Closed dneise closed 8 years ago
I tested it on ~250MB input of raw data, i. e. 200k events of roi=40. It needs ~6.5GB when reading the data into memory, later during fitting ~2GB are enough. The run time is 5minutes. compared to 11minutes, when using the other two programs. (using 7cores for fitting the power laws in both cases)
May I merge this?
No :D
I combined the two scripts, which are currently needed to create a DRS4 calibration file, which can be used by the
TimelapseCalibration
. working title is "calc_timelapse_constants"So instead of calling:
dragonboard_dataextraction /path_to_rawdata/some_files_*.dat --outpath data.h5 --memory 60
and then:dragonboard_calc_calib_constants data.h5 calib.h5 -n 7
one can call:
calc_timelapse_constants calib.h5 /path_to_rawdata/some_files_*.dat
By default it uses all CPUs minus one (or one if only one is there). It uses all the memory it can get and this might be a lot... It is a little bit faster than the other two. It does not even try to extract calibration constants for the 8th channel.