While Windows Vista has a whole host of new features to offer, it has one major problem that just won’t go away: it’s totally FUBAR’d after you resume from sleep or hibernate. Unfortunately, many of these issues weren’t present during the beta stage, and were somehow introduced in the RTM build of Windows Vista. This exclusive NeoSmart Technologies report describes some of the symptoms in detail, and we even provide links to possible fixes by Microsoft. All issues have been duly reported and confirmed by Microsoft, so this isn’t just some figment of our imagination. A number of these patches are scheduled to be included in Windows Vista SP1 (Codename Fiji).
Few computer “enthusiasts” turn off their PCs. Even with Bill Gate’s promised 6-second-boot (we’ve clocked an average of 42.6 seconds here on 8 different PCs), turning on a PC via a cold-boot requires waiting for all the various programs to load, the network to establish, the security policies to propogate; and you don’t get to brag about never turning off your PC – plus your uptime restarts. The alternatives were either hibernation (for laptop owners) or “Deep Sleep” for the rest.1
Throughout the beta, Deep Sleep in Windows Vista went great. It’s the default option (so long as it’s configured in the BIOS) when you click the shutdown button.2 It would put your computer in a low-power mode that recovered in a matter of 2 or 3 seconds, and didn’t crash! But in the final version of Windows Vista, something is very, very majorly wrong. On 6 of the 8 tested systems,3 recovering Windows Vista from a hibernate or Deep Sleep results in one of the following:
- When recovering from a hibernate: “Cannot find uxtheme.dll” appears whenever you attempt to run (almost) any program. No matter what you do, you can’t even run Task Manager. What’s worse, a restart doesn’t fix it, and because Windows Explorer also fails to launch with this error, you need to boot from the DVD and use System Restore – Safe Mode won’t work!4
- Failure to establish a network connection. Everything looks OK, but you can’t connect to the internet. Your LAN signal will be there, but the internet just doesn’t work. You must restart to fix it.
- Poor performance: though Task Manager will show normal CPU load, some of the drivers (they don’t appear in TaskMan) will attempt to use 100% of the CPU, resulting in a very laggy PC. You need to restart to fix it.
- No DWM. For no reason, DWM just won’t re-appear. This happens on ATi and nVidia, with or without the latest official drivers from the companies themselves. Manually running “dwm.exe” doesn’t work, you need to restart to fix it.
- BSOD on recovery. This is usually caused by the video drivers, and may or may not indicate something wrong with the kernel itself.
- No sound. Vista goes mute. Nothing you can do about it, no way to revive it, you just have to restart and let the re-done sound-stack load-up the way it should.
All of the above errors and more occur randomly and make using hibernation down-right impossible (unless you’re willing/eager to run System Recovery from the DVD!) and Deep Sleep a waste of time (seeing as you have to restart to “quick recover”). Most of the errors are indicative of a problem somewhere deep in the kernel, and it’s not going to be easy to fix it. Some people are blaming this on the PC/Hardware/BIOS itself, but it’s not the BIOS’ job to support the OS,5 and the only thing to blame here is a buggy ACPI model.
We’ve notified Microsoft of each of these errors, we’ve been told they’re real bugs and a fix is in the works for some issues, others are just as much of a mystery. Some of these can be solved when ATi and nVidia release their final (hopefully bug-free) drivers for Vista. Others may not be as willing to go away. Either way, an operating system that you have to shutdown in order to save on power isn’t exactly the biggest business model. The only good news is, this bug only recently made its way into Vista, so that may just mean it won’t be too hard to squash. For now, if you really need to keep your PC on all day and all night, check the list below for hotfixes that may work for you.
Patches by Microsoft (Updated 06/12/07)
Here’s a list of patches by Microsoft related to Vista and wake/resume problems. You may have to call MS directly for access to some of these patches. Stop errors are blue screens (BSODs).
