Message boards : Sixtrack Application : SIXTRACKTEST
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 5 · 6 · 7 · 8 · 9 · Next

AuthorMessage
Profile Ray Murray
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 281
Credit: 11,859,285
RAC: 1
Message 39822 - Posted: 4 Sep 2019, 12:11:41 UTC - in response to Message 39653.  

I have 2 sixtracktests that I returned on 15th Aug. My wingman aborted them just as they were about to timeout on 29th Aug. With lots of sixtracktests in the queue, and the resends going to the back of that queue, they are still waiting to be resent.
If there are similar delays in returns of split 10^7 jobs, it may take a looong time to get the full job results returned.
CPDN, which runs for months, uses a "Trickle" system to upload intermediate data. Could something similar be implemented here for these so that the full job could be run to completion but the intermediate results could be used if necessary if the job is lost or to pinpoint any failure?
I generally run 24/7 so my longest job at 4.6 days wasn't a problem for me.
ID: 39822 · Report as offensive     Reply Quote
Profile Magic Quantum Mechanic
Avatar

Send message
Joined: 24 Oct 04
Posts: 1114
Credit: 49,501,728
RAC: 4,157
Message 39834 - Posted: 5 Sep 2019, 21:59:12 UTC

(I haven't had time to check this entire thread)

It looks like X86 does not like running the Test v502.05 (sse2) tasks but can run the SixTrack v502.05 and never got any of the v502.05 (avx) so no idea if that would run on a X86 so I switched the X86 back to regular Sixtracks and Theory VB tasks.
ID: 39834 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 15 Jun 08
Posts: 2386
Credit: 222,932,171
RAC: 137,662
Message 39835 - Posted: 6 Sep 2019, 5:44:15 UTC - in response to Message 39834.  

... and never got any of the v502.05 (avx) so no idea if that would run ...

At least this can be explained.

This is the only host on your list that is running x86_32 instead of x86_64:
https://lhcathome.cern.ch/lhcathome/show_host_detail.php?hostid=10447575

It's CPU doesn't have AVX extensions:
http://www.cpu-world.com/CPUs/K10/AMD-Phenom%20II%20X3%20720%20-%20HDX720WFK3DGI%20(HDX720WFGIBOX).html
ID: 39835 · Report as offensive     Reply Quote
Alessio Mereghetti
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 29 Feb 16
Posts: 157
Credit: 2,659,975
RAC: 0
Message 39839 - Posted: 6 Sep 2019, 9:00:57 UTC - in response to Message 39822.  

Hi Ray,
thanks for the suggestion. We are looking into something of this kind, i.e. we split the job in sub-steps, and we collect the partial results together with the checkpoint-restart files.
The machinery in SixTrack is almost ready, for BOINC the process will be completely transparent; we are now in the process of modifying the software on the side of the scientists - it will come in few months, with a major code re-writing. Slowly but moving :)
Happy crunching
ID: 39839 · Report as offensive     Reply Quote
Profile Ray Murray
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 281
Credit: 11,859,285
RAC: 1
Message 39868 - Posted: 8 Sep 2019, 17:15:18 UTC
Last modified: 8 Sep 2019, 17:21:44 UTC

Those 2 x 10^7 tasks from 15 Aug have at last been resent and I've got couple of 10^6 _2 myself so we must have gotten to the end of the queue.
ID: 39868 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 1268
Credit: 8,421,616
RAC: 2,139
Message 39870 - Posted: 8 Sep 2019, 17:48:01 UTC - in response to Message 39868.  

Those 2 x 10^7 tasks from 15 Aug have at last been resent .....
I've got a *_4 resent of a 10^7-task from the initial replication of 2 one month ago from the 8th of August.

https://lhcathome.cern.ch/lhcathome/workunit.php?wuid=120081701 Another 4 days of work. :)
ID: 39870 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 1268
Credit: 8,421,616
RAC: 2,139
Message 39911 - Posted: 12 Sep 2019, 16:47:49 UTC - in response to Message 39870.  

I've got a *_4 resent of a 10^7-task from the initial replication of 2 one month ago from the 8th of August.

https://lhcathome.cern.ch/lhcathome/workunit.php?wuid=120081701 Another 4 days of work. :)
Finally finished ;)
ID: 39911 · Report as offensive     Reply Quote
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 857
Credit: 1,619,050
RAC: 0
Message 39914 - Posted: 13 Sep 2019, 8:45:16 UTC - in response to Message 32365.  

Thanks, excellent feedback. Memory requirement could be reduced
by tracking fewer particles in the task. Memory management in SixTrack
is now dynamic. There must be something wrong with our memory estimate
I think. In the first place we should avoid using all your memory and a
potential deadlock........We shall see. Eric.
ID: 39914 · Report as offensive     Reply Quote
Profile Magic Quantum Mechanic
Avatar

Send message
Joined: 24 Oct 04
Posts: 1114
Credit: 49,501,728
RAC: 4,157
Message 39921 - Posted: 14 Sep 2019, 0:24:44 UTC - in response to Message 39914.  

Thanks, excellent feedback. Memory requirement could be reduced
by tracking fewer particles in the task. Memory management in SixTrack
is now dynamic. There must be something wrong with our memory estimate
I think. In the first place we should avoid using all your memory and a
potential deadlock........We shall see. Eric.

