> Thanks for the feedback, I thought I had seriously screwed up a build. I see 
> the correct output for the looping process above, but most other processes
> except compute bound ones like dnetc report 0:00.00. Yet a copy of ps from 
> last year seems to report times which tally with top. I have put the latter
> binary on an ftp server (ftp://ftp.liv.ac.uk/pub/dtyson/ps), it would be 
> useful if you could see if it reports more sensible times on your system.
> I'll have a look at differences between the versions and see if I can spot 
> where the error lies and put in a PR.

I didn't try your binary but I think you should send a PR.
Before send it you can ask about ps in tech-userlevel@.

-- 
Best regards, Aleksey Cheusov.