101% system resource usage

101% system resource usage

[Nikolai Weibull]

It's true!  My system can execute at 101% efficiency!  Sounds like
something taken from michaelscomputers.com [1] perhaps, but here's what
a timing of grep gave:

::: dir: ~ :: user: pcp@puritan :::
time grep -c 'a*' test.file
1
grep -c 'a*' test.file: real: 1.212 user: 1.140 system: 0.090 (101%)
::: dir: ~ :: user: pcp@puritan :::


I don't know if anyone really cares, but there seems to be a seldom
occuring rounding error perhaps?

::: dir: ~ :: user: pcp@puritan :::
zsh --version
zsh 4.2.0-pre-1 (i686-pc-linux-gnu)

If that's of any help.  (Sorry, but don't know if there's any better way
to generate debugging information.)
	nikolai

[1] http://www20.tomshardware.com/column/20040317/index.html

--
::: name: Nikolai Weibull    :: aliases: pcp / lone-star / aka :::
::: born: Chicago, IL USA    :: loc atm: Gothenburg, Sweden    :::
::: page: www.pcppopper.org  :: fun atm: gf,lps,ruby,lisp,war3 :::
main(){printf(&linux["\021%six\012\0"],(linux)["have"]+"fun"-97);}

Re: 101% system resource usage

[Peter Stephenson]

Nikolai Weibull wrote:
> grep -c 'a*' test.file: real: 1.212 user: 1.140 system: 0.090 (101%)

In fact the calculation (printtime() in jobs.c) always rounds down
(though I'm not sure why it's done the way it is; it effectively
recalculates three floating point numbers it's already calculated).

It may be an artifact of the way the start of real time is only recorded
by the shell after the process is forked (and the end of real time after
it finishes), while the time of process execution is recorded by the
operating system itself and so is more accurate.  If there's a
significant gap due to scheduling the `real' time may be too short.

-- 
Peter Stephenson <pws@pwstephenson.fsnet.co.uk>
Work: pws@csr.com
Web: http://www.pwstephenson.fsnet.co.uk