Development for LHC@home

The 2 tasks I requested this morning created 2 dual core VM's, but both are running 1 single core job.

4-Core CMS-Task, on a Win11pro-Workstation.
https://lhcathomedev.cern.ch/lhcathome-dev/results.php?userid=378

I have a couple more running and finished late tonight

https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3313349
Run time 17 hours 16 min 10 sec
CPU time 2 days 11 hours 43 min 19 sec
Validate state Valid
Credit 243.77 (4 core)

How hard would it be to make native version of CMS?

Not that easy. To my knowledge we have running versions of CMSSW on x64, NVidia (CUDA), XeonPhi (Kights Landing), and Arm. The original idea to run in VirtualBox was because the x64 architecture existed in Windows, Linux, and MacOS, so we could deploy to those environments with minimal additional effort. Given that the effort for years has been mostly Laurence (on the BOINC side), me from CMS, and lately Federica from CMS, I don't think we can support a large number of environments. You also have to realise that there are a heck of a lot of other things running behind the scenes, CVMFS and containerisation to name but two, so that folding them into a VM rather than expecting Volunteers to maintain them as well makes participation rather easier for Joe Average.

---

Ivan,
we have 4-Core CMS and 1-Core CMS.
When this is transfered to production and running well, it's ok.

My i7-8650U laptop and i7-4790 desktop both seem to be using between 3.5 and 4 CPU cores on their 4-core MT tasks. These computers are not running anything else.

My Xeon E5-2699 v3 is using almost exactly 4 CPU cores pretty consistently on its 4-core MT tasks. I do have other BOINC projects running on the other CPU cores on this computer.

My i7-8650U laptop and i7-4790 desktop both seem to be using between 3.5 and 4 CPU cores on their 4-core MT tasks. These computers are not running anything else.

My Xeon E5-2699 v3 is using almost exactly 4 CPU cores pretty consistently on its 4-core MT tasks. I do have other BOINC projects running on the other CPU cores on this computer.

Fardringle you need to go in your account here under *Preferences for this project* and set your CMS tasks to run with 4 cores per task because all of the ones you have been running are only using one core per task.

You can see that when you check your tasks STDERR where it says * Setting CPU Count for VM. (1)*
And when you do that and run a task here you will see that the CPU time is much longer than the Run time when you check your finished tasks.

They will then say they finish in close to 17hrs Run time and the CPU time will be over 2 days

Given that the effort for years has been mostly Laurence (on the BOINC side), me from CMS, and lately Federica from CMS, I don't think we can support a large number of environments. You also have to realise that there are a heck of a lot of other things running behind the scenes, CVMFS and containerisation to name but two, so that folding them into a VM rather than expecting Volunteers to maintain them as well makes participation rather easier for Joe Average.

Well here at -dev we have a few of us with plenty of VB and GPU experience that will be happy to test doing that here in the future Ivan.

Of course we do have members here that might be on that Average Joe list so you could just ask a few of us to do that or even make another url like the one we used to have for Atlas-Test if this is ever made possible to run tests for since that is how we did it with Atlas.......and CMS when we used call this place CERN CMS-dev
And of course way back in the beginning with T4T

Todays results (looks like they are ready)
https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3313414
Run time 16 hours 54 min 2 sec
CPU time 2 days 11 hours 39 min 50 sec
Validate state Valid
Credit 216.70
Device peak FLOPS 16.68 GFLOPS

https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3313433
Run time 17 hours 15 min 29 sec
CPU time 2 days 13 hours 9 min 30 sec
Validate state Valid
Credit 278.10
Device peak FLOPS 16.63 GFLOPS

https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3313421
Run time 17 hours 49 min 7 sec
CPU time 2 days 13 hours 13 min 46 sec
Validate state Valid
Credit 228.52
Device peak FLOPS 16.68 GFLOPS

I think you are looking at older tasks. I switched to 4 thread tasks about 2 days ago, and all tasks since then are registering and using 4 threads. And the result logs and actual CPU usage confirm that.

If you are looking at my task history, everything returned after 29 Mar 2024, 8:38:12 UTC is running the new 4-thread CMS tasks.

Yes as you can see your first ones are after I posted that so all I could see were the ones that you finished before that time on the 29th

I see the ones you now have finished as multi-core

---

Mad Scientist For Life

https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3313451


Yeah the Credits after all these years here are still a joke BUT I'm sure if it is like this in production you will get all of the members to jump on this one.

hoping to be the first with 20k ;-).

hoping to be the first with 20k ;-).

You better hurry Axel and it looks like it takes 32 cores for over 19K
https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3312030

But then if you are lucky you can get 17K with only 4 cores
https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=3313663

It would be nice to change the Limit of 10 days to get this 25k (+5k) ;-)
Ivan, have your Team more of this CMS-Tasks?
https://home.cern/news/news/physics/first-observation-photons-taus-proton-proton-collisions-cms

My personal record goal is 4000 Valid CMS here with less than 475 Credits each

It looks like our new batch of CMS are back to the old version without actually running multi-core so far.

https://lhcathomedev.cern.ch/lhcathome-dev/server_status.php

New batch of CMS so lets see if it will run multi-core or not.

It doesn't