GSI Forum
GSI Helmholtzzentrum für Schwerionenforschung

Home » PANDA » PandaRoot » Bugs, Fixes, Releases » TPC digitization for more than 1000 events
TPC digitization for more than 1000 events [message #11926] Fri, 27 May 2011 14:05 Go to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *kph.uni-mainz.de
Hi all,

I want to get some advice from all PANDA user!
I'm trying to optimize events numbers for simulation with may11 stable version in my local farm.

I have electron particle simulated mainly in central part, TPC make huge amount of avalanche electron at some particular events in digitization process.

Of course it should take really time....but also not likely, usage of memory increase more and more if the events are going to more than ~1000 events. Then finally jobs crash...without output.

Memory RAM set 8 GB and swap allocates 10.0G in my local machine.
But roughly 5% of run, which has 1000 events, have such kind of problem in my machine.

In order to avoid this inefficient memory handling at TPC dig,
I reduce my event number as 200 in run, then looks fine for all.

Does anyone test already TPC Digi part with stable PANDAroot and with more than 10000 events for instance?


Thanks,
Donghee
Re: TPC digitization for more than 1000 events [message #11928 is a reply to message #11926] Fri, 27 May 2011 14:11 Go to previous messageGo to next message
Felix Boehmer is currently offline  Felix Boehmer
Messages: 149
Registered: May 2007
Location: Munich
first-grade participant

From: *natpool.mwn.de
Hi Dnonghee,

could you be more specific of what you are doing and what exaclty happens?
I just ran 300 jobs with 5000 pion events each on our cluster, memory allocation LIMITED to 1.0 GB per job, without problems...


Cheers

Felix
Re: TPC digitization for more than 1000 events [message #11929 is a reply to message #11928] Fri, 27 May 2011 14:34 Go to previous messageGo to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *kph.uni-mainz.de
Hi Felix,

I'm trying now with may11, that is before last stable version at two days ago. It is not really last version at today.
Just one electron in central part ( 100 - 160 degree), gamma and proton only into very forward direction ( 10 degree).
Please don't ask me why there are proton... that is my hobby event. And interaction point is (0,0,30).

So then I have seen some avalanche creation and digitiztion in print out during run of digi macro. It looks like typically this...

Quote:


-------------------------------------------------
2020 Avalanches created
0 aggregations done.
3577 Signals created
PndTpcElectronicsTask::Exec
Building up padmap ...finished. 116 pads hit
...........
119 Digis created




If TPC procuded really big number of "avalanches created", sometimes more than 200000 for example, here I need quite a time to go further digitization or next event, then increasing memory more and more.

I submit jobs ~100 runs with each 1000 events, got 5% no output due to this problem. Simply machines are down!

If the situation will be improved with really last stable may11 version, then I'm happy, but I think there are not so much diffrences between last and before last.

Best wishes,
Donghee



Re: TPC digitization for more than 1000 events [message #11930 is a reply to message #11929] Fri, 27 May 2011 14:40 Go to previous messageGo to next message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *kph.uni-mainz.de
Hi Felix,

Honestly I didn't mention that I don't know exactly whether the printing out relates to only TPC part or some other detectors Smile

Quote:


-------------------------------------------------
2020 Avalanches created
0 aggregations done.
3577 Signals created
PndTpcElectronicsTask::Exec
Building up padmap ...finished. 116 pads hit
...........
119 Digis created



If 1-3 lines are related to TPC, then my complain is still valid...

Donghee
Re: TPC digitization for more than 1000 events [message #11931 is a reply to message #11929] Fri, 27 May 2011 14:40 Go to previous messageGo to next message
StefanoSpataro is currently offline  StefanoSpataro
Messages: 2736
Registered: June 2005
Location: Torino
first-grade participant

From: *to.infn.it
This problem in tpc digitization is known since several months. It depends mainly on the machine you are using. With my SL4s I could not run more than 2000 events, with my new machine I can run 10k events without problems.
Re: TPC digitization for more than 1000 events [message #11932 is a reply to message #11931] Fri, 27 May 2011 14:45 Go to previous messageGo to next message
Felix Boehmer is currently offline  Felix Boehmer
Messages: 149
Registered: May 2007
Location: Munich
first-grade participant

From: *natpool.mwn.de
Hmm,

what momenta do your tracks have? Can you have a look at the clusters or simply the MC points - do you see anything weird?

If not then I don't know what's the problem.


Re: TPC digitization for more than 1000 events [message #11933 is a reply to message #11932] Fri, 27 May 2011 15:00 Go to previous message
donghee is currently offline  donghee
Messages: 385
Registered: January 2009
Location: Germnay
first-grade participant
From: *kph.uni-mainz.de
Hi all,

The electron has mostly momentum below 3 GeV. only 2-3% exceed more than 3 GeV.
I'm slightly confusing because of argue of felix.
You produced pion without any problem at (0,0,0).
I have also seen that TPC is very stable with 6 outgoing pion or kaon tracks of psi(3770) decay at interaction IR(0,0,0).

I will inform you further studies after updating of may11.
And I think this is good reference view to know, if we change IR(0,0,30) region, how will TPC perform.

Anyhow, thank you for your listening.

Best regards.
Donghee
Previous Topic: Meaning of Core dump due to Geane application at rec. part
Next Topic: Reconstruction macro crush - PndMCTrackAssociator
Goto Forum:
  


Current Time: Thu Mar 28 22:38:55 CET 2024

Total time taken to generate the page: 0.00993 seconds