Stories
Slash Boxes
Comments

News for nerds, stuff that matters

Slashdot Log In

Log In

Create Account  |  Retrieve Password

Boot Windows Faster, Using Linux

Posted by timothy on Sat Jan 24, 2004 07:30 PM
from the linux-to-the-rescue dept.
BiOFH writes "TechNewsWorld is reporting that InterVideo has a solution for slow boot times runing Windows XP MCE. 'The new Linux-based InstantOn software -- designed to help Windows XP Media Center Edition PCs boot more quickly -- is aimed at taking advantage of the power of Intel's Pentium processors, not at fixing fragmented hard drives. The software integrates into the computer's BIOS and the operating system.'" According to this article, the software uses a small Linux partition on the user's hard drive. I wonder how BIOSes with hard-wired Microsoft-based DRM would cooperate with this scheme.
+ -
story
This discussion has been archived. No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More
Loading... please wait.
  • Vaporware! (Score:5, Informative)

    by LostCluster (625375) * on Saturday January 24 2004, @07:32PM (#8078017) Homepage
    It doesn't make Windows boot faster. It's just a stripped-down version of Linux which of course is going to boot faster because it provides far less functionality. If you want to get to full Windows, you'll have to wait out the remainder of the boot process you interrupted.

    Any CD-based Linux distro can achieve the a similar effect with far more functionality.
    • "Any CD-based Linux distro can achieve the a similar effect with far more functionality."

      Maybe I'm just ignorant, but I really doubt you can show me a live CD that can hand off to an installed Windows.

      Furthermore, live CDs, while great, are not the solution to slow boot times.

      And just cause it's related, I set up my VectorLinux (with kernel 2.6.1) to boot right into KDE, and timed it against Windows XP (on the same computer, so there's no hardware advantages). From pressing enter in Grub to being inside
      • If its been on for 92 days then you are surely vulnerable to a number of bugs whose patches require restarts.
      • Re:Vaporware! (Score:4, Informative)

        by AstroDrabb (534369) on Saturday January 24 2004, @08:38PM (#8078350)
        It is not pointless. Linux is USED for all the multimedia in the device and not MS Windows. The DVD, TV, FM, etc are all handled by LinDVD. This allows the device to boot really fast when you want it for multimedia purposes. If you want/need to do normal desktop stuff, that is when you boot up full MS Windows.
      • Re:Vaporware! (Score:5, Informative)

        by ortholattice (175065) on Saturday January 24 2004, @09:13PM (#8078569)
        But there is NEVER a reason to shut down a Windows XP computer (if you're not installing anything or changing settings). My computer has been on for 92 days and is still as stable and fast as it was on day 1 (super fast and 100% stable).

        Since you don't install the security updates (which require a reboot) I certainly hope your computer isn't connected to the Internet. Otherwise it's probably been turned into a zombied relay for spammers who are all too happy with your 92 day uptime.

      • Re:Vaporware! (Score:4, Informative)

        by Rasta Prefect (250915) on Saturday January 24 2004, @10:54PM (#8079084)
        But there is NEVER a reason to shut down a Windows XP computer (if you're not installing anything or changing settings).My computer has been on for 92 days and is still as stable and fast as it was on day 1 (super fast and 100% stable).

        You know, when you're not using your computer for long periods of time (say, 92 days) you should shut it off. Needless to say, my experiences with XP's stability have been a bit less sterling than yours. (Generally after a few days it's good and ready for a reboot. If I'm developing, once a day minimum).

        • Tell me, how is it 100% stable if you have to restart it to install things or change settings?

          Stability and having to re-boot to install *certain* software packages have nothing to do with each other. Yes, the need to re-boot for certain installs is a weak point for Windows, but that's not the same thing as stability.

          It's understandable that many here do not like Windows. But many people also understand that certain applications don't run on Linux, nor have *nix equivalents. We who must use these app

          • Re:Vaporware! (Score:4, Informative)

            by aardvarkjoe (156801) on Saturday January 24 2004, @09:10PM (#8078550)
            so turning a computer off doesnt 'save' any energy.

