Message: Execution Speed and Parameterization Not Logged In (login)
 Next-in-Thread Next-in-Thread
 Next-in-Forum Next-in-Forum

Question Execution Speed and Parameterization 

Forum: Medical Applications
Date: 14 Sep, 2004
From: Ken Sutherland <ken@cme.hokudai.ac.jp>

Dear Geant4 experts,

I am experimenting with the DICOM example. I am using Geant4.6.1. I am shooting protons into the CT volume.

The execution speed is around 2 or 3 minutes per event on my 3GHz P4 machine. I can improve the time by increasing the compression (subsample) value, but that makes the result rough.

I am measuring time between RunAction::BeginOfRunAction and RunAction::EndOfRunAction, so as to ignore Geant4 and volume initialization time.

If I set the compression value to 2, the speed improves to about 20 to 30 seconds per event. I wonder why the number of voxels has such a drastic effect on the execution time. It seems that there is a heavy penalty when transfering from one parameterized volume (voxel) to the next.

I have gone over the source code, making trivial improvements here and there, but is there anything I can do, other than parallel processing, to significantly improve the execution speed?

Thank you!

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

1 None: Re: Execution Speed and Parameterization   (Hongyu Jiang - 17 Sep, 2004)
1 None: Re: Execution Speed and Parameterization   (Ken Sutherland - 21 Sep, 2004)
3 Sad: Re: Execution Speed and Parameterization   (Takashi.Sasaki@kek.jp - 08 Dec, 2004)
1 Feedback: Re: Execution Speed and Parameterization   (Ken Sutherland - 10 Dec, 2004)
...
3 Idea: Re: Execution Speed and Parameterization   (Makoto Asai - 13 Oct, 2004)
 Add Message Add Message
to: "Execution Speed and Parameterization"

 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 ]