- [KB-928135] – Windows Vista hangs on resume/wake
- [KB-928135] – USB-Related crashes on resume/wake
- [KB-929734] – Many problems on resume/wake
- [KB-927341] – “Manage Discs” WMP feature slow to respond after resume/wake
- [KB-933872] – Default Gateway missing after wakeup
- [KB-933778] – Applications with HotStart fail to load after wakeup
- [KB-929685] – No (HD) audio upon resume/wake (possibly permanantly)
- [KB-929577] – No bluetooth on resume/wake
- [KB-929762] – Stop error 0x9F on machines with firewire (IEEE1394) upon resume/wake
- [KB-929909] – Intel 945GM Chipset PCs won’t wake/resume
- [KB-930311] – No network with stop error 0x0000007E after resume/wake
- [KB-930495] – No firewire (IEEE1394) after resume/wake
- [KB-930570] – usbhub.sys stop error 0x00000044 on resume/wake
View on [digg] [slashdot] [slashdot again]
“Deep Sleep” is configured in the BIOS, and refers to the use of S3 power-saving mode instead of the default S1 setting. In S3, you’re machine actually kills the power to everything but the memory, and uses up about as much power as a single, tiny light (like those powering up your Christmas lights). And you get to instantly turn your PC back on, with all your programs running and in a matter of seconds (for real!). At least, that’s how it’s supposed to work. ↩
As opposed to using the shutdown menu, and selecting an option from there. ↩
All systems are modern, 100% x86 ACPI-compatible systems running Windows Vista Ultimate Edition, RTM. No systems are using non-signed Vista drivers, and no buggy software is installed. ↩
We obviously didn’t mess with uxtheme.dll, didn’t patch it, nor did we try to install any of XP’s theming software… ↩
Assuming, of course, that the BIOS conforms to the basic standards already defined and used by BIOS manufacturers everywhere ↩
Talk about broken! What does Vista have to offer that isn’t
Half-Baked
Completely Broken
Stolen from somebody else
Useless
??????
Microsoft was going on and on about how Vista’s power settnings have been totally revamped, I’ve seen the screeenshots in your gallery, then you guys came and claim it sucks…
Either MS is way worse off than anyone thought or you guys are talking about somehting else…
Lol, Michael, there is such thing as grammar you know 😉
But really, I don’t see a conflict. It seems to me (took a quick look at the Vista gallery at NST really fast), that Microsoft has offered flexibility and customization and control, but they just left out stablitiy and speed, and that’s what (to me at any rate) this article is all about.
Interesting article, scary reflections. Vista won’t fail because Windows is already number 1, but let’s just say if this was Linux 2.7 we’re talking about, we could write off as a joke and say that Linux as we know it was over… But it’s Vista, so we have to be nice. :-{
Odd – I’m running Vista Enterprise on about 30 machines and none of them have S3/S4 resume issues.
They’re all bits of HP hardware though – several models of laptops & tablets, and d530, dc7100/7600/7700 business desktops. Maybe we’re just lucky…
Now I’m no Microsoft fan boy (writing this on my Mac as we speak) but bow arrogant can one person be, making such a large blanket statement.
I have the final version of Vista running on every one of my companies workstation models (A mixture of 10 different models mixed between Dell, IBM, and HP). Not one of them has this issue. Either the author is completely ignorant and has no idea how to test a operating system (like installing on more then one machine, with different hardware configs and making sure that all the drivers are correct) or the author is just trying to generate traffic by writing a inflammatory article. Either way a complete was of time.. Be smart people get your information from a trusted source, not some random yahoo who clearly has no clue.
What a load of bullshit! There are so many things wrong with what you are saying, it is hard to know where to start. There is no credibility to this story as the footnotes are vague and largely incorrect.
1. Deep Sleep is not S3. Deep Sleep is S4. Look it up for yourself. Support for any hibernation mode relies on the hardware itself to be fully compliant.
3. “All systems are modern” – List the specifications, including the manufacturers of the systems tested. List the bios settings used, drivers, installation options, etc. etc.
4. There are a number of other recovery options, and it is not a reasonable assumption to associate the .dll problem with the hibernation.
5. That is a rather large assumption. Further, there is no evidence to suggest that this is a problem at the kernel level. With so many different manufacturers of motherboards, different hardware configurations, etc. etc. etc. it is not reasonable to assume that every configuration would be supported on the new O/S. Further, which of the tested systems was badged “Vista Ready” ?
Dear Dusty,
Did you RTFA?
Sincerely,
Me
Actually, if it were Linux 2.7, “we” would write it off as being a part of the development series of the Linux kernel.
