Let me try to explain a problem. I am testing exactly the same workbench on the to computers. One has installed FME 2012 without SP, another has the same but with SP4. Tested workbenches behaviour are different. First, one without SP4 process data min. 10x faster. Other thing is difference in information about processing in log window.
For example, for FME 2012 without SP:
Reading table/feature class 'name1' Reading table/feature class 'name2' Processed CANDIDATE 7070 Reading table/feature class 'name33' Processed CANDIDATE 19460 Processed CANDIDATE 27470
And for FME 2012 SP4 at the same part of log:
Sorting 566 features... Calculating inside points... Determining uppermost containment... Constructing donuts from 566 features...
And that takes long time. Input data are in File geodatabase API format. Any clues what should I do? Why constructing donuts and holes? I was quite satisfied with the way and processing time before. Thank you.