            A computer is hardly an efficient space heater. (An effective one, yes, but mostly because current systems use huge amounts of power.) If you want to save energy, turn off the PC when you're not going to be using it and use the furnace.
                    • Most modern machines can only be completely shut down by pulling the power plug.

                      Hibernate is called the S4 sleep state. It is still using power because some peripheals can wake the machine. Wake-on-LAN, Wake-on-Ring, etc.

                    • Re:Vaporware! (Score:3, Interesting)

                      Yes. S5 and/or pulling the plug was what I was alluding to when I said "turn it off," and you can't get through airport security (or at least you shouldn't be able to) with a hibernating laptop.

                      I'm afraid I've been feeding a troll who came in yelling "moron," but who clearly has never measured current draw of computers in various states.

                      If the computer consumed no power how would it be possible to Wake-on-Lan and/or without a full software reboot.

                      Any form of "Instant On" is using some power for something
              • If you put your computer in "hibernate" it is the same as power-off, which is 3 watts of energy.

                Your may very well be right, but what exactly is the 2 or 3 watts used for when the computer is powered off? I can put my (fairly new) Linux-laptop into hibernation (writing all memory pages to disk and turn off), remove its battery and go mountain climbing for a week. It will still power up and recover quite nicely. Of course the battery to keep the clock running is there, but in my experience it uses a great
      • Re:Vaporware! (Score:5, Informative)

        by Anonymous Coward on Saturday January 24 2004, @08:48PM (#8078424)
        Not quite. On my machine (1.8Ghz P4) Redhat 9 and Windows XP Professional both take about the same amount of time from power on to CPU idle (ie: waiting for something to do). Windows XP is faster to present a login screen, but the machine is still not useable until quite a while longer after login.
  • by spun (1352) <loverevolutionary&yahoo,com> on Saturday January 24 2004, @07:34PM (#8078021) Journal
    It boots Linux faster, offering a choice of several entertainment related programs, as well as the choice to boot windows, which takes as long as usual.
      • Parent wrote: "Someone should really ask them when LinDVD will be launched to the public".

        Yeah, that would be cool. I've seen it play, and it's pretty much just like the Windows WinDVD.

        Another article on their LinDVD based Instant On product in NewScientist [newscientist.com]

  • Common sense (Score:4, Informative)

    by Anonymous Coward on Saturday January 24 2004, @07:36PM (#8078032)
    What About the old "Don't load programs you don't need to load at startup"? Prefetcher tweakage. (yay for bootvis) Killing ad / spyware, tweaking services? My XP boot fairly quick (if I *enter* out of my 30sec countdown from my Xp bootloader asking me if I want Linux Or windows today.) Who doesn't know that isn't very likely to install a seperate Linux partition just to boot quicker?
    • Re:Common sense (Score:5, Insightful)

      by kfg (145172) on Saturday January 24 2004, @08:03PM (#8078181)
      Slow start up times are the price of using programable, general purpose machines. The ultimate way to reduce startup times is to hardwire the specific functionality you're looking for, as in conventional TVs or stereos.

      Ya pays yer money and ya takes yer choice, as they say.

      KFG
      • The bootvis page explicitely says that it's a poor choice of a tool by end users for improving boot times. So, um... what do you use it for that makes you impressed with it?

        It impresses me by shaving significant time off boot times. However Microsoft describe it bootvis can do a lot, depending of course on how fast your boot is already and the factor that is slowing it down. If you load a lot of services at startup then bootvis is a great help.

        Or didn't you realise that what Microsoft say, and what

  • LinuxBIOS (Score:5, Insightful)

    by wowbagger (69688) on Saturday January 24 2004, @07:37PM (#8078033) Homepage Journal
    Sounds like they are using LinuxBIOS plus some apps for the quick boot option.

