Page File Tweak Is False

Discussion in 'Windows Desktop Systems' started by Dirk Diggler., May 29, 2002.

  1. Let me ask something, and don't get knackered

    Does anybody here actually believe that for the past 3 os's, microsoft hasn't heard it all about this tweak...all about not letting it reamain dynamic...that it will cause fragmentation?

    Of cource they have heard about this...so I'm realizing this, and then the penny dropps.

    Obviously, everybody s wrong, cause they would surely make this simple adjustment, unless they felt there was a perfmance benefit to leaving the page file dynamic.

    So, I do a search, and sure enough, I find the following article

    Ed Bott's Windows Tips
    Tip #181: Swap File Secrets, Part 1
    Windows tries its best to keep you from running out of memory. Its most useful memory-managing trick is the Windows swap file, also known as virtual memory. By default, Windows 95/98 and Windows Me create this swap file automatically and manage its size dynamically. If you find a large file called Win386.swp in your Windows folder, you've just found the swap file. (If this file is located in the root directory of your system drive, that indicates that the swap file size has been manually adjusted.)

    Normally, Windows creates the swap file on the same drive as the one on which Windows is installed, and it manages its size automatically, from a minimum of 0 bytes to a maximum size that uses all available space on your hard disk. If you use Windows 98 or Windows Me, these are the correct settings. Don't be fooled by bogus Windows tips recommending that you create a permanent swap file of a certain size! That advice applies only to older systems running Windows 3.1 or Windows 95. In newer Windows versions, Windows is tuned to manage the swap file efficiently. You may actually harm performance by changing the swap file settings.


    and then this article;

    Ed Bott's Windows Tips
    Tip #182: Swap File Secrets Part 2
    In yesterday's tip, I recommended that most Windows users leave swap file settings alone. However, there is one specific set of circumstances where you may be able to improve system performance by manually adjusting swap file settings.

    If you have a second physical hard drive (not just a second partition on a single drive), you can improve performance by moving the swap file to a different drive from the one that contains your Windows files.

    To do so, open Control Panel and double-click the System icon. Click the Performance tab and click the Virtual Memory button. Choose the Let me specify my own virtual memory settings option. Finally, open the drop-down Hard drives list and choose the drive letter where you want your swap file to be located.

    Do not adjust the minimum and maximum sizes of the swap file, and above all, do not check the Disable virtual memory box! Click OK to save your changes, and reboot to make the new swap file settings effective.




    so, this long old wives tale might soon come to rest...huh?
     
  2. dreamliner77

    dreamliner77 The Analog Kid

    Messages:
    4,702
    Location:
    Red Sox Nation
    Remember, that's just one person's opinion. I could find plenty of stuff on the web that would say the exact opposite
     
  3. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    Actually, this isn't one persons opinion...it seems microsoft has well researched this tweak, and they insist it hurts performance...they make this case simply by desighning the os with the dynamic memory...they have every reason to make the memory static, yet they don't...that's a very strong case, isn't it
     
  4. Dirk Diggler

    Dirk Diggler Guest

    In defense of Mr. Diggler.,
    he wasn't himself when he posted this stuff. You could almost say he was totally out of his mind.
    lol
     
  5. xsivforce

    xsivforce Prodigal Son Folding Team

    Messages:
    8,547
    Location:
    Texas, USA
    I see he has his period. Get it? Look again and you will. :D
     
  6. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    oh man, I thought this was going to be fun, you guys found me right out...anyway, it is a fair discussion, isn't it?
     
  7. Dirk Diggler

    Dirk Diggler Guest

    I would think the reason why it isn't static is because Micro$oft can't know a persons memory (RAM) amount or a persons memory usage. If you have lots of RAM and don't use any memory hungry apps whats the point in having a large Swap/Paging file.
    If you have only limited amounts of RAM and use heavy memory apps, it would be daft to have only a small Swap/Paging file. Therefore to cater for the majority Micro$oft always use a non-static Swap/Paging file.
     
  8. xsivforce

    xsivforce Prodigal Son Folding Team

    Messages:
    8,547
    Location:
    Texas, USA
    [​IMG] <-----BUSTED!!!
     
  9. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    then it seems this is the answer for everybody that has tons of ram...let the os adjust your pagefile, and it'll remain as small as possible, and possibly page less then a static file...someone should experiment
     
  10. Dirk Diggler

    Dirk Diggler Guest

    Hey X,
    your arrows in the wrong place.
    Yours should read "TUSHIED"
     
  11. allan

    allan Guest

    I agree that until I see a white paper from MS, the info posted above by Dirk are opinions, not fact. It still seems logical to me that a static paging file will be more efficient than a dynamic paging file and for the time being I think I'll stick with that.
     
  12. Dirk Diggler

    Dirk Diggler Guest

    Yes but only if your prepared to defrag it every so often, say maybe every hour or so.?????????
     
