Chandra-MARX / marx

Chandra X-ray Observatory ray-trace simulator
http://space.mit.edu/cxc/marx/
5 stars 4 forks source link

Investigate performance if CALDB files are used directly #5

Open hamogu opened 9 years ago

hamogu commented 9 years ago

JD wanted MARX to be fully self-contained, but as the required calibration files grow, that might not be practical any longer. Files in /data/caldb are taken from CALDB directly or with minimal processing (re-order, bin-up). If at least those files could be read from CALDB (which will be installed on essentially every MARX user's system) directly, I would not have to maintain them. On the other hand, the performance of MARX will go down and any change in the CALDB format would require an immediate MARX release. Also, accessing CALDB through the normal access routines is potentially a rabbit-hole of complex logic and more dependencies.

Using CALDB directly would mitigate #4 automatically.

To address this issue, I should first do some experiments and measure the performance hit from using the big contam file in CALDB. That much can be tested with no code change.

hamogu commented 8 years ago

Speed tests on melkor indicate that the performance impact is < 10%. Need to test machines with less RAM, though.