    Now, the question is, will Joe User start asking himself "Why can't EVERYTHING run this quickly?", and will the companies start realizing that everything CAN, IF they port their stuff to Linux?

    (NOTE: Obviously there is one company that is unlikely to take this action, but perhaps others might.)

    Of course, there is always the option of embedding Windows into the system ROM as well.

    (shudder)
  • by Valar (167606) on Saturday January 24 2004, @07:37PM (#8078035)
    Which is what I thought when I read the writeup. It is actually a minimal media-distro designed to boot quickly. To do windows stuff, you still have to wait for windows start time.
  • by a XOR b XOR a XOR b (744728) on Saturday January 24 2004, @07:38PM (#8078045)
    Here it is!

    F000:E05B call check_for_linux
    F000:E061 jc do_error_beep_and_halt
    F000:E063 nop
    F000:E064 nop
    F000:E065 nop
    F000:E066 int 19

    All kidding aside... I write BIOS code for a living, and this scares the crap out of me. What Microsoft wants is to basically eliminate the BIOS, except for the jump to the OS code (the "int 19" above). Windows already does just about everything that we do in the BIOS, like PCI device enumeration, etc...

    No doubt, this would make Microsoft's life a lot more simple, but I think it would give them too much control -- way too much. DRM would just be the start of it.

    I wonder what the EFI proponents (Intel) think about this deal...
    • by adrianbaugh (696007) on Saturday January 24 2004, @07:45PM (#8078082) Homepage Journal
      I thought Linux also re-did (or had the ability to re-do) pretty much everything the BIOS did, purely to fix up cretinous BIOSes that didn't do their stuff properly. I can see why that would scare you as a BIOS programmer (not knocking your personal ability, you understand) but surely the simple answer is for the BIOS industry to improve its standards so that OSes don't have to incorporate numerous workarounds.
      • by runderwo (609077) <runderwo@mail.w i n . org> on Saturday January 24 2004, @08:59PM (#8078484)
        Some things that the BIOS does can't be re-done. For instance, Athlon 64 SMP configuration must be done by the BIOS and cannot be done after the kernel has been booted, like x86 MPS can. Other types of things absolutely have to be done at initialization time for you to even get to the kernel.

        LinuxBIOS project has the right idea by ideally cutting out as much cruft from the system firmware as possible and leaving it up to the OS to perform initialization, but in reality some tasks are forced onto the firmware by design.

    • by ctr2sprt (574731) on Saturday January 24 2004, @08:47PM (#8078415)
      I don't understand why it scares you, or why it gives MS any more control than they have now. If you more or less eliminate the BIOS, it means the operating system needs to do more work itself. Big deal. All modern operating systems ignore 95% of the BIOS anyway. It wouldn't be a significant change from the current situation, and OSes are much smarter than BIOSes anyway. They do a far superior job of resource allocation.

      Now consider the scenario where BIOSes get bigger. Remember that BIOSes are on a chip, which makes them damn hard for normal home users to replace or modify. If some DRM crap gets put in there, it's nearly impossible to remove. Now that's the part that's scary. The BIOS might refuse to boot unrecognized OSes, in which case you're SOL. But if it's the OS that's handling DRM, well, someone will have a crack for it a week before the OS comes out. Or you can uninstall the OS and run one without DRM, like Linux. Or you can install Linux and write some DRM software. Or whatever.

    • In the XBox. Which is kinda-sorta a PC, but not quite, because it dosent have a compatable BIOS.

      If MS was to start producing BIOSes, which Im sure they could do, they would have to maintain compatability with the existing BIOSes of the world.

      There are pleanty of things that are not MS OSs that use the BIOS. Ghost. PXE. DOS before Netware (do they still do this?). Recovery CDs. And of course the OSS OSs.

      I have no idea how much the license for something like Phoenix BIOS costs. Less then a dollar per mobo, Im sure. Lets say that MS starts giving away their BIOS: How many PC hardware manufacturers are going to switch, to save pennies, at the risk of no longer making PC hardware? The hardware world has settled on using industry standards a long time ago. Not even MS can change that.

