After you have entered the appropriate parameters for any of the Torch programs / transactions, the job will be executed immediately when run via “Execute” or started through a background job with a variant. The execution sequence is as follows:
1) Creation of “intervals” that represent the work to be processed (either from a pre-generated interval variant or dynamically created at run-time)
2) Processing of each interval to determine the business logic required dependent upon the program being executed
3) Collation of results for aggregation of what has been done across the parallel processes
4) Display of aggregated result as an output (ad-hoc)
Step 2 is done with parallel processing. Steps 1 and 3 in the above sequence are done in a sequential manner with a single work process. The final step 4 is optionally performed if a Torch program has a result to be output for analysis by the user.