Beta testing - Calibrated Boinc core

[trux]

OK, thanks, Mr. Pernod - good to know.

[Mr.Pernod]

if you want me to stop testing, just say so, otherwise I will be bugging you with this kind of stuff for a long time

[trux]

No problem, Mr.Pernod, I am happy for the feedback. I have no chance to test everything, especially because I run just very few backup projects.

[Neil]

Running core 5.3.11.tx31

On my HT machine when i have the priority project set for seti and there is no work, it requests 1 second of work from the backup for one half of the processor, which is working as intended, but it won't request work for the second half. Is there any way to get it to pull another for ht machines? or shall i just add another backup project?


EDIT: Never mind i just made my backup project my second "priority project" , did the trick

[Honza]

Well, when BOINC realize that there is no new work for another core, it should request new work [when contacting scheduler].
I guess it's a bit tricky to request new work for just 2 WUs among the wide porfolio of projects based on seconds and benchmark numbers

[trux]

Neil: Not sure if assigning as the second priority project will work as you expect, especially regarding the downloading of work. Correctly, it should not - the behaviour for priority and non-priority projects is different, even if they are not assigned as the first one. I'll have to add handling of multi-cpu systems for the backup projects

[Mr.Pernod]

my dedicated SETI host just hosed 200+ results with this error:

Kód: Vybrat vše

<core_client_version>5.3.11.tx31</core_client_version>
<message>There are no child processes to wait for. (0x80) - exit code 128 (0x80)
</message>

directly after it finished this result

Kód: Vybrat vše

<core_client_version>5.3.11.tx31</core_client_version>
<real_cpu_time>124</real_cpu_time>
<corrected_cpu_time>162</corrected_cpu_time>
<corrected_Mfpops>3206.9</corrected_Mfpops>
<stderr_txt>
Windows optimized S@H application by Crunch3r
Improvements by Tetsuji Maverick Rai, Hans Dorn, Harold Naparst, Ned Slider, Crunch3r, trux,...
$Using cache implementation by Hans Dorn $
$Build: Windows SSE2 Intel Pentium 4 V2.10 by Crunch3r $
$Rev: 166.10 Windows SSE2 Intel Pentium 4 V2.10 $
$Internal: +16SMA;+PA;+IA;+EFL-P4P $
SETI@Home Informational message -9 result_overflow
NOTE: The number of results detected exceeds the storage space allocated.

</stderr_txt>

any idea if this could be related to 5.3.11, the optimzed application or is it an OS-related error?

[rsisto]

Im having troubles running it in a PC with Windows Server 2003, does it run under this OS?

Thanks
Roberto

[trux]

2 Mr. Pernod: this problem is unrelated to the core client, it is an error in the app (most likely a noisy WU)

2 Roberto: yes, it does, with no problems. Myself, I am testing on XP and W2003. What the "problems" exactly mean?

[rsisto]

2 Roberto: yes, it does, with no problems. Myself, I am testing on XP and W2003. What the "problems" exactly mean?

I copied the new boinc exe and dll and visual studio dll in the zip file to boinc directory. When trying to start again the service it does not start and I get an error saying that the configuration is not right. It think it has to do with some missing or incorrect dll version.

[trux]

Yes, it is certainly some missing lib. Unfortunately I have the Visual Studio installed on the 2003 machine too, so do not catch such messages. Can you try starting boinc.exe from a command line window to see if you get a more verbose message about the missing lib? Or maybe looking into the Events Viewer or the stderrdae.txt may reveal the missing lib name.

[Mr.Pernod]

2 Mr. Pernod: this problem is unrelated to the core client, it is an error in the app (most likely a noisy WU)

I know the -9 result_overflow is most likely a noise result, I have had these before.
What never happened before is that the next 200+ results that were started directly after the noisy one were trashed with the <message>There are no child processes to wait for. (0x80) - exit code 128 (0x80) </message> error.
From what I can see it looks like a timing problem somewhere between BOINC and the science application as it resumed normal operation after I restarted BOINC.

[Mr.Pernod]

I gave SIMAP another test run on the dual MP, but it is the same issue again.
realistic claims should be between 5 to 10 credits for the current results, but it is up to 30+ claimed and climbing in less then 10 hours.

[trux]

In that case they probably have the unit value estimation set wrongly. Simply do not enable the calibration at that project.

[Mr.Pernod]

In that case they probably have the unit value estimation set wrongly. Simply do not enable the calibration at that project.

seems like the only option for SIMAP, maybe put a note on the downloadpage?