  13. Dirk Diggler

    Dirk Diggler Guest

    Hey allan are you talking about Dirk's post or Dirk's posts

    Confused, you will be :D
     
  14. xsivforce

    xsivforce Prodigal Son Folding Team

    Messages:
    8,547
    Location:
    Texas, USA
    Good one DoubleD. ;)
     
  15. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    well if you have tons of ram, the page file will remain static, even though it's set to be dynamic...it'll never change, as you'll never need the virtual memory...fragmentation will therefore not be an issue
     
  16. Dirk Diggler

    Dirk Diggler Guest

    The trouble is, Windows will use the Paging file. It doesn't matter how much RAM you have, it will still use the Paging file. Its just one of those odd special features of Windows I suppose.
    :p
     
  17. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    well, we're all used to seeing posts from people that actually know enough to actually change the settings of this file, so, I don't have enough ram to experiment, but someone with say a gig, should try this experiment, and actually see just how small the page file gets, and then monitor whether or not the box is paging
     
  18. ghayes

    ghayes Microsoft MVP

    Messages:
    94
    In the grand scheme of things from a performance perspective, fragmentation of the pagefile is far down on the list of things to worry about. When Windows reads the pagefile, it does so in "chunks" of a finite size - no matter how many pieces the pagefile is in.

    What is the downside of a dynamic sized pagefile? If Windows has to take time to change the size of the pagefile on the fly - it takes time & resources to do and you can notice applications or Windows "stalling" while it does this. If the pagefile is static (fixed in size), this will not occur. The primary benefit of defragmenting the pagefile (which can NOT be done online and requires a boot time defrag) is that it allows defragmenters to do a much better job of defragmenting the files that fragmentation causes more of a performance problem with - your frequently accessed data and applications.

    - Greg/Raxco Software

    Disclaimer: I work for Raxco Software, the maker of PerfectDisk - a commercial defrag utility, as a systems engineer in the support department.
     
  19. Dirk Diggler

    Dirk Diggler Guest

    I have one and a quarter gigs of RAM, and my Paging File usage goes from a minimum of around 25 Megs up to 256 Megs (as I speak).
     
  20. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    Well you're not going to believe this...I actually asked jaimie, of the io page lock fame, and he adds to this discussion in the following email to me;

    Hi,

    Here's a thread over at 2cpu.com where I've covered this and related issues pretty thoroughly:

    http://forums.2cpu.com/showthread.php?threadid=11689

    Although it's in a Win2K forum it's all still true for XP.

    It's true that, if automatic pagefile expansion is enabled, the expansion area (anything allocated past the default) will probably be fairly fragmented, as it will have to use whatever chunks of free space happen to be in the partition at the time expansion happens. BUT -

    - this doesn't affect the original extent(s) of the pagefile's default size. Those stay in whatever fragmentation state they're in. If the original default size of the pagefile was one extent, then that much of the pagefile remains as one extent, even if the system expands the file later.

    - the system only expands the pagefile if and when it needs to. It only needs to if your "commit total" starts approaching the "limit". This is also when you see the "system is running low on virtual memory" popup. Before that, the system spends no extra time "managing" the file just because expansion is enabled. Nor after that, if the first expansion gave it enough space.

    - the (possibly fragmented) expansion regions get deleted the next time you reboot, so the pagefile reverts to its default size -- and its original fragmentation state. So the "fragmentation" effect of the expansion mechanism is completely transient.

    - so, once the pagefile is defragmented there is really no need to defrag it ever again, unless you change its default size. And possibly not then either, because...

    - even if the pagefile IS fragmented, this hurts performance a lot less than most believe, on most systems. Most of us really don't do that much IO to the pagefile, and even with a fairly badly fragmented pagefile, most pagefile IOs don't cross extent boundaries anyway, so aren't made slower by fragmentation of the file.

    - in fact, fragmentation of files in general hurts performance a lot less than most believe, for MOST uses of the disk.

    - all that said, you really don't want the system to have to expand the pagefile, so you do want to set the default size to whatever you think you'll ever need, plus a bit of extra. 1.5x the RAM size is a good initial estimate for that. But there is no penalty for having pagefile expansion enabled. There IS a penalty if the system routinely needs to expand the pagefile, but it's better to let it do that than to have apps crash for "out of virtual memory". i.e. a slow-running app is better than an app that dies.

    - the best way to optimize your pagefile performance is to stop worrying about where it is and so on; simply add more RAM.

    btw, re something else you wrote - there's no case in which XP (or NT, or 2K) EVER creates a pagefile on each disk! I routinely install XP on systems with multiple physical disks and I've never seen that behavior. Left to itself it simply creates one pagefile in the boot partition (that's the one with the \windows\... or \winnt\... directory), default size 1.5x RAM, max size 3x RAM.

    --- Jamie Hanrahan
    Kernel Mode Systems
    Windows Driver Consulting and Training
    http://www.cmkrnl.com/