[Gate-users] root file size issue

Zhang,Yuxuan yuxuan.zhang at mdanderson.org
Wed Jun 6 23:16:01 CEST 2012


Dear All,

In my recent simulations I noticed that the root files become 
much larger than it used to be. After some investigation, these are
what I have found:

1. The change of the root file size seems happened after I upgraded
the GATE from 5.0 to 6.1, and the Root is upgraded from 5.21 to 
5.28 at the same time.

2. For a similar simulation (geometry, activity, time), the root file 
from the new version can be 10 to 18 times larger than that from 
the old version. 

3. The root file from the old version cannot be compressed by the 
software such as RAR (only a few percent), while the new one can be 
compressed with the ratio of 5 times.

4. When I open the new root file with a Hex editor, I can see large 
continuous areas of memory with keyword of TBastek and "moduleID1 Coincidence"
, "layerID2 Coincidences" etc. It seems just empty records. If I open
the file with Root and use f.Map() command, here is part of the results:

20120605/163443  At:299969  N=89        TBasket
20120605/163443  At:300058  N=91        TBasket
20120605/163443  At:300149  N=88        TBasket
20120605/163443  At:300237  N=455       TBasket        CX =  3.15

There are a lot of those TBasket records with the sizes from  88 to 98
without the CX values. The last line tthat has a "CX=3.15" seems to be
a correct record.

Is there anybody else noticed this phenomenon in your simulation? Though this thing
dose not affect the output and the results looks fine, however, it waste
too much disk space so it is better to get it solved. Anyone have some
ideas how this happened?  Thank you.


Yuxuan


More information about the Gate-users mailing list