@Mike: We spoke with Microsoft, they told us the .dll with hibernation was a bug. The rest are also reported bugs. The machines don’t have a Vista-Capable sticker, but Vista installed on them, didn’t it? And they all receive above a 3.5 in System Rating. This is using the default S3 settings, except the default S1 sleep has been changed to S3. There’s a reason why “Deep Sleep” is in quotes, we’re talking about S3 – S4 is rather temperamental, and we didn’t want that to bias our results.
The drivers installed don’t matter, what does is that they’re all signed and out-of-the-box (except the WDDM drivers, and we talked about that).
Maybe Vista doesn’t support S3 on this hardware, is that any better? These are all modern machines that S3-Sleep just fine on XP, 2k3, and Linux – what more do you need?
Lol Antares, you know what he means 😉
If it’s Linux, it’s the end of the world. If it’s Vista, it’s just another small thing we’re supposed to ignore. For how long!?
@CG
“@Mike: We spoke with Microsoft, they told us the .dll with hibernation was a bug.”
Who at Microsoft did you speak to? What KB article is this documented under?
“The machines don?t have a Vista-Capable sticker, but Vista installed on them, didn?t it?”
I can get Vista to install on just about any piece of 486+ hardware, that does not mean that it is supported, or that all functionality is available.
“There?s a reason why ?Deep Sleep? is in quotes, we?re talking about S3 – S4 is rather temperamental, and we didn?t want that to bias our results. ”
Fair enough. I agree that S4 is very poorly supported, by both O/S and hardware manufacturers.
“The drivers installed don?t matter, what does is that they?re all signed and out-of-the-box (except the WDDM drivers, and we talked about that).”
Unfortunately, drivers do matter. Study after study has concluded that the majority (as high as 99.xx%, and I apologize for not having a link to this information) of BSOD scenarios are a direct result of poor drivers. The fact that they get signed, well now that is a valid point for contention. Many drivers should never have been signed, and the testing process is not as open and conclusive as many IT professional would like.
“Maybe Vista doesn?t support S3 on this hardware, is that any better? These are all modern machines that S3-Sleep just fine on XP, 2k3, and Linux – what more do you need?”
Better? No. But certainly more accurate than a blanket statement calling Windows Vista Wakeup Support hideous.
Fair point. I guess if Microsoft drivers with it’s “Ultimate” operating system that shouldn’t be there and shouldn’t be even signed in the first place, that might explain the problems we’re facing on our widely-varying hardware here. However, that’s rather worse than just saying “Vista’s wakeup-controller is borked” because it means a hell of a lot worse can go wrong.
About the hardware: I agree, that’s why I pointed out that bit about all machines scoring > 3.5 on WinSAT.
I really think you should publish the list of hardware that failed, so in the least everyone can avoid that manufacturer. While it is easy, and possibly fun, to blame Microsoft and windows for every short coming of a OS the fact of the matter is that there are many shoddy hardware implementations. For instance the wireless connectivity issues detailed http://windowsvistablog.com/blogs/windowsvista/archive/2006/12/04/how-windows-vista-rtm-build-addresses-wireless-connectivity-issues.aspx they can be nailed down to shoddy hardware, which isn’t, by a long shot Microsoft’s fault.
Of course the line saying that simply because Vista installs on some hardware it should support all the hardware is quite horrible precedent to take. Take for instance the fact that people will use a microwave to dry their clothes (not recommended), simply because you can fit a baby in one doesn’t mean that it is suitable for drying. While that is a very extreme example I think it gets the point across, there is no substitute for common sense even if the sense isn’t so common.
Hey Lyme,
I just want to point out that the Microwave doesn’t come with a list of “tested clothing brand names” to back that up – nor does it ship with “sample clothing” to try it with.
Vista does. They’re called signed drivers, and their very definition is Microsoft-approved, tried, and tested. Some come in the box, some are available on Windows Update. If Microsoft explicitly says they’re OK…
The ACPI S3 suspend mode is called Suspend to RAM (“STR”), not “deep sleep”. S4 (Hibernation) is Suspend to Disk (“STD”) and S1 is Power on Suspend (“POS”).
I can concur with the myriad of sleep/wakeup issues in RTM. I have a homebuilt system using an Asus P5W DH board (latest 1607 Bios), Pentium D 805, and an Nