Delivering fast boot times in Windows 8
blogs.msdn.com
blogs.msdn.com
Granted the computer was a couple years older than average users' these days, but psychologically it was just infuriating to think the computer is ready for me, and then having to sit through the "second boot." Instead of a super-fast boot they should focus on a super-simple boot which isn't torture to watch. I can walk away from my machine and do something else for a minute while it's booting, just like anyone does when they're popping popcorn. But the popcorn should be easy to open once it's "done."
Have a look at Autoruns[1], which is a great SysInternals tool to show you what is set to run at startup, and lets you selectively disable them.
[1] http://technet.microsoft.com/en-us/sysinternals/bb963902.asp...
I think of this every time: http://www.flickr.com/photos/ebwolf/498371086/
But I wonder how big a difference a smarter disk-elevator algorithm would make.
Or something that logged file-accesses, made a list of disk-block-no of that, sorted the list, and early at next boot read of blocks from that list. This is to avoid boot-time being spent waiting for random disk-accesses.
http://blogs.technet.com/b/askperf/archive/2008/03/28/startu...
I don't know how much this affects startup performance, but on my work Thinkpad Windows 7 boots up and is usable rather quickly.
Is the state of software engineering so poor that no one has considered staggered or delayed startup for some of these borderline useless apps? I mean, startup is a busy time in the computer world and on the server end of things we use delayed services all the time, but on the dekstop the assholes at Apple and Google and the rest just assume that their little startup app should run immediately and with full priority and so what if you have 5 others just like this. Heaven forbid someone consider the end user experience here.
I think MS had to step in and set these things to low priority starting with Vista. What a mess when the OS guys are fixing what the app guys are simply unable or unwilling to do.
Currently, the OS simply doesn't tell you when it's actually finished booting.
This is Vista I'm talking about, perhaps half the fault is mine for not updating to 7. And a clean reinstall might have ridden the computer of a few of these, but only a few. Also, I left out Adobe update because it's completely 3rd party but in light of your second paragraph, I agree. That shit got annoying fast.
And yes, the point I'm making and that others have reinforced is that the OS shouldn't pretend to be done booting up when it's clearly not. These things are part of the OS.
The key change seems to be a clever kernel session hibernation. They discard user sessions as those involve writing a lot of data (full hibernation) but save kernel session which can be written out and read back fairly quickly. Then there is multicore enabled resume - doesn't help with disk io but with decompression of hibernated data.
Microsoft does have some intelligent people working there. It's such a bummer it can't get some real work done outside of Office and Windows. I'd say their games unit is pretty solid too but everything else seems to be a real hit or miss scenario.
Neat reading these things though.
Seen elsewhere with improved battery life of the iPhone vs android phones, for example.
More precisely they do both. Close the lid and they save RAM to disk, then sleep. Open the lid and they resume instantly from sleep and discard the hibernation content, but wait long enough to drain the battery (or pop off the battery e.g for a swap on older Macs) and they will resume from disk.
IIRC Win 7 knows how to do that, though it might not be default.
That said, Windows Phone 7 is much more interesting to me than Android. Honestly, if I had the choice between Android and a WP7 device, I'd pick the WP7 device. I am a Mac guy so it makes the most sense to be on the iPhone. But if things change I am quite glad that Microsoft rethought their mobile strategy.
My only hope is that they don't get bypassed by other consumers so much that they can't maintain keeping the OS moving forward. I truly honestly believe that WP7 is a better choice than Android in the long run.
From what I've seen it is a really interesting experience. I've seen enough videos to see how it functions from a user standpoint and I have to say it is a really unique way of thinking things. I see features on it that I wish the iPhone had.
So while I dismissed the mobile unit from my original comment, I should ammend it to state that I really do like the way the mobile unit in Microsoft is going.
I haven't had the same problems since I switched to mac, aside from occasionally having the screen not come up immediately. This lets me run many things and not feel fearful about it all getting lost when I resume.
* that said, I did watch a non-techie guy always shut down his imac when he didn't use it, seemed awkward to me but eh.
http://blogs.msdn.com/b/e7/archive/2008/08/29/boot-performan...
In Windows 95 I had two options: reboot and poweroff. In Windows 2000 I had three options: reboot, poweroff and sleep. In Windows XP/Vista I have four options: reboot, poweroff, sleep and hibernate. Finally, in Windows 8 I will have five options: reboot, poweroff-which-is-not-really-a-poweroff, poweroff-which-is-really-a-poweroff, sleep, sleep-and-then-hibernate-if-battery-is-low.
True Microsoft Way. It remembers me funny examples from Alan Cooper book 'The Inmates Are Running the Asylum'.
I'm sorry but i have to call you out here; i would be suprised if MS hasn't thought of this in some way; whether detecting changes and initating a more complete boot or dealing with it in the GUI.. but even if it required you to select a 'full' shutdown/boot, i would say it was completely worth it for a 7s boot time.
Further, i find your imagination regarding the practical handling of these power states rather lacking. The way i see it there will simply be 'reboot', 'power off' (which hibernates the kernel as described in the article, reloading it where checks show necessary), and 'standby' (which also writes ram to disk allowing resuming from hibernate with power loss -- i would imagine this could be turned off with a setting). So, contrary to your rant, i am going to bet that there will be one less standard option. But i guess things were better in the w95 days?.