  • by TubeSteak (669689) on Saturday January 24 2004, @07:39PM (#8078048) Journal
    A partition on the hard disk houses the Linux software in an area separate from the Windows operating system. The developer claims that with the new software, the boot time is brief -- a mere 10 seconds. When a user turns on the PC, the InstantON software preempts the Windows boot sequence and takes over, quickly loading basic entertainment functions.


    Business Applications

    Moving from the quick-boot entertainment functions into Windows, however, will take users more time than the initial 10-second boot because the InstantON software must hand off the user to the Windows operating system at that point. Still, the company believes that most users inclined to use PCs as entertainment machines in their living rooms are accessing those machines mainly for entertainment-related functions, rather than to run business applications like Microsoft Word or Excel.
    1st step towards seperating 'entertainment centers' from general computers. "Ro foresees a market emerging for computers that don't have conventional operating systems but are used in the living room as entertainment devices, right beside TVs." And since this technology is being marketed towards OEMs, HP and Gateway etc, I can't imagine that it will compromise WINXP Media Center's DRM.
    • ". "Ro foresees a market emerging for computers that don't have conventional operating systems but are used in the living room as entertainment devices, right beside TVs.""

      He's a genius! Or would have been if he made that statemnt 25 years ago.

      Shesh. VCR, has a computer(albiet primitive) cd player, dvd player, stereo, Tivo.
      The market is here Ro.
  • by LostCluster (625375) * on Saturday January 24 2004, @07:39PM (#8078049) Homepage
    Windows Media Center is meant to be a TiVo clone. In order for it to record the shows you want, you need to leave it up at all time. This stripped-down Linux just isn't going to make the cut... the proper mode of operation is to simply avoid rebooting by leaving it always-up.
  • Didn't we just (a few months ago) have a /. story about how to speed up linux boottimes (by parallelizing stuff) so it could stay competitive with WinXP? Somehow I don't think this is going to help.

    Meanwhile, my shiny new RHEL 3.0 box isn't mounting NFS filesystems on boot because the network hasn't finished initializing yet. Apparently it takes the network about 30 seconds to come up. Wonder if that's a gigE thing. :(

  • by Anonymous Coward
    Who does it more than twice a year? Oh wait... this is Windows. Nevermind.
  • by techvd (744989) on Saturday January 24 2004, @07:50PM (#8078114)
    When the PC is hibernated, it comes back up much much faster than a normal boot. Most PC/laptops on market have had support for hibernation for a while. Except when necessary, why not get rid of a complete boot process and just stick to hibernation? It's no Instant On, but a lot better than a complete reboot!
    • That was my first thought. XP resumes from hibernation very quickly. The only reason not to do this would be if instability and memory leaks progressively mess things when the system isn't rebooted frequently. XP seems stable enough that this isn't a problem. It might be an issue for applications though.

      There's also suspend, and it can be almost instant. Suspend to RAM can use less than 5 watts. That's definitely acceptable for a home entertainment system. It's within the range of power used by TVs

  • by Our Man In Redmond (63094) on Saturday January 24 2004, @07:50PM (#8078121)
    "I'll take Irony for $1000, Alex."
  • by Anonymous Coward on Saturday January 24 2004, @07:53PM (#8078132)
    The solution to the long boot time problem isn't quicker boots, it's getting rid of the need to boot or reboot! Think about it, Handheld devices are designed so that they don't need regular reboots. Embedded devices are the same way. My Tivo takes forever to boot up, much longer than my laptop running Windows or Linux, but it doesn't bother me because the only time I reboot it is when it's moved! The solution to these issues is not faster on time, it's always on! This is where Linux has a big lead too, even though WinXP is much better than 9x in terms of stability it still can't beat Linux when measuring stability in weeks and months...
  • by digitalhermit (113459) on Saturday January 24 2004, @07:54PM (#8078138) Homepage
    There was a project a while back to take a snapshot of a boot state then load this snapshot directly into memory. Any modern harddrive can move the 40M or so in a few seconds. The sticking points were mainly due to hardware that needed initialization and some OS design issues (beyond my understanding, but had to do with how control is passed to the operating system). If not for these issues, the machine could boot completely in seconds.
    • by nacturation (646836) <nacturation@@@gmail...com> on Saturday January 24 2004, @09:07PM (#8078536) Journal
      Check out notes for the KeyKOS project:

      http://www.eros-os.org/project/novelty.html#persis tence [eros-os.org]

