[Nottingham] Time

[Martin]

Joshua Lock wrote:
[---]
> An initial guess is that system measures things like disk i/o. Which
> would explain why after the first of 5 consecutive runs it's consistent.

Well, that follows my understanding...

"user" is the userspace cpu time used, "system" is the kernel cpu time
used servicing your job's requests.

What are you benchmarking?

Algorithms costing or compiler optimisations or?

Cheers,
Martin

-- 
----------------
Martin Lomas
martin at ml1.co.uk
----------------