I followed Simona's recipe, running everything except the final histogram comparison on the native architecture, for the files br042ecb.0453phys and br042ecb.03b7phys. This time I see differences, and odd differences they are, too.
First look at the "03b7" file. If I ask that the histograms be bin-wise identical, between SL3 versions and SL5 versions there are 353 differences. Some of them look like this one (which actually comes from a comparison of the "0453" files):
The naked eye suggests that there are a number of notable differences here.
If I require only that the histograms vary at more than the .99999999 (8 9's) level in the KM test, only 1 histogram differs: Lummon/event_1/ntracks.
The following is SL3
The following is SL5
It looks like a simple bin shift.
Other examples of differences in the "0453" files follow. These are from a "3 vs 4 comparison."
| comparison | # NOT identical | KM cut .99999999 |
| SL3 vs SL4 | 350 | 0 |
| SL3 vs SL5 | 350 | 0 |
| SL4 vs SL5 | 117 | 0 |
| comparison | # NOT identical | KM cut .99999999 |
| SL3 vs SL4 | 352 | LumMon/event_1/ntracks (see above) |
| SL3 vs SL5 | 353 | LumMon/event_1/ntracks |
| SL4 vs SL5 | 65 | 0 |
Modified 27-March-2010 at 11:46
http://hep.physics.wisc.edu/~jnb/cdfcode/27Mar2010
| Previous notes | Next notes | Main slide directory |