home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!ira.uka.de!chx400!urz.unibas.ch!bioftp.unibas.ch!comp.bioz.unibas.ch!doelz
- Newsgroups: comp.sys.sgi.misc
- Subject: Re: REALLY nice processes?
- Message-ID: <1993Jan4.093800.17457@comp.bioz.unibas.ch>
- From: doelz@comp.bioz.unibas.ch (Reinhard Doelz)
- Date: Mon, 4 Jan 1993 09:38:00 GMT
- Reply-To: doelz@urz.unibas.ch
- Sender: usenet@comp.bioz.unibas.ch (NEWS transaction account)
- References: <NS14.93Jan4013229@crux2.crux2.cit.cornell.edu>
- Distribution: comp
- Organization: EMBnet Switzerland [BASEL]
- Nntp-Posting-Host: biox.embnet.unibas.ch
- Lines: 26
-
- In article <NS14.93Jan4013229@crux2.crux2.cit.cornell.edu>, ns14@crux2.crux2.cit.cornell.edu (Nathan Otto Siemers) writes:
- ...
- |> Even when I nice my processes as much as possible, there is a
- |> noticable drop in performance in interactive work. Is there any way
- |> to eliminate this? I could envision the swapper swapping out
- |> processes when an interactive job is running; is this configurable?
-
- First, you could purchase a batch queuing system (like NQS) which does
- automatic non-aging priorities. Second, you could do it yourself with
- the npri(1) call. See the man page for this. Last, you could make a daemon
- which greps the CPU time of, lets say, more than 1 hour and sends a STOP signal
- to there if you log in to the console.
-
- Regards
- Reinhard
-
-
- --
- +----------------------------------+-------------------------------------+
- | Dr. Reinhard Doelz | RFC doelz@urz.unibas.ch |
- | Biocomputing | DECNET 20579::48130::doelz |
- |Biozentrum der Universitaet | X25 022846211142036::doelz |
- | Klingelbergstrasse 70 | FAX x41 61 261- 6760 or 267- 2078
- | CH 4056 Basel | TEL x41 61 267- 2076 or 2247 |
- +------------- bioftp.unibas.ch is the SWISS EMBnet node ----------------+
- -----------------------------------------
-