      There's an interesting story regarding Novell there. Anyway, that OS would take snapshots of the entire memory state every N seconds so that even if you pulled the plug out of the wall while the machine was running, you'd be back up to where you left off (minus some seconds) as it simply reloaded everything from disk again.
      • by Jerf (17166) on Saturday January 24 2004, @10:42PM (#8079035) Journal
        Current hibernation modes would not be able to do that because they end up storing various memory-based file-system caches in the hibernation as well. You do not want those caches to get out of sync with the real contents of the disk; big lossage.

        "So avoid saving those caches...", well there's a reason they are saving them. If you don't save them, you have to first close all open files, and now your "suspension" isn't transparent. So you've opened the door to two "modes" of program starting, "suspended" and "not suspended", or some such other crap. And re-loading the cache with new, valid data on startup takes time, which defeats the point of this in the first place.

        My point is not that there are no solutions to these problems but to try to give a taste of how these things cascade rapidly. OS design is a subtle and tricky work, which only becomes truly apparent once you actually try to sit down and code solutions to these problems.
  • by Ageless (10680) on Saturday January 24 2004, @08:06PM (#8078197) Homepage
    This has to be one of the most misleading articles, and even more misleading /. blurbs I have ever seen. This software has nothing to do with Windows. It's a stripped down version of Linux that has basic media center programs. It "integrates" with the BIOS by "booting" like every other operating system.
  • by cmacb (547347) on Saturday January 24 2004, @08:27PM (#8078288) Homepage Journal
    My recollection is that Windows boot times first started getting bad (WFW booted pretty fast comparatively) when publications like CNet were bending over backwards (or was it the other way) to show how much faster Office was than competing products. The benchmarking consisted of: (1) boot both systems, (2) start timers, (3) start application, (4) start benchmark series, (5) end application, (6) stop timers.

    Lo, and behold, more and more initialization work for Office, and then IE, started showing up in the Windows boot sequence.

    Merging applications into the OS is BAD DESIGN, but it won the poorly thought out benchmarks that many organizations used to select their "productivity" tools. Now Windows, and Windows users will pay the price. Serves them right.

    Sure, leave your Windows machines running 24/7 to avoid the boot delay. Linux and OS X users have that option too, but for them it is truly an option, not something they NEED to do.
      • Where does cmacb ever mention XP? Cmacb is talking about something that started to happen in Windows long before XP came out. XP is a fairly quick booting OS, I'll admit, but previous versions of Windows weren't.

        However, I think a lot of it is really the GUI and memory resident utilities loading. Try booting Linux into a GNOME session running Enlightenment as the window manager with a ton of applets, buttons and suchlike gewgaws. Almost every Windows box I ever had to service had far too much crap like that running, not to mention spyware. It's a wonder some of them even booted at all, and given a clueless user, a Linux system could be just as bad.

        God knows mine was when I first started playing around with X Windows. :-) Look everyone, Enlightenment v0.4 with the Aliens theme and a dozen applets running on a 386 with 32 megs of RAM. Hey Rocky, watch me pull an X Server hard lockup out of my hat! (but try telneting in to reboot a frozen windows machine...)
      • by cmacb (547347) on Saturday January 24 2004, @09:55PM (#8078775) Homepage Journal
        Well, I DO know something about optimizing large systems, even though I don't work at Microsoft. Judging by your photo, most of my work in that area was done when you were very very young.

