61) Message boards : ATLAS Application : ATLAS vbox v1.10 (Message 7325)
Posted 14 Jun 2022 by David Cameron
Post:
Thanks for the feedback! The tasks here are very short, just simulating a couple of events instead of 200 on the production server.

One thing you should notice is that the start up time is faster since there is no copy of the vdi image to the tasks working area.
62) Message boards : ATLAS Application : ATLAS vbox v1.10 (Message 7322)
Posted 13 Jun 2022 by David Cameron
Post:
Thanks to computezrmle there is a new VirtualBox wrapper available which takes advantage of "differencing images". See details here: https://github.com/BOINC/boinc/issues/4602

v1.10 for Windows is using this new wrapper, so please test it out and give feedback!
63) Message boards : ATLAS Application : ATLAS native 1.07 and vbox 1.08 (Message 7284)
Posted 22 Dec 2021 by David Cameron
Post:
The vdi file is identical to the previous version, but I'm not sure if BOINC is clever enough to know this and not download it again...
64) Message boards : ATLAS Application : ATLAS native 1.07 and vbox 1.08 (Message 7278)
Posted 21 Dec 2021 by David Cameron
Post:
Hi all,

We just released two new versions of the ATLAS app:

Native 1.07

To reduce the time and disk space used copying the input files around, the copy of data from slots/n/shared to slots/n has been replaced by a hard link.

Vbox 1.08 (Windows)

Some problems were reported recently on the LHC@Home forum which may be related to the old vboxwrapper version used by ATLAS tasks. This new version uses the latest version of vboxwrapper 26203.
65) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7163)
Posted 25 Mar 2021 by David Cameron
Post:
Long tasks are now available on LHC@Home (in a beta application): https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5625
66) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7162)
Posted 24 Mar 2021 by David Cameron
Post:
finished the tasks of the first wave of March 18 and can no longer get new ones
are there any restrictions on host parameters?

since we are already talking about the official release - note new tasks do not count as ATLAS for badges


Good point, I'll think about how to include the credits for ATLAS badges.

I'm going to stop submitting new tasks here now and set up the app on the production LHC@Home server. I'll announce on the message boards there when it's ready.

Thanks everyone for all your help here!
67) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7157)
Posted 24 Mar 2021 by David Cameron
Post:
Thanks for testing this out. The computations in each task are the same as in the regular tasks on LHC@Home but process 5 times as much data. The idea of testing here was to see if anything strange happens and to see if anything needs tweaked in the BOINC server configuration to handle these tasks. It seems like it mostly runs smoothly so we can go to the next step of releasing the app officially on LHC@Home.

The long tasks are mainly for powerful dedicated machines so old laptops would be better put to work on the regular short tasks. People should only opt-in to the long task app if their machines can handle it.
68) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7155)
Posted 24 Mar 2021 by David Cameron
Post:
This should be better now. I have removed all the tasks with missing input files.
69) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7150)
Posted 23 Mar 2021 by David Cameron
Post:
Let's say you have a 32 core machine. Is it better to run a single task using 32 cores? Or 4 tasks using 8 cores each? Or 8 tasks using 4 cores each? I have tried all three configurations and the credits per hour work out to be roughly the same. As far as I can tell, there is no advantage to me regardless of configuration.

So my question is, does the project have a preference?


Thanks for trying this out! It seems your 32-core tasks worked fine. As computezrmle said, the choice of cores should be based on your individual disk and memory set up. A single-core task uses ~3GB of memory and your 32-core tasks used 7GB because a lot of memory is shared between the different processes. In addition each task requires 8-10GB of disk space. There is a slight efficiency gain in running fewer cores but it is probably only a few percent.
70) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7149)
Posted 23 Mar 2021 by David Cameron
Post:

David Cameron wrote:
ATLAS systems cancel any tasks which have been queued for more than two days

I would suggest to take this as a hint and use a setup that allows a task to finish within 1-1.5 days.
Beside that it's a matter of personal preference.


Sorry I wasn't clear here before. I meant that unsent tasks are cancelled after 2 days. Once a task is taken by a volunteer the usual 1 week deadline applies, and then if the deadline expires the task is available for another host.
71) Message boards : ATLAS Application : ATLAS very long simulation v1.01 download errors (Message 7140)
Posted 22 Mar 2021 by David Cameron
Post:
Most likely the "task-has-been-revoked" error.
Just check if your wingcomputers get the same error.


Yes, this is the problem. ATLAS systems cancel any tasks which have been queued for more than two days so that is what is happening here. There were too many tasks queued for too long, so I've configured a lower number of queued tasks in order that this doesn't happen as often.
72) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7105)
Posted 19 Mar 2021 by David Cameron
Post:
I've increased the flops estimate by a factor 5 for new tasks. I think BOINC's estimation is also based on previous tasks and since they have been very short the estimation is very small.

Note that for this app the max CPUs is 48, so if you have a big machine please give 48-core tasks a try!
73) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7098)
Posted 19 Mar 2021 by David Cameron
Post:
There are more long tasks in the queue now. I finished one task successfully on one of my hosts and it took 12 hours using 4 cores.
74) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7096)
Posted 18 Mar 2021 by David Cameron
Post:
There are still some short test tasks in the system which process 2 events. I've stopped submitting those short tasks now. The first batch of 10 real long tasks with 1000 events is now available. Once those are finished there will be more available.
75) Message boards : ATLAS Application : ATLAS long simulation 1.01 (Message 7094)
Posted 18 Mar 2021 by David Cameron
Post:
Hi all,

We are testing long tasks which run over 1000 events instead of 200, so please select "ATLAS very long simulation" from the app list and give us feedback here. It is currently native linux-only and a minimum of 4 CPUs is required.
76) Message boards : ATLAS Application : Atlas native 1.03 (Message 7069)
Posted 10 Jul 2020 by David Cameron
Post:
I started redirecting some test tasks here, but the problem is that they are cancelled upstream before they run and this means the input data is deleted from BOINC servers. I will stop submitting the tasks to avoid the failures.
77) Message boards : ATLAS Application : ATLAS native 1.01 (Message 6986)
Posted 6 Feb 2020 by David Cameron
Post:
Thanks, the new patch is in 1.03.
78) Message boards : ATLAS Application : ATLAS native 1.01 (Message 6981)
Posted 6 Feb 2020 by David Cameron
Post:
Ok, I just released 1.02 with this change.
79) Message boards : ATLAS Application : ATLAS native 1.01 (Message 6978)
Posted 6 Feb 2020 by David Cameron
Post:
Hmm, it doesn't seem to work for me: https://lhcathomedev.cern.ch/lhcathome-dev/result.php?resultid=2868679
80) Message boards : ATLAS Application : ATLAS native 1.01 (Message 6976)
Posted 6 Feb 2020 by David Cameron
Post:
computezrmle has provided a better method of stderr logging for the new native script, which is now in version 1.01. I have injected a few test tasks here to make sure everything is ok before moving to the production server.


Previous 20 · Next 20


©2024 CERN