When is a pagefile not a pagefile?

Discussion in 'Windows Desktop Systems' started by a1ehouse, Aug 16, 2002.

  1. a1ehouse

    a1ehouse Chamone M*tha Fu*ka

    Messages:
    288
    Hmmm.

    I have set my pagefile temporarily to 0. Coolmon reports that I still have a 482 MB pagefile. WTF?!?

    Anyone any ideas?
     
  2. cruiser78

    cruiser78 Me

    Messages:
    235
    instead of setting your pagefile size to 0, try selecting the "No Page File" option... it's under the custom size setting
     
  3. JJB6486

    JJB6486 Retired Mod Political User

    Messages:
    1,207
    Location:
    West Lafayette, IN, USA
    Did you reboot after changing the setting?

    JJB
     
  4. dreamliner77

    dreamliner77 The Analog Kid

    Messages:
    4,702
    Location:
    Red Sox Nation
    Coolmon gives me wrong pagefile size too.

    while we're on the subject of coolmon, I slected to hide the tray icon. How do I get it back? I can't do anything with coolmon now.
     
  5. a1ehouse

    a1ehouse Chamone M*tha Fu*ka

    Messages:
    288
    Thanks for replies - in answer:
    cruiser78 - Yup that's what I've done!
    JJB6486 - Yup that's what I've done!
    dreamliner77 -
    (from the Coolmon help file) :

    Must just be a Coolmon thang
     
  6. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    setting the pf to 0, or dissabling the pf, or having the pf too small will always make your putor slower...xp is a paging os...live with that...just understand that an active page file does not indicate using the hard drive instead of ram...ram is always used first...what you're looking at is the os getting ready...that's it...if it didn't prepare, then the virtual memory would be so much slower.

    Also, "no paging file" in no sense means no virtual memory.

    "virtual memory" is a few different mechanisms;...address translation, fetching of data and code from backing store, etc ("demand paging"), there's working set replacement, and also temporary storage of modified pages in backing store.

    If you eliminate the paging file, all you eliminate is this particular backing store file...exe's, dll's, data files, etc., will still act as backing stores, and they will do so more if there is no pf...you aren not eliminating "paging to disk", only paging to the pagefile...the other paging mechanisms are still going.

    Now, when xp does not have the pagefile, it will still write to disc, however, the os will not be able to discriminate, and instead of oldest ago information being written to disc, the os will be forced to write information that is not appropriate.

    believe me...it is definately counter productive to eliminate the pagefile no matter how much ram you have
     
  7. a1ehouse

    a1ehouse Chamone M*tha Fu*ka

    Messages:
    288
    Thx for that dealer, however, having no pagefile is not, or was never my end goal. I was having a mess around and noticed this anomaly in that I still have a pagefile even though it's set to 0 or "No Page File". I'd be interested if someone else has had the same problem, or has come across this before.
     
  8. dreamliner77

    dreamliner77 The Analog Kid

    Messages:
    4,702
    Location:
    Red Sox Nation
    Heres a ? for dealer-

    I moved my pagefile to my secondary hard drive (physical not partition). I also use Diskeeper7 for defragging. When I analyze the drive, I see that the pagefile is stuck in the middle of the drive. Shouldn't it be at the beginning to be optimized? What if I turn the pagefile off, run diskeeper, and then turn it back on? Would that help?
     
  9. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    ACTUALLY, no, it's a myth that files are optimum in the beginning of the drive...it's still repeated, much like the static page file thing... it's false

    this is so obvious, it's surprising the myth started in the first place.

    the heads are accessing info all the time, obviously, the heads are going to be located towards the center of the disc on average, more often then anywhere else, therefore, that's the most efficient place for a file.

    of course if you have a huge drive that's empty, it won't be in the middle, but it surely won't be at the beginning of the drive
     
  10. dreamliner77

    dreamliner77 The Analog Kid

    Messages:
    4,702
    Location:
    Red Sox Nation
    You're right. Never thought of it that way. Dealer, I think you have become the resident pagefile expert.
     
  11. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    my favorite topic...(could you tell?)
     
  12. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    well, I did some reading on your behalf, alehouse

    get this...from the excellant book..."inside windows 2000"...I read this from a library, so I have to paraphrase...if you have an interest, please read the text in the context from the source

    the nt kernal will create a temporary pagefile if yours has been deleted...it does this to give you back the stability that you would loose without a pagefile...it's not the same, or as good as the permanent pagefile, but it's created to keep you going, until you set the pagefile to a proper value
     
  13. 2z

    2z OSNN Gamer

    Messages:
    2,439
    Location:
    England
    Sorry dealer gotta disagree on this 1
    Your HDD works a bit like a the old vinyl record players only quicker & the heads dont touch the disk.
    If the data is on the outer tracks, more can be read/written in one revolution.
    see pic
    [​IMG]
    The time taken for the hard disk head to read from point A to point B is exactly the same as the time taken for the head to read from C to D. But because the areal density of the platter is constant, therefore more data can be read from the outer tracks compared to the inner tracks in the same amount of time.
     
