Date: Sun, 12 Oct 1997 16:02:20 +0100 (MET) From: Luigi Rizzo <luigi@labinfo.iet.unipi.it> To: itojun@itojun.org Cc: xaa@stack.nl, joerg_wunsch@uriah.heep.sax.de, freebsd-hackers@FreeBSD.ORG, denny1@home.com Subject: Re: wish /bin/sleep handled fractions of a second. Message-ID: <199710121502.QAA03647@labinfo.iet.unipi.it> In-Reply-To: <10514.876671975@coconut.itojun.org> from "itojun@itojun.org" at Oct 13, 97 00:59:16 am
next in thread | previous in thread | raw e-mail | index | archive | help
> Is it so meaningful to provide very high resolution timing with
> sleep(1)? There will be error in timing due to disk access
> for sleep(1) binary, process fork/exec time, and so forth.
> IMHO, 1/10sec resolution may worth it, but beyond that becoming more
> and more meaningless.
Good point. So I made some quick tests now on a P5133 with ide disk:
prova# time ps # first execution
... ps output
0.007u 0.023s 0:00.14 14.2% 358+1000k 2+0io 8pf+0w
prova# time ps # second execution, so "ps" is in core
... ps output
0.014u 0.014s 0:00.03 66.6% 152+492k 0+0io 0pf+0w
and...
prova# time sleep 0
0.000u 0.004s 0:00.00 0.0% 0+0k 0+0io 0pf+0w
Considering that those who use sleep(1) with sub-second resolution
presumably will run it from a warm file cache, then even nowadays it
might make sense to have 1ms of resolution. After all, at every scale
you look at, there is no guarantee that a process will complete within
a deadline.
Cheers
Luigi
-----------------------------+--------------------------------------
Luigi Rizzo | Dip. di Ingegneria dell'Informazione
email: luigi@iet.unipi.it | Universita' di Pisa
tel: +39-50-568533 | via Diotisalvi 2, 56126 PISA (Italy)
fax: +39-50-568522 | http://www.iet.unipi.it/~luigi/
_____________________________|______________________________________
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199710121502.QAA03647>