I have gathered some performance data for the calculation of the intersecting volume between an elliptical cross-section cylinder (tgc) and a hexahedron (arb8) using gqa in MGED. I created a 100x scaled version of the same model and again used gqa to calculate the volume. I used the following command to get the volume and time (time "gqa ..."):
gqa -Av -g XX int.c
where XX is the grid spacing I used except for the default case.
I see that the default 0.25 lower limit for the grid size is not optimal for either the original or the 100x scaled version of the model. Also, I can get to within 0.3% error in 0.05s for the small model but for the 100x model requires more than 0.15s to get to the same error level. Here is the data in a table (hoping the formatting will stay as it is)
Characteristic Size 1 100
V Actual 0.7853982 785398.2
Grid Size (% of charc. Size) Volume Time (microsec) % error Volume Time (microsec) % error
Def. Grid Size (0.25 abs lower limit) 0.731724 23535 -6.8% 787744 368938 0.3%
10.0% 0.76912 38972 -2.1% 834063 25095 6.2%
5.0% 0.78306 51085 -0.3% 811749 41145 3.4%
2.5% 0.78539 85981 0.0% 799036 67201 1.7%
1.0% 0.785166 187265 0.0% 790717 162074 0.7%
0.5% 0.785536 404860 0.0% 788228 298249 0.4%

I also need to calculate the center of gravity for the combination but I can't seem to find out how to assign the density and then get the center of gravity/moments of inertia information.
Any help with the volume calculation accuracy and speed as well as the density assignment and center of gravity calculations is much appreciated.