home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!usc!zaphod.mps.ohio-state.edu!uwm.edu!ogicse!das-news.harvard.edu!cantaloupe.srv.cs.cmu.edu!lindsay
- From: lindsay+@cs.cmu.edu (Donald Lindsay)
- Newsgroups: comp.arch
- Subject: Re: Interrupts vs Polling---WAS: cooling
- Message-ID: <BtMC8n.B6H.2@cs.cmu.edu>
- Date: 27 Aug 92 01:31:34 GMT
- Article-I.D.: cs.BtMC8n.B6H.2
- References: <BtED7C.5II@exnet.co.uk> <BtEy50.Ix4.2@cs.cmu.edu> <BtIDsH.8x3@exnet.co.uk>
- Sender: news@cs.cmu.edu (Usenet News System)
- Organization: School of Computer Science, Carnegie Mellon
- Lines: 28
- Nntp-Posting-Host: gandalf.cs.cmu.edu
-
- In article <BtIDsH.8x3@exnet.co.uk> dhd@exnet.co.uk (Damon) writes:
- >>High-reliability systems (and high priced
- >>systems) can justify computer monitoring to detect e.g. a failed fan.
- >>Aha, /dev/muffin! - well, actually, it's more likely an interrupt
- >>flavor.
- >Let's not use interrupts for this sort of thing. Just how quickly is the
- >temperature going to change. A once per second poll should be more than
- >enough. B^>/2.
-
- Traditionally, power fail was reported via interrupt, since many
- power supplies could only hold the system up for milliseconds. Big
- systems (which have had computerized monitoring for decades)
- typically were controlled from maintenance processors, which is why
- you don't see maintenance features in the mainframe's architecture
- spec. (Actually, at one time the biggest "feature" was a microfiche
- reader for the schematics.)
-
- It *is* sometimes important to get fast notice of cooling failure.
- Some systems, stopped abruptly, get hot enough to damage themselves.
- Their power-fail sequence tries to have residual power, not for the
- CPU or disks, but for the fans/pumps.
-
- >PS. If I replace /dev/heatsink with a plain file can I cool the system by
- > dumping the file across the net? B^>
-
- No, but pizzas that you FTP in will arrive hot.
- --
- Don D.C.Lindsay Carnegie Mellon Computer Science
-