Processing Speed
The time in seconds taken to process the first 10,000 events of run 1028 on a standard desktop machine is shown below for five of the calibration options. The calibrations which do not take account of trigger jitter all take a similar amount of time, the fast (interpolation based) trigger jitter correction is marginally slower, whilst the clock fitting method trundles in six times slower than the fast clock method.
| Calibration Name | CPU Time | Real Time |
|---|---|---|
| kVoltageTime | 14.06 | 46.2389 |
| kVTFullRG | 11.96 | 12.1132 |
| kVTFullJWPlus | 12.46 | 12.5758 |
| kVTFullJWPlusFastClockZero | 14.77 | 15.2354 |
| kVTFullJWPlusClockZero | 90 | 93.6464 |
Note: The long real time for the first method is due to NFS caching of the data.
Copyright © 2006 Ryan Nichol, (last modified 06 Jun 2007)