osmigo@ut-emx.UUCP (05/31/89)
It seems to me that if you used Virtual Memory as your "standard, everyday setup," it would involve INSTENSE, CONTINUOUS hard disk accessing, in contrast to the occasional "chirp" of the drive we now experience. Given that the more frequently something operates, the sooner it breaks,I'm wondering if the upcoming virtual memory enhancement would significantly reduce hard disk longevity. I'm also wondering how such longevity is related to the amount of accessing. In other words, if accessing increases 100%, does the wear rate increase 100% as well? If so, a 100% increase in accessing would reduce the life of the hard disk by HALF, and I'm certain that many virtual memory setups will increase accessing far in excess of 100%. Ron =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+ > Ron Morgan {ames, utah-cs, uunet, gatech}!cs.utexas.edu!ut-emx!osmigo < > Univ. of Texas {harvard, pyramid, sequent}!cs.utexas.edu!ut-emx!osmigo < > Austin, Texas osmigo@ut-emx.UUCP osmigo@emx.utexas.edu < =+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+=+
cgw@cadre.dsl.PITTSBURGH.EDU (Gray Watson) (06/02/89)
In article <13548@ut-emx.UUCP> osmigo@ut-emx.UUCP writes: >It seems to me that if you used Virtual Memory as your "standard, everyday >setup," it would involve INSTENSE, CONTINUOUS hard disk accessing, in contrast >to the occasional "chirp" of the drive we now experience. > >Given that the more frequently something operates, the sooner it breaks,I'm >wondering if the upcoming virtual memory enhancement would significantly >reduce hard disk longevity. >Ron Morgan {ames, utah-cs, uunet, gatech}!cs.utexas.edu!ut-emx!osmigo I guess the question here is: what part of the hard-drive usually fails first due to longevity or use over a long period of time?? As far as I know, most of the hard disk failures, I've heard of, have been bering and drive-motor problems. If this was indeed the case then intensive use by virtual memory drivers, which don't increase usage of these parts, would NOT decrease the HD's life expectancy. On the other hand if most of the failures come from the head or any other part that will be affected by increased io then this would be a problem. gray
flatmas@ladder.cs.orst.edu (Scott Timothy Flatman) (06/02/89)
In article <2872@cadre.dsl.PITTSBURGH.EDU> cgw@cadre.dsl.pittsburgh.edu (Gray Watson) writes: [ question about hard drive failure deleted ] >>Ron Morgan {ames, utah-cs, uunet, gatech}!cs.utexas.edu!ut-emx!osmigo > >I guess the question here is: what part of the hard-drive usually fails first >due to longevity or use over a long period of time?? > >As far as I know, most of the hard disk failures, I've heard of, have been >bering and drive-motor problems. If this was indeed the case then intensive >use by virtual memory drivers, which don't increase usage of these parts, would ^^^^^^^^^^^^^^^^^^^^ [ some more stuff deleted ] >gray What?! How can this be true? Virtual memory has to constantly switch pages in and out of memory.In order to accoplish this the motor has to be on! This means more wear and tear on the motor and bearings. ----------------------------------------------------- Scott Flatman INTERNET: flatmas@ladder.cs.orst.edu UUCP: hplabs!hp-pcd!orstcs!ladder.cs.orst.edu!flatmas ---------------------------------GO BULLS!
