as Cluster said. Mine are usually 5 to 5 1/4 hours. It should take a few to get the estimated time correct.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

It's been running now about 2 hours 45 min, time left is a bit over 24 hours. It started at just over 37 hours. At that rate it will likely be a little more than 8 hours total. I do run 2 projects at once (other one switches between World Community and Climate Prediction. I'm also usually on my computer (I'm working on getting disability and still getting unemployment) and turn it off when I go to bed or am away from home.)

Well, the previously mentioned one is due tomorrow at 4:30 pm about - nearly 12 hours so far and less than an hour to go. A new one also started with a given time of about 8 hours. The app apparently works.

Now it seems the tasks are smaller and a week is given to complete.

Now it seems the tasks are smaller and a week is given to complete.

Yes some are now.


I think part of your problem with slow workunits is in your completed wus it lists 12 cores for your p4. Do you have 11 virtual cores setup? Whatever is causing that is why you are running these tasks so slow. The 10 hour task should have been under 5. Just run the one and I'm sure you'll come out with more credits.

<core_client_version>6.2.28</core_client_version>
<![CDATA[
<stderr_txt>
Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40001 GHz (43860ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40001 GHz (37214ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40006 GHz (5402ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40005 GHz (1607ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40002 GHz (4210ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40005 GHz (33629ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40005 GHz (7542ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40005 GHz (3005ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40003 GHz (5135ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40002 GHz (4120ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40003 GHz (2785ms)

Running Milkyway@home version 0.20 (Win32, SSE2) by Gipsel
CPU: Intel(R) Pentium(R) 4 CPU 3.40GHz (2 cores/threads) 3.40002 GHz (3641ms)

WU completed. It took 2299.16 seconds CPU time and 2552.73 seconds wall clock time @ 3.40003 GHz.

</stderr_txt>

Your cpu time is given seperately as

CPU time 53879.67

, but in the list it gives 2300 secs.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Do you have 11 virtual cores setup?

I have no idea what that means.

Do you have 11 virtual cores setup?

I have no idea what that means.

Then probably not. Some reason Boinc is using/thinks you have multiple cores, and that could account for your extralong runtimes.

---

Doesn't expecting the unexpected make the unexpected the expected?
If it makes sense, DON'T do it.

Well, my computer died Wednesday - fried motherboard. I got a new one hooked up yesterday which is 5+ years newer and much faster so things might work better with BOINC now.