        In a character mode system you can put a prompt up for the user almost immediately. You don't actually have to be ready to process his input yet, just grab keystrokes. You finish initialization while the user is still thinking and typing. You don't need to initialize memory at all. Unless your code is buggy to begin with and you need zeros there for debugging, allocate space as needed and allow it to start out random. I know most modern systems don't do this, but it's been done.

        GUI systems are of course a lot more involved. Can the part of the system that draws the screen and all the icons be isolated so as to get that up before loading all the common elements from Internet Explorer, Word, etc (and NO, they don't show up as separate entries on the task list!)? Yes, they probably could, but as those components get more and more bloated they each have their own initialization requirements.

        I'm quite sure people at Microsoft don't sit around twiddling their thumbs and TRYING to make the system less efficient. They DO have a vested interest in making ordinary PCs less and less desirable each year so that the hardware upgrade cycle needs to continue. Every new version of Windows has a larger memory footprint and that has to do with putting more and more stuff on the critical path between turning the system on hand handling that first mouse movement.

        I was doing my OWN benchmarking of these systems during the NT 3.5-4.0 timeframe and I DO know what I am talking about.
  • Fujitsu Lifebook (Score:4, Interesting)

    by jmichaelg (148257) on Saturday January 24 2004, @09:02PM (#8078502)
    My business partner has a Fujitsu P1100 Lifebook. [fujitsupc.com] It runs on Windows and takes quite awhile to boot up but once it's up, she never turns it off. When she's done using whatever program, she just closes the lid and it goes to sleep. When she needs to use it, she opens the lid and 10 seconds later she's back doing whatever she was when she last closed the lid. She bought the extra large battery so it'll run for 7 hours or so between charges. It's pretty neat.

    The only downside is the screen is very small so if you're at all far sighted, it's hard to read. Not a problem for her so she's happy.

  • Too complicated... (Score:3, Informative)

    by ByteSlicer (735276) on Saturday January 24 2004, @09:13PM (#8078568)
    If you really want to boot Windows XP fast, configure your BIOS to do a suspend to RAM on sleep. When you hibernate XP, the computer will be completely off (except for a tiny current for self-refreshing the DRAMS). From this state, booting will take only about 5s. And all programs you had previously running will still be there. Even music will continue playing where it left off.
    The only drawback is: if you lose power, the DRAMS will be cleared. That could be solved by a UPS or maybe some built-in battery.
    • I meant 'Standby' instead of 'Hibernate'. Hibernating will write your RAM to disk. It is controlled by the OS instead of by the BIOS. Reading half a GiB worth of RAM back is significantly slower than resuming from suspend to RAM. But in hibernation state the computer is really off, so no risk of losing DRAM content.
  • by gmuslera (3436) * <gmuslera@@@gmail...com> on Saturday January 24 2004, @09:26PM (#8078640) Homepage Journal
    That is nothing new... with Linux I accelerated Windows to 9.8 m/s^2
  • by Animats (122034) on Sunday January 25 2004, @03:31AM (#8080060) Homepage
    Just think of all the things that have to happen during boot-up today:
    • Displaying Microsoft logo.
    • Enumerating peripherals.
    • Waiting for the nonexistent floppy drive to time out.
    • Checking file signatures for files covered by "file protection".
    • Re-homing the scanner.
    • Restarting the print queue.
    • Loading Internet Exploder.
    • Loading Microsoft Office.
    • Loading every DLL that contains anything those two ever need.
    • Starting services nobody except attackers ever use.
    • Loading fonts into memory and generating bitmaps for them.
    • Bringing up the PPPoE connection.
    • Checking for new updates to Microsoft software.
    • Downloading new virus signatures.
    • Loading spyware and adware.
    • Loading latest ads for home page in browser.
    • Every dealt with the Power On Self Test of Sun systems? My E3500 takes minutes to get to loading the kernel, because it has to check all eight processors, all 32M/cache, and all 4G/ram.

      That 25 seconds is a blessing.