usenet@cps3xx.UUCP (Usenet file owner) (06/02/89)
In article <10954@orstcs.CS.ORST.EDU> flatmas@ladder.CS.ORST.EDU (Scott Timothy Flatman) writes: [ Gray Watson notes that virtual memory drivers don't increase bearing and drive-motor wear ] >What?! How can this be true? Virtual memory has to constantly switch pages in >and out of memory.In order to accoplish this the motor has to be on! This means >more wear and tear on the motor and bearings. The motor on a hard drive is always running (unlike a floppy drive). The only increased wear, as far as I can see, would be to the stepper motor and related parts (which move the head back and forth). +---------------------------+------------------------+ | Anton Rang (grad student) | "VMS Forever!" | | Michigan State University | rang@cpswh.cps.msu.edu | +---------------------------+------------------------+
ags@mentor.cc.purdue.edu (Dave Seaman) (06/03/89)
In article <10954@orstcs.CS.ORST.EDU> flatmas@ladder.CS.ORST.EDU (Scott Timothy Flatman) writes: >Virtual memory has to constantly switch pages in and out of memory. Not true, as any NeXT user can testify. When there is page swapping going on on my NeXT, I can hear the disk activity. Right now, as I type this, there is only the motor hum. That means that the disk is rotating, as always, but there are no reads or writes. That's not surprising, because I have no memory-intensive background processes (such as Mathematica) running now, and it does not take 8 megabytes to run a terminal window. Depending on your usage patterns, virtual memory may lead to less disk usage, rather than more. If I am running Mathematica on my NeXT and I want to do something else, there is a brief clatter of disk activity as Mathematica is swapped out to make room for another process. When I go back to Mathematica, there is another brief clatter as pages are swapped back in. The total amount of disk activity in swapping Mathematica out and back in again is far less than what it would take to re-launch Mathematica from scratch and fire up the kernel again. By contrast, that is exactly what I have to do when I run Mathematica on my Mac II, using a 4 Mb Multifinder partition on a 5 Mb machine. There is hardly anything other than the finder that is small enough to run while Mathematica is active, so I have to kill it and re-launch it whenever I need to do something else. This causes a lot of unnecessary activity on my Mac disk.
amanda@intercon.UUCP (06/03/89)
In article <10954@orstcs.CS.ORST.EDU>, flatmas@ladder.cs.orst.edu (Scott Timothy Flatman) writes: > What?! How can this be true? Virtual memory has to constantly switch pages in > and out of memory.In order to accoplish this the motor has to be on! This means > more wear and tear on the motor and bearings. The motor's on whenever the power is on. That's what makes the cute little whirring sound... The only piece of the drive that would get any more wear would be the head stepper motor, and I've never heard of one of them failing. In my experience, anyway, hard disk failures seem to come in a few major flavors: media failure, main power supply failure, and controller failure. I can't think of any reason that paging would make any noticeable difference on any of these. -- Amanda Walker <amanda@intercon.UUCP> InterCon Systems Corporation
landman%hanami@Sun.COM (Howard A. Landman) (06/14/89)
In article <13548@ut-emx.UUCP> osmigo@ut-emx.UUCP writes: >It seems to me that if you used Virtual Memory as your "standard, everyday >setup," it would involve INSTENSE, CONTINUOUS hard disk accessing, in contrast >to the occasional "chirp" of the drive we now experience. No, it only involves disk accessing when you want to read or write something and it's not in memory. That means a "page fault". With enough memory, or a program that doesn't really use all the memory it has allocated, you just don't page fault that often. In a non-virtual system, you MUST have everything in memory at all times (unless there's some other form of swapping to disk), so if you need just a little more, BANG you're dead. Virtual allows you to organize your program in the simplest way without having to deal with such messy issues except in the general sense of trying not to use memory too wastefully. And if you overstep the limit, it degrades relatively gracefully - it runs slower, but it still runs. Virtual memory is like growing old - it doesn't seem so bad when I consider the alternatives ... Howard A. Landman landman@sun.com
kff@kesmai.COM (Kelton Flinn) (06/17/89)
In article <109873@sun.Eng.Sun.COM>, landman%hanami@Sun.COM (Howard A. Landman) writes: > In article <13548@ut-emx.UUCP> osmigo@ut-emx.UUCP writes: > >It seems to me that if you used Virtual Memory as your "standard, everyday > >setup," it would involve INSTENSE, CONTINUOUS hard disk accessing, in contrast > >to the occasional "chirp" of the drive we now experience. > > No, it only involves disk accessing when you want to read or write something > and it's not in memory. That means a "page fault". With enough memory, or a > program that doesn't really use all the memory it has allocated, you just > don't page fault that often. Actually, this same thought first occurred to me as I listened to the disk drive rumbling, when I switched from MPW to Versaterm for the first time. Unlike our MicroVax, the Mac does seem to thrash a LOT when you switch from one large task to another. I have 2 Mb of memory, about enough to hold 1 1/2 applications. Obviously having Virtual is better than not being able to run SADE and MPW at the same time at all! But it does grind more than I had hoped... Kelton P.S. A warning: Never run MS Word under Virtual under the Finder. It seems to want to initialize all 8 Mb of memory, and then checks it periodically. It took me 15 minutes just to get out! Beleive them when they say use MultiFinder.