Hey everyone,

I just wanted to let everyone know that new Nbody runs have been put up on MW@home. The last set of runs started converging nicely, so we are confident this next set of runs will converge as well. The names of these runs are:
-de_nbody_07_29_2020_v176_40k__data__1
-de_nbody_07_29_2020_v176_40k__data__2
-de_nbody_07_29_2020_v176_40k__data__3

Please let us know if you have any trouble with these runs. Thank you all for your continued support.

-Eric

Yes

What can I do to get the NBody tasks? I only receive Separation tasks. I have been with Milkyway@home for a while I have never received NBody jobs.

Have you configured your account to accept cpu and N-body work?
https://milkyway.cs.rpi.edu/milkyway/prefs.php?subset=project

---

Got a hang on de_nbody_07_29_2020_v176_40k_data_3_1596028202__76964_1
stops making progress for hours (task should be 5 to 10 minutes). Was not problem on previous groups of Nbody. After short suspend runs to completion in reasonable time. Best if these sorts of things errored out rather than tie up all Nbodys until human intervention.

---

Also when Nbody gets buffed up, ability to run it on the gpu would be good, as a lot of machines these days have Radeon VII, Titan V, etc. that can really chew through double precision tasks and have a lot of memory, not being partly crippled for compute tasks like purely gameing cards. They don't do so well on the separations as they might as each only takes 12 to 20 sec and a fair amount of time is spent on loading and unloading them in the gpu, and the number sent gets used up fast and there is no work available for a while. Probably would not work well for Nbody if quad or greater precision is needed.

---

Does it mean that the NBody application works on CPU only? If not, how can I enable the app on my account? I cannot see any option to do that.

Does it mean that the NBody application works on CPU only? If not, how can I enable the app on my account? I cannot see any option to do that.

CPU only.

I have a 12 core (24 thread) CPU and am getting 16 thread tasks, so its only running one at a time. CPU utilization is very low for a couple minutes, and then peaks around 50% for the rest of the task. Is there an app_config that would allow specifying running 2 x 12 or 3 x 8 thread tasks? It would improve efficiency greatly.

Edit I think I figured it out.

<app_config>
<app>
<name>milkyway_nbody</name>
<max_concurrent>3</max_concurrent>
</app>
<app_version>
<app_name>milkyway_nbody</app_name>
<plan_class>mt</plan_class>
<avg_ncpus>8</avg_ncpus>
</app_version>
</app_config>

Yes. I run only one core per work unit. They don't take much memory anyway (only 16 MB).
I don't see the point of running more cores. They just disrupt the other projects.

But I run mine 24/7, and they sometimes run for over a day. If you need to complete them faster, then of course more cores would help.
But I think you lose a little efficiency with more cores.

<app_config>
  <project_max_concurrent>16</project_max_concurrent>
  <app>
    <name>milkyway_nbody</name>
    <fraction_done_exact/>
    </app>
  <app_version>
    <app_name>milkyway_nbody</app_name>
    <plan_class>mt</plan_class>
    <avg_ncpus>1</avg_ncpus>
  </app_version>
</app_config>

I have a 12 core (24 thread) CPU and am getting 16 thread tasks, so its only running one at a time. CPU utilization is very low for a couple minutes, and then peaks around 50% for the rest of the task. Is there an app_config that would allow specifying running 2 x 12 or 3 x 8 thread tasks? It would improve efficiency greatly.

Edit I think I figured it out.

<app_config>
<app>
<name>milkyway_nbody</name>
<max_concurrent>3</max_concurrent>
</app>
<app_version>
<app_name>milkyway_nbody</app_name>
<plan_class>mt</plan_class>
<avg_ncpus>8</avg_ncpus>
</app_version>
</app_config>

Actually the correct method of defining mt app is provided in the BOINC configuration document.
https://boinc.berkeley.edu/wiki/Client_configuration#Application_configuration

They provide an example.

[<app_version>
<app_name>Application_Name</app_name>
[<plan_class>mt</plan_class>]
[<avg_ncpus>x</avg_ncpus>]
[<ngpus>x</ngpus>]
[<cmdline>--nthreads 7</cmdline>]
</app_version>]

So if you want to run 3 tasks with 8 cores each your app_config should look something like this.

<app_config>
<app>
<name>milkyway_nbody</name>
<max_concurrent>3</max_concurrent>
</app>
<app_version>
<app_name>milkyway_nbody</app_name>
<plan_class>mt</plan_class>
<cmdline>--nthreads 8</cmdline>
</app_version>
</app_config>

The --nthreads parameter is the actual controller of how many cpu threads are utilized per each task.

---

If you want to receive NBody jobs, click on the "Project" tab and select "Preferences". From there, you can click on "Edit preferences" and check the box to accept NBody jobs.

Does it mean that the NBody application works on CPU only? If not, how can I enable the app on my account? I cannot see any option to do that.

Currently NBody only runs on CPU. We are currently working on releasing a GPU application.

Actually the correct method of defining mt app is provided in the BOINC configuration document.
https://boinc.berkeley.edu/wiki/Client_configuration#Application_configuration

They provide an example.

[<app_version>
<app_name>Application_Name</app_name>
[<plan_class>mt</plan_class>]
[<avg_ncpus>x</avg_ncpus>]
[<ngpus>x</ngpus>]
[<cmdline>--nthreads 7</cmdline>]
</app_version>]

So if you want to run 3 tasks with 8 cores each your app_config should look something like this.

<app_config>
<app>
<name>milkyway_nbody</name>
<max_concurrent>3</max_concurrent>
</app>
<app_version>
<app_name>milkyway_nbody</app_name>
<plan_class>mt</plan_class>
<cmdline>--nthreads 8</cmdline>
</app_version>
</app_config>

The --nthreads parameter is the actual controller of how many cpu threads are utilized per each task.

Thanks a lot! Looks like it solved my issue as well!

By default the milkyway_nbody process was single instance, distributed over all my threads, taking several days to finish. The CPU utilisation was almost nada, ie idling around 1% overall.

The latter variant of the xml-file seems to work well in my case. After Jim's advice, I restricted it to only a single thread per process, and increased the maximum number of parallel processes accordingly.

After saving the app_config.xml in the projects app-data folder, I suspended the milkyway project, waited a minute or so, then resumed. The ongoing process then resumed without any visible change, and finished in about 25 mins what it had spent several days doing already. Full CPU utilisation on all cores/threads.

So, great! However, weird that this isn't solved on the server side as others have pointed out. It's a real time waste for the other projects when the milkyway instances are run so inefficiently..

For reference, here's my C:\ProgramData\BOINC\projects\milkyway.cs.rpi.edu_milkyway\app_config.xml

<app_config>
	<app>
		<name>milkyway_nbody</name>
		<max_concurrent>8</max_concurrent>
	</app>
	<app_version>
		<app_name>milkyway_nbody</app_name>
		<plan_class>mt</plan_class>
		<cmdline>--nthreads 1</cmdline>
	</app_version>
</app_config>

Hello.
I had to stop using your project because is hugging all processors, living nothing for other projects.
Please allow in the project preferences the option to specify the number of processors and how many units to download at a time.
If this happens I will join again and maybe you get more volunteers.
Thank you.

Hi. Es, gibt, einen, Neuen, Treiber, für, NVidia GPU - 460.89 notebook win10 64bit , auf, www.nvidia.de, herunterladen, Boinc, Beenden, und Updaten. :-).

Treiber, installieren, Boinc Starten, und Benchmark, durchfühtren.:-)

MfG :-)

Sascha Becker  \ oetker201 :-)