Message: Re: AnalysisManager root performance for large numbers of ntuples Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

None Re: AnalysisManager root performance for large numbers of ntuples 

Forum: Analysis
Re: None AnalysisManager root performance for large numbers of ntuples (Sebastian Heil)
Date: 21 Jan, 2014
From: Guy Barrand <Guy Barrand>

   Hi

 In the case of an ntuple with a large number of columns, there is today a bottleneck in performance
due to the reallocation of internal buffers and "baskets" (ROOT terminology) sizes that appears to be
too short by default for this case.  I am going to propose to Ivana a patch for a further release in order
to improve the default behavior of these internal buffers. Waiting that you can test by doing the
following patch in your own g4tools (assuming that you have some control on your G4 installation).
 Then in : 
     g4tools/tools/wroot/buffer

have the public method :

     bool expand2(uint32 a_new_size) {return expand(mx<uint32>(2*m_size,a_new_size));}

and replace in this file "expand" with "expand2" in all the "write" methods as in the one :

    template <class T>
    bool write(T x){
      if(m_pos+sizeof(T)>m_max) {
        if(!expand2(m_size+sizeof(T))) return false;
      }
      return m_wb.write(x);
    }

 This should improve things (at least it clearly improve things on a standalone example).

   Cheers. Guy

Inline Depth:
 1 1
 All All
Outline Depth:
 1 1
 2 2
 All All
Add message: (add)

1 None: Re: AnalysisManager root performance for large numbers of ntuples   (Sebastian Heil - 23 Jan, 2014)
 Add Message Add Message
to: "Re: AnalysisManager root performance for large numbers of ntuples"

 Subscribe Subscribe

This site runs SLAC HyperNews version 1.11-slac-98, derived from the original HyperNews


[ Geant 4 Home | Geant 4 HyperNews | Search | Request New Forum | Feedback ]