Message boards : ATLAS application : Very long lead times in the final stages of [Atlas Simulation] units
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Marco Del Dottore

Send message
Joined: 21 May 10
Posts: 2
Credit: 117,361
RAC: 0
Message 47198 - Posted: 29 Aug 2022, 14:48:43 UTC

Hello everyone.

I would like to ask your opinion about the behavior of the computation units I have been processing lately.
It seems that in the final stages of processing, when the estimated remaining time is a few tens of minutes, the execution speed decreases a lot. The example figure shows eight minutes remaining, but the actual processing will still take several hours.

https://i.imgur.com/yJEDBOw.png



The configuration that behaves in this way is as follows. Has anyone experienced the same problem and can give me any suggestions? Thank you.


    Libraries: libcurl/7.84.0-DEV Schannel zlib/1.2.12
    OpenCL: Intel GPU 0: Intel(R) HD Graphics 620 (driver version 30.0.101.1340, device version OpenCL 3.0 NEO, 6500MB, 6500MB available, 192 GFLOPS peak)
    Windows processor group 0: 4 processors
    Processor: 4 GenuineIntel Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz [Family 6 Model 142 Stepping 9]
    OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19044.00)
    Memory: 15.87 GB physical, 26.87 GB virtual
    Disk: 465.15 GB total, 168.11 GB free
    VirtualBox version: 6.1.36

ID: 47198 · Report as offensive     Reply Quote
maeax

Send message
Joined: 2 May 07
Posts: 1692
Credit: 113,243,381
RAC: 309,756
Message 47200 - Posted: 29 Aug 2022, 16:58:12 UTC - in response to Message 47198.  

Have you checked Yeti's Checklist in this folder?
ID: 47200 · Report as offensive     Reply Quote
Harri Liljeroos
Avatar

Send message
Joined: 28 Sep 04
Posts: 600
Credit: 36,111,761
RAC: 4,442
Message 47201 - Posted: 29 Aug 2022, 17:54:51 UTC

The final stage (and also the starting phase) use only one CPU core even if you have setup the task use multiple cores for crunching.
ID: 47201 · Report as offensive     Reply Quote
Crystal Pellet
Volunteer moderator
Volunteer tester

Send message
Joined: 14 Jan 10
Posts: 1149
Credit: 7,035,906
RAC: 669
Message 47202 - Posted: 30 Aug 2022, 7:19:49 UTC - in response to Message 47198.  
Last modified: 30 Aug 2022, 7:22:52 UTC

I would like to ask your opinion about the behavior of the computation units I have been processing lately.
It seems that in the final stages of processing, when the estimated remaining time is a few tens of minutes, the execution speed decreases a lot. The example figure shows eight minutes remaining, but the actual processing will still take several hours.


Hello Marco,

For your type of CPU and running ATLAS on 4 threads that long runtime is not normal.
The 200 events to process should take between 8 and 12 hours at the most on that machine.
I noticed from the 2 results you returned that the Virtual Machine is paused and restarted very often.
ATLAS (and CMS) don't like that. They want to run in one flow without interruption. They also always want to have connection to the network.
1 of your results was succesfull (HITS file was successfully produced), the other was valid, but not successful (No HITS file was produced)
BOINC's 'tempo mancante' is useless for VM-tasks progress.
You may better use the VM's Console to watch the progress with the ALT-F3 and ALT-F2 keys.
ID: 47202 · Report as offensive     Reply Quote
Profile Marco Del Dottore

Send message
Joined: 21 May 10
Posts: 2
Credit: 117,361
RAC: 0
Message 47212 - Posted: 30 Aug 2022, 15:20:14 UTC - in response to Message 47202.  

Thank you Crystal, Harri and maeax for your useful advices.
I'm checking the configuration, hope to start with the tasks again soon.
Have a great day!
ID: 47212 · Report as offensive     Reply Quote

Message boards : ATLAS application : Very long lead times in the final stages of [Atlas Simulation] units


©2023 CERN