No problems with mine Eric
Ran about 1200 of them so far and have another 444 to run and have been easy to finish way before the deadline.
Have them running X8 with 16GB ram and 24GB ram and X4 with 12GB ram and over no problems running X3 with 8GB ram on the old 3-core.

No errors

(good to see you here too Eric)

-Samson
ID: 39921 · Report as offensive     Reply Quote
Erich56

Send message
Joined: 18 Dec 15
Posts: 1686
Credit: 100,372,896
RAC: 102,138
Message 39923 - Posted: 14 Sep 2019, 6:15:41 UTC - in response to Message 32505.  

I just downloaded a Sixtracktest task (for the first time), it runs perfectly, memory use is 40MB constantly.
ID: 39923 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,109,667
RAC: 104,249
Message 40548 - Posted: 20 Nov 2019, 8:38:25 UTC

This is one of two - waiting to finishing sixtracktest:
https://lhcathome.cern.ch/lhcathome/workunit.php?wuid=123954511
ID: 40548 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,109,667
RAC: 104,249
Message 40550 - Posted: 20 Nov 2019, 12:06:22 UTC - in response to Message 40548.  

This is one of two - waiting to finishing sixtracktest:
https://lhcathome.cern.ch/lhcathome/workunit.php?wuid=123954511

Is now finished, thank you.
ID: 40550 · Report as offensive     Reply Quote
Profile Ray Murray
Volunteer moderator
Avatar

Send message
Joined: 29 Sep 04
Posts: 281
Credit: 11,859,285
RAC: 1
Message 40678 - Posted: 26 Nov 2019, 13:10:37 UTC
Last modified: 26 Nov 2019, 13:11:48 UTC

I can't get any of the 88,000 sixtracktest jobs advertised on the Server Status page. I tried disallowing all other apps but still "No tasks available for sixtracktest"
Neither Windows nor Linux hosts getting any but 7 users shown as active returners so are these all Macs, with no other flavours available?
ID: 40678 · Report as offensive     Reply Quote
Alessio Mereghetti
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 29 Feb 16
Posts: 157
Credit: 2,659,975
RAC: 0
Message 40703 - Posted: 27 Nov 2019, 8:41:01 UTC - in response to Message 40678.  

Hi Ray,
thanks for pointing this out, good catch.
Yes, all the tasks pin v5.02.07, which has only Mac exes.
I am plugging in Win/Lin 64 bit exes, just to enlarge the parterre of volunteers and crunch this wave of tasks faster.
You should be able to get tasks soon.
Cheers,
A.
ID: 40703 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,109,667
RAC: 104,249
Message 40741 - Posted: 1 Dec 2019, 3:42:52 UTC

Hi Alessio,
There is a sixtracktest running with 0.5 CPU(?) in a Linux VM.
Is this now possible?
ID: 40741 · Report as offensive     Reply Quote
Alessio Mereghetti
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 29 Feb 16
Posts: 157
Credit: 2,659,975
RAC: 0
Message 40748 - Posted: 2 Dec 2019, 8:07:53 UTC - in response to Message 40741.  

that's interesting, though a-priori not excluded to me - but I have no experience with the VM set-up of the experiments (i.e. CMS, ATLAS)
...unless it is a VM interacting directly with the server - I have a windows one on my Ubuntu laptop...
ID: 40748 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,109,667
RAC: 104,249
Message 40749 - Posted: 2 Dec 2019, 9:45:04 UTC - in response to Message 40748.  

Have watched it yesterday. It works also in sixtrack. Two tasks with always 0.5 CPU's parallel.
No idea why it is active. It's a CentOS7 VM.
https://lhcathome.cern.ch/lhcathome/results.php?hostid=10591881
ID: 40749 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 1268
Credit: 8,421,616
RAC: 2,139
Message 40752 - Posted: 2 Dec 2019, 10:29:19 UTC - in response to Message 40749.  

Two tasks with always 0.5 CPU's parallel.
Number of processors 1
ID: 40752 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 2071
Credit: 156,109,667
RAC: 104,249
Message 40753 - Posted: 2 Dec 2019, 10:35:53 UTC - in response to Message 40752.  

Yes Crystal,
It's a AMD A10 with two true CPU's. Can't make a screenshot, sorry.
In Windows is a Atlas running under Boinc.
The VM is for Cosmology and sixtrack or sixtracktest.
ID: 40753 · Report as offensive     Reply Quote
tullio

Send message
Joined: 19 Feb 08
Posts: 708
Credit: 4,336,250
RAC: 0
Message 40765 - Posted: 3 Dec 2019, 12:24:20 UTC

I too have an AM10-6700. It runs 4 tasks plus one GPU task. It is sold as 4 cores but the Windows Task Manager of Windows 8.1 says it has two cores an 4 logical processors.It runs also a SuSE Linux Tumbleweed Virtual Machine with kernel .5.3.12. I have enlisted it in Science United and it runs mostly Milkyway@home and Asteroids@home plus SETI@home tasks. Their credits are not given to me, this is the main disadvantage of running Science United tasks.
Tullio
ID: 40765 · Report as offensive     Reply Quote
Previous · 1 . . . 5 · 6 · 7 · 8 · 9 · Next

Message boards : Sixtrack Application : SIXTRACKTEST


©2024 CERN