> Hello Robert, > I did reproduce the time you reported and it seems to me my first guess ("garbage collection") > was a wrong one. The collection is not activated for TNode's. > I have no explanation yet. > I hacked in enough #includes, forward declaration and fixed up an what turns out to be an attempt to modify a constant string and got everything to compile using .L time_geom.C++. This reduces the overhead where I'd expect it, but did nothing for the destruction time (as one would expect since the interpreted form simply has to find/call the TGeometry destructor). -robert ================================== build geometry nplanes=48 nstrips=192 cummulative time basics Real time 0:0:0, CP time 0.000 matrices Real time 0:0:0, CP time 0.000 shapes Real time 0:0:0, CP time 0.250 nodes Real time 0:0:0, CP time 0.600 done building shape list has 9266 entries world node has 51 sub-nodes destroy geometry Real time 0:0:15, CP time 14.910 done destruction root [2] time_geom(484,0); build geometry nplanes=484 nstrips=192 cummulative time basics Real time 0:0:0, CP time 0.010 matrices Real time 0:0:0, CP time 0.010 shapes Real time 0:0:3, CP time 3.110 nodes Real time 0:0:6, CP time 6.440
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:22 MET