[gate-users] Root Output - Dealing with the file size limit

[Irène Buvat]

Dear Irene,

>
>First of all, I experience problems sending e-mails from my e-mail address,
>every time I try to send one, the gate-users-owner replies me saying that I
>am not allowed. When I send an e-mail to the gate-users-owner, then the
>answer is the address is unknown. That is why I use the address of my
>colleague V. Spanoudaki. What can I do??

You've first to post to the new mailing list, 
gate-users at lists.healthgrid.org since the old one 
(gate-users at lphe1pet1.epfl.ch)  will soon not be supported anylonger.
If you were a member of gate-users, we have been automatically set as 
a member of the new list, but your passwd has been regenerated for 
security reasons. You can change your setting by first asking for 
your new passwd.

Hope this helps, and if you encouter any other problem, please feel 
free to contact me.

Irene


>  >>Hi, I'm trying to generate a root output file.  I seem
>>>to be exceed the 2GB file size limit very easily and
>>>I'm not sure how to get around this.
>
>Regarding the size of the root file I had the same problem last year, I
>send you what I did in order to work with larger file sizes than 2G. The
>following was suggested by Nerina Gimenez (IFIC-Valencia), she found it in
>the GATE talk and added some modifications in order to use it.
>
>This is the web page where she found it:
>
>http://www-lphe.epfl.ch/~PET/GATEusers/documentation/GATE-USERS_TALK/html/2004-December/000395.html
>
>********************************************************************************
>
>Yeap, a 3.7 Giga bytes ROOT file. This file contains singles and
>coincidences.
>I then processed the file with a .C program to create my listmode data
>and everything went smooth. The listmode file was smaller than 2Giga
>though.
>But In theory the latest version of ROOT can read and write files
>larger
>than 2 Gigabytes
>
>What to do:
>
>1) update ROOT to version v4.01.02.
>2) edit the petsim/include/GateToRoot.hh file as follows:
>
>     In the class SingleOutputChannel : public VOutputChannel,
>     in the inline method Book, add the SetMaxTreeSize call to set the
>     maximum size of the Singles tree.
>
>       inline void Book()
>         {
>            m_collectionID=-1;
>            if (m_outputFlag) {
>               m_tree  = new GateSingleTree(m_collectionName);
>               m_tree->SetMaxTreeSize(17179869184L);
>               m_tree->Init(m_buffer);
>            }
>         }
>
>     Do the same for
>     class CoincidenceOutputChannel : public VOutputChannel
>
>       inline void Book()
>         {
>           m_collectionID=-1;
>           if (m_outputFlag) {
>             m_tree  = new GateCoincTree(m_collectionName);
>             m_tree->SetMaxTreeSize(17179869184L);
>             m_tree->Init(m_buffer);
>          }
>      }
>
>
>3) Recompile Gate
>
>The method TTree::SetMaxTreeSize(Long64_t maxsize) takes a 64 bits
>Long,
>so in theory you could
>go all the way to 2^64=18446744073709550000 bytes, if unsigned (I'm
>not
>sure, maybe is only to 2^63).
>Right now, the value I 'm using is 16Gb.
>
>
>*****************************************************************************************
>****************************************************************************************
>The only change that I did following Nerina suggestions was to modify
>17179869184L for 17179869184LL in both cases. It worked fine for me.
>
>I hope this can help you.
>
>>  A way that you can get around the 2GB limit for now is to limit the
>>  acquisition time for your simulations, and run multiple simulations,
>>  making sure the seed is changing each time you run a new sim.  I have a
>>  little program that generates the JamesRandom seed that GATE uses to
>>  run, if you would like.
>
>By the way, I would be interested in using this program (if there is no
>problem), we are starting simulations in a cluster and we are
>investigating how to handle the problem with the random seeds.
>
>Thanks a lot,
>
>Irene
>
>
>
>--
>Irene Torres-Espallardo
>Klinikum rechts der Isar
>Nuklearmedizinische Klinik und Poliklinik
>Technische Universität München
>email: i.torres at lrz.tu-muenchen.de
>Tel  : 0049-89-4140-4569
>Fax  : 0049-89-4140-4938
>
>_______________________________________________
>gate-users mailing list
>gate-users at lphe1pet1.epfl.ch
>http://lphe1pet1.epfl.ch/mailman/listinfo/gate-users


-- 
-------------------------------
Irene Buvat, PhD
Equipe "Imagerie in vivo de la perfusion et des échanges moléculaires"
http://www.guillemet.org/irene/equipe4
U678 INSERM
CHU Pitie-Salpetriere
91 Boulevard de l'Hopital
75 634 Paris Cedex 13
France

Tel : (33) (0)1 53 82 84 19
Fax : (33) (0)1 53 82 84 48
e-mail : buvat at imed.jussieu.fr
http://www.guillemet.org/irene