Job CPU time vs. job runtime

I know job runtime is wall clock time, but what is CPU time measuring? If a job were to use 100% of a single core, single CPU's processing resources for the entire runtime of the job would the CPU time always = runtime?

In multi-CPU, multi-core setups if all the CPU's/cores were utilized at 100% would the job CPU time exceed the job runtime? Would the difference between the runtime CPU time be an even multiple of the number of CPU's/cores in this case?

You talking about in a mainframe?

Regardless the answer is similar...it is how much time the cpu is actually used processing the job/task, or even the os. This does not count for idle states or waiting for I/O, etc... active cpu time.

Edit (more details):https://en.m.wikipedia.org/wiki/CPU_time

EarthDog said:

You talking about in a mainframe?

Regardless the answer is similar...it is how much time the cpu is actually used processing the job/task, or even the os. This does not count for idle states or waiting for I/O, etc... active cpu time.

Edit (more details):https://en.m.wikipedia.org/wiki/CPU_time

Click to expand...

Thanks ED, this pretty much confirms what I thought.

These stats aren't for a mainframe but for two HPC clusters, I had thought I'd have to explain the differences between the aggregate runtime stats and CPU time stats (because both stats are usually almost exactly the same, but always differ by a couple of hours).