[Gate-users] GATE 5 performance vs. GATE 4

[Bryan McIntosh]

I've successfully installed GATE 5 alongside GATE 4, and noticed that
it is significantly slower.  The bnechmarkPET macro takes 70% longer
to run under GATE 5 than under GATE 4, which was my first clue that
there was a difference.  I also found that a simulation I've made of
two Siemens Inveon detector modules with no external source aside from
intrinsic LSO activity runs 16% slower in GATE 5 than in GATE 4.

The support files I am using are:

-CLHEP 2.0.3.2
-GEANT4 9.1 patch 03
-ROOT 5.20
-LMF 3.0
-ECAT 7

I've tried changing the random number generator between the JamesHep
and Twister algorithms, as well as using GEANT4 9.2 p01, but neither
of those steps has changed the computing time noticeably.  Has anyone
else run into this yet, and is it expected to happen?  Any ideas as to
why the new code is so much slower than before?

Thanks a lot,

-Bryan