Problem 1308 - Geant 4.9.5.p01 slower to compile application and to execute than Geant 4.9.4.p02
Summary: Geant 4.9.5.p01 slower to compile application and to execute than Geant 4.9.4...
Status: CLOSED WORKSFORME
Alias: None
Product: Geant4
Classification: Unclassified
Component: processes (show other problems)
Version: 9.5
Hardware: PC Linux
: P5 normal
Assignee: Ben Morgan
URL:
Depends on:
Blocks:
 
Reported: 2012-04-26 08:04 CEST by jerome.roccaz.cn
Modified: 2018-05-08 14:47 CEST (History)
0 users

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this problem.
Description jerome.roccaz.cn 2012-04-26 08:04:32 CEST 
Hi everyone,

I want to report a possible bug I've already reported on Geant4 forum (http://hypernews.slac.stanford.edu/HyperNews/geant4/get/installconfig/1457/1.html ) and maybe related to problem 1286 (http://bugzilla-geant4.kek.jp/show_bug.cgi?id=1286).

I'm working under SL 6.0 running on VMWare 4.0.1 build-528992, 2Gb memory, Core i5. I've recently installed last Geant4 release.  

Considering a program I ran under Geant4.9.4.p02, I've compiled it with Geant4.9.5.p01 after some small modifications so to be compilable. While with previous realease it took few seconds for linking process, whith last realease it takes minutes!

When I use the system monitor, occupied memory before compiling is 0.24Gb. When ld process starts, occupied memory by this process does not stop increasing till the end of linking, up to 420Mb and total occupied memory increased up to 1.4Gb. %CPU occupied by ld process is around 10% with raise up to 20%. At the same time Xorg process is working at total occupied %CPU can increase up to 100%.
After end of linking process, occupied memory decrease down to 1Gb, and all system is slowed.

When I run my programm, time to execute is nearly 3 times longer than the same program compiled with Geant 4.9.4.p01.

My program simulates a neutron source, so I use NeutronHP libs and hadronic libs. 

If you need more information, please let me know.
Comment 1 Gabriele Cosmo 2012-04-26 11:58:49 CEST 
We do not have directly access to VMWare systems, but cannot reproduce the issue reported here.
You should verify if compilation options used are the same, i.e. it looks like you're compiling 9.5.p01 in debug mode, in which case a factor 3 at run-time and bigger size of the binaries would explain what you observe...
Comment 2 Gabriele Cosmo 2012-04-26 11:59:48 CEST 
We do not have directly access to VMWare systems, but cannot reproduce the issue reported here.
You should verify if compilation options used are the same, i.e. it looks like you're compiling 9.5.p01 in debug mode, in which case a factor 3 at run-time and bigger size of the binaries would explain what you observe...