Here are flow diagrams showing for each bombarding energy the cpu
time for hitfinding and tracking, as well as some relevant information,
like the average number of hits, tracks, track length, etc.
Notes:
- For 2d hitfinding, this is for the first pass only.
- No corrections (t0, distortion...) were used.
-
On the charts are the results for BTF with "tight" and "loose" switch
settings. Goto the bottom of this page to
see the switches.
Look at the chart for 2 GeV data.
Look at the chart for 4 GeV data.
Look at the chart for 6 GeV data.
Look at the chart for 8 GeV data.
Notice that:
- For the low-energy events, 2d hitfinding takes less time
than 1d, but for higher energy, it takes more time. This may
be why experiment E910 finds that 2d is faster-- because they
have proton beams, which is more like our lower energies.
- The cpu times are really not that much different between the
codes (although the 2d hitfinder will require another pass!).
- The hitfinder and tracker take comparable amount of time.
- It seems that BTF does better (% hits uses, % track seen)
with loose cuts than with tight ones, although it takes somewhat more
cpu.
btf_sw table
| Switch | Tight setting (default) | Loose setting
|
|---|
| itolst | 2 | 2
|
| itolsg | 3 | 3
|
| nmismx | 3 | 3
|
| npacpt | 5 | 3
|
| npconf | 5 | 5
|
| toladp | 0.3 | 0.5
|
| resfac | 2.0 | 2.0
|
| sangmx | 0.998 | 0.999
|
| chipmx | 10.0 | 10.0
|
| tolsta | 3.0 | 5.0
|
| tolseg | 3.0 | 3.0
|
Return to the Hitfinder home page..