As was pointed out, unless the "soaker job" is running only a single task, the expectation is flawed. TIMEUSED provides the time used by the issuing task, not the time for the job step. Only when the job step consists of a single task might you expect TIMEUSED to be the time for the job step.
A time value of 00000000B57E2F00, as was pointed out, that is x'B57E2' microseconds.. Dividing the number of microseconds by 10000 gives the number of hundredths of seconds. As was pointed out, this is 74 (i.e., about 3/4 of a second), and this is what the display showed. >I would expect it to be somewhere around 17. I don't know why you would expect that. The CPU time for the job was reported as 17.17 seconds. Therefore you might expect 1700 but not 17. If you wanted something other than the number of hundredths of seconds, then 10000 was the wrong dividend to use. I don't know specifically what the OP is trying to track, but ASCBEJST or ASSB_TIME_ON_CP (the latter is not always set, depending on machine, release, and zIIP configuration) might provide the value of interest. Those will not include the time that each task has accumulated since it was dispatched the last time but will include the time from all previous dispatches. Peter Relson z/OS Core Technology Design