  14. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    no...I'll be back with the ms documentation later...don't have time now.

    basically, there's burst speed, and seek time...the moost important place for the most accessed files is in the center of your information, that's where the heads are the majority of the time.

    I'll be back
     
  15. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    before we go further with this discuussion...let's point out, the differances we're talking aobut here are absolutely unnoticeable, the hardware today pretty much makes these points moot as far as any noticeable differance in performance on the consumer level.

    nevertheless, ms has done the research;...

    from this we get the following paragraph...

    "In Windows 2000 and earlier versions of Windows NT, the MFT was typically placed at the start of the disk space available to the file system. In Windows XP, the NTFS format utilities place the MFT 3 GB further into the disk space, which has been found to improve system performance by 5 to 10 percent."


    here's what's going on...

    The total time to complete a disk IO consists of the arm movement time, the rotational delay, and the actual transfer.

    pagefile reads and writes are usually very short (64Kbytes) ... the actual transfer time is about a millisecond at 66 Mbyte/sec, or two msec at 33 MByte/sec.

    Average rotational delay on a 7200 rpm drive is 4.2 msec...average seek time on a consumer HD is around 9 msec.
    See... The actual burst transfer rate is not the important factor.


    So if you want to make those IOs go faster, the most important thing to work on is the access time, that getsshorter if the arms fewer cylinders to cross.

    You've will cut the average access time from anywhere else on the disk to lets say the pagefile just about in half. Now, you are correct, you've increased the actual data transfer part of the IO time by let's say 30% as compared to having it the file on the outside, BUT...it's more than made up for by the improved access time. sort of like 1 step back, three ste[s forward...a net win

    BUT NOT SO ANYONE WOULD EVER NOTICE THIS DIFFERANCE, SO DON'T SWEAT IT
     
  16. 2z

    2z OSNN Gamer

    Messages:
    2,439
    Location:
    England
    Ok

    lets assume for a moment that all these ppl are correct
    http://partition.radified.com/partitioning_2.htm
    http://www.etestinglabs.com/bi/cont1998/199807/diskgrf2.asp
    in the old days of Win9x I would set a fixed pagefile defrag using Norton Speed Disk that would put the pagefile at the beginning of the disk BEFORE windows & all was well :)
    Now under XP you cant do that. & the perfomance gain from this will be that windows has the most prominent place on the disk this would give you the alleged **% increase in performance.
    Mmmm ;-(
    So much conflicting evidence its know what to believe. (seeing is believing)
    A little benchmark test for you pagefile
    Minimize a window (AOL is the worst on mine)
    Run a RAM Optimizer like ClearMem http://www.simtel.iif.hu/pub/pd/50971.html
    Maximize - the time it takes your App (PC) to become usable depends on the effectivness of the pagefile.
    Next time I do a fresh install I'll install XP to "D" & resize "C" to 1gig & format it FAT16 32KB cluster & put a 512meg pagefile in there & see how it goes.
    Later

    ;)
     
  17. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    ya...conflicting evidence.

    here's a case where the theory was oposite of the practical application...the theory being the heads start at the beginning of the disc, so that's the most efficient place for an often accessed file.

    the practical application being, no, it's not where the heads start, it's where the heads are located the majority of the time.

    no, if I designed a hardware configuration, I would just make the heads start in the middle of the information by defaultc...now this would actually translate into a noticeable differance...heads starting in the middle, most accessed files in the middle...best idea...I should submit this

    always remember, when it comes to high tech, the most recent information should always carry more weight then old information

    here's the most important point of our discussion

    I don't think anyone will the differance, regardless, the hardware is just so efficient.
     
  18. 2z

    2z OSNN Gamer

    Messages:
    2,439
    Location:
    England
    not from my experience :-(
    the benchmark test I mentioned using ClearMem & AOL
    AOL was taking up to 10secs to become useable
    Since moving the pagefile to a seperate drive in the primary partition (1gig FAT16 16KB cluster - 512meg pagefile)
    AOL is now useable in about 3secs.
    but I'm sure I'm can do better than that ???
    Later
    ;-)
     
  19. Perris Calderon

    Perris Calderon Moderator Staff Member Political User

    Messages:
    12,332
    Location:
    new york
    I really don't think you could notice the differance either way twoz..I'm pretty sure there is another factor in your equation....this to me seems obvious, since you are not maxing your ram when you launch aol
     
  20. 2z

    2z OSNN Gamer

    Messages:
    2,439
    Location:
    England
    Quite often my games will max out RAM
    I just used AOL as a benchmark to see how effective the pagefile is,
    Last thing I need when playing games is a slow retreival of data from the pagefile making the game less responsive & in some cases unstable.