Development for LHC@home

The new attributes in the project preferences are just cosmetic for now. Nothing acts upon them. Some of the possibilities are being discussed in this thread.

EDIT: I don't understand why BOINC assigns 1.5 cores.

What do you gave in your cc_config.xml in the BOINC directory? I've sometimes had to play with that to get BOINC to "simulate" the proper number of cores.

---

EDIT: I don't understand why BOINC assigns 1.5 cores.

It's configured by Rasputin in his app_config.xml.

In principle the partial use of a core was designed for supporting GPU's.

Example: you could have 3 GPU's or 3 tasks on 1 GPU claiming 1 CPU-core.

Claiming for 3 tasks 3 times 1.5 cores is rounded down to 4.

It's configured by Rasputin in his app_config.xml.

No, it is not. It does that, even when i reset the project.

The new attributes in the project preferences are just cosmetic for now. Nothing acts upon them. Some of the possibilities are being discussed in this thread.

EDIT: I don't understand why BOINC assigns 1.5 cores.

If you haven't been using the XML plan class stuff it is probably falling back to some hard coded values in the compiled server scheduler dating back years at this point.

See:
https://github.com/BOINC/boinc/blob/master/sched/sched_customize.cpp#L888

If you all want to switch over to the XML based plan-class we can go ahead and remove the hard-coded stuff put in for the theory app.

I guess we always assumed the intuitional knowledge about what was added for the various CERN projects would be passed on from one generation to the next.

----- Rom

I have a case, where i am running a 4 core tasks. For whatever reason, it decided after 3h or so, to discontinue 3 of them and runs only on one.
( as if it had hit the 12h mark, which it has not.)

I have a case, where i am running a 4 core tasks. For whatever reason, it decided after 3h or so, to discontinue 3 of them and runs only on one.
( as if it had hit the 12h mark, which it has not.)

I had the same thing yesterday with a 2 core task when one of the two job streams went quiet while the other continued to run jobs.

I have a case, where i am running a 4 core tasks. For whatever reason, it decided after 3h or so, to discontinue 3 of them and runs only on one.
( as if it had hit the 12h mark, which it has not.)

I had the same thing yesterday with a 2 core task when one of the two job streams went quiet while the other continued to run jobs.

I had that this morning, but no multi-core VM's.
I had 4 single core VM's running. 2 of them out of jobs, but they were not killed after 5 minutes idling.
Looking into the results, I see something has change to the logging (DEBUG-info) too.
http://lhcathomedev.cern.ch/vLHCathome-dev/result.php?resultid=229639
http://lhcathomedev.cern.ch/vLHCathome-dev/result.php?resultid=229103
The other 2 single core VM's still running and getting jobs.

The new attributes in the project preferences are just cosmetic for now. Nothing acts upon them. Some of the possibilities are being discussed in this thread.

EDIT: I don't understand why BOINC assigns 1.5 cores.

If you haven't been using the XML plan class stuff it is probably falling back to some hard coded values in the compiled server scheduler dating back years at this point.

See:
https://github.com/BOINC/boinc/blob/master/sched/sched_customize.cpp#L888

If you all want to switch over to the XML based plan-class we can go ahead and remove the hard-coded stuff put in for the theory app.

I guess we always assumed the intuitional knowledge about what was added for the various CERN projects would be passed on from one generation to the next.

----- Rom

Just to be clear, removing line:
https://github.com/BOINC/boinc/blob/master/sched/sched_customize.cpp#L894

Should setup a multi-threaded VM to use as many cores as preferences allow and not cap it at 1.5 CPU(s).

----- Rom

The new attributes in the project preferences are just cosmetic for now. Nothing acts upon them. Some of the possibilities are being discussed in this thread.

EDIT: I don't understand why BOINC assigns 1.5 cores.

If you haven't been using the XML plan class stuff it is probably falling back to some hard coded values in the compiled server scheduler dating back years at this point.

See:
https://github.com/BOINC/boinc/blob/master/sched/sched_customize.cpp#L888

If you all want to switch over to the XML based plan-class we can go ahead and remove the hard-coded stuff put in for the theory app.

I guess we always assumed the intuitional knowledge about what was added for the various CERN projects would be passed on from one generation to the next.

----- Rom

Just to be clear, removing line:
https://github.com/BOINC/boinc/blob/master/sched/sched_customize.cpp#L894

Should setup a multi-threaded VM to use as many cores as preferences allow and not cap it at 1.5 CPU(s).

----- Rom

I've gone ahead and removed the old CERN Theory policy from the stock source code.

See:
https://github.com/BOINC/boinc/commit/eac15d6982fc34ca3c781e0580d45b8f32039c22

----- Rom

Thanks, Rom.
How is determined now, how many cores to be used?
By the project preference file? What if it is undefined?

Thanks, Rom.
How is determined now, how many cores to be used?
By the project preference file? What if it is undefined?

I believe g_req->effective_ncpus is calculated from the global_prefs.

See:
http://lhcathomedev.cern.ch/vLHCathome-dev/prefs.php?subset=global

I would need to hunt down where exactly, I don't have a Visual Studio project for the server code. So I cannot easily cheat by having Visual Studio do all the hunting.

----- Rom

Thanks, Rom.
How is determined now, how many cores to be used?
By the project preference file? What if it is undefined?

...

Anyway be careful as Laurence hasn't yet fully implemented the new project preferences for number of cores and tasks… You may not get what you ask for until he's finished coding it.

I know, it isn't functional, yet.

Would be nice, to get a message, when it is.