home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!ames!olivea!sgigate!odin!sgi!rhyolite!vjs
- From: vjs@rhyolite.wpd.sgi.com (Vernon Schryver)
- Newsgroups: comp.sys.sgi
- Subject: Re: how to interrogate ftimer status from a program?
- Message-ID: <s1po0j4@rhyolite.wpd.sgi.com>
- Date: 7 Nov 92 00:02:08 GMT
- References: <23907@hacgate.SCG.HAC.COM> <1992Nov6.213506.12264@odin.corp.sgi.com>
- Organization: Silicon Graphics, Inc. Mountain View, CA
- Lines: 22
-
- In article <1992Nov6.213506.12264@odin.corp.sgi.com>, probins@bubba.wpd.sgi.com (Paul Robins) writes:
- > ...
- > Fast clock does not incur nearly the overhead of timers, but I don't have a
- > relative number. I've been assured that it is of neglible impact.
-
-
- A few years ago the number was either 5% or 15% of a CPU (I've
- forgotten which) to run the fast clock. That is why it was not turned
- on by default. More precisely, that is why ASD changed it from
- being on by default to being off by default.
-
- Maybe that cost has changed. Maybe some people now consider 5% (or 15?)
- "neglible." Some did back then, until the issue was raised.
-
- You can measure it for yourself. Pick any CPU stupid benchmark
- (e.g. drhystone). Run it with `ftimer -f off`, and then run it with
- `ftimer -f on`. Do this several times to get consistent numbers. As
- long as you change nothing else, the difference is the cost of fast
- timers.
-
-
- Vernon Schryver, vjs@sgi.com
-