Sure, they can detect hw changes and make full cold boot, therefore losing all RAM data which was saved to the disk at the point of hibernation. (It would be like a quick dirty reset.) Or they can restore that data back to RAM and crash the kernel instantly. In either case, RAM contents are lost.
So each time I turn off the PC, I have to plan ahead, "do I want to use poweroff-type-I or poweroff-type-II option?" If I won't guess right, then I lose my data.
This is not a change from now; if you hibernate and then change your hardware, yes, you will lose data.
Still there could be some edge cases - like hibernating filesystem cache as a part of kernel-related RAM , then shutting down, booting Linux, changing the filesystem and then booting windows (which does not know that filesystem pages were changed from outside and working with obsolete copy from cache).
This one is a non-issue actually, because Linux ntfs-3g driver deletes hiberfil.sys when mounting NTFS partition in R/W mode, but there could be other edge cases like this one.
I don’t think the firewall adds anything significant to the startup time; it is just one of the dozens of services that start automatically.
Then, for many people an antivirus program may also be useful. Would that add 5 to 10 seconds to the startup time of a modern machine?
For the other 3—anti-rootkit, anti-spyware, keyboard scrambler— why do you think they are necessary for every user of Windows?
Think how long iptables takes to load, plus installing a couple of system hooks to detect unknown programs and a GUI window to warn about them. This stuff only takes a long time to load if you're loading in a huge framework and JIT at startup - startup times are very quick for C++ programs using Win32 API calls or WTL.
I'd say anti-virus and anti-spyware are mostly treated as one these days (anti-malware) - with things like Microsoft Security Essentials, Norton Internet Security, etc. I use MSE and it sure as hell doesn't add 5 seconds to startup. It doesn't even add 1. But perhaps it's because I'm running a 'post-modern' machine with an SSD...
I like what they did with WP7 in that regard.
And until XP they had a nasty habit of not doing updates on the consumer OS. ie. 95 and 95B, 98 and 98SE.
Also, I haven't used Windows laptops in a while, but the Sleep mode on XP was really terrible - if I selected Start -> Shutdown and then closed the lid, when I came back and opened the lid I would find Windows resuming the shutdown procedure due to automatic sleep mode.
Fast boot times are very important because there are lots of people who do need to turn their machines off. Maybe Microsoft have fixed various usability problems, but the electricity ones still stand.
Why have running electronics that you're not using?
More so than 8+ hours of use, do you think?
I've heard the myth often enough, though never any evidence either way.
Machine off, cold start to login screen: 6.7s
At desktop clicking restart to login screen: 5.9s.
Wake from sleep to login screen: 3.2s
Lift lid to login screen: 2.8s.
Got a citation for that? My PC seems to connect instantly.
Lion has also cut battery life in half for a minority of users, and my own machine refused to go to sleep until -- get this -- I turned off Internet Sharing in the Sharing preferences.
What?
I've never understood why I should care since it's something I do once a week or less for a few seconds.
Why would I want my home PC turned on at 3AM, exactly?
Ok, maybe that's just me, but I suspect I can't be the only HN reader who uses his home PC at 3 AM more than he does at 3 PM.
My father's laptop, larded with the corporate, err, "value add software", once took five minutes to pass that gauntlet, without even any local network activity. You do want to make sure you can pull open the browser and do a bit of real work, though; the desktop appeared in about a minute, but you couldn't actually do anything. With the Maverick release of Ubuntu, the same test could be passed in 45 seconds. (Still not what I'd call speedy, I just include that for reference. Yes, it's a laptop and the hard drive hurts, but the hardware wasn't that bad. Also, this was XP, which I know from experience on that hardware should have taken roughly 45 seconds too.)
You might understand why it's a bit of a sore point with many people whose wallets Microsoft is listening to after that experience.
(It appeared to me, based on the process list and the way the machine was performing, that the simultaneous presence of backup software trying to scan every file access and virus scanning software trying to scan every file access, and for all I know trying to each perform other real scans of the disk, was causing major disk thrash, which combined with the fact the virus scanner insisted on scanning various files before the OS could execute them caused major, major slowdown. I'd never put up with this on my machine, but somehow I suspect this is not an uncommon case in the corporate world...)
Not "and now you know why Windows is fundamentally incapable of starting up quickly". Phrased it that way on purpose.
I am not sure I like the idea.
Really so? It is not that people want to "start fresh", it is because the boot times from hibernate have been pathetic.
Now I am on a MacBook and I never almost shutdown, just sleep. Yes sleep consumes power, but what I mean is that I would not want to "start fresh" each time. No way.
How many people here want to "fresh start" at each boot?
For Windows? I only restart because I need a fresh start.
I do not use Windows in recent times, either Ubuntu or Mac. But still want to see a better Windows 8 so I can use that too.
And that is why in the article they state that the "RESTART" option will still give you a "fresh start". It is only when you "turn off" the computer when you get this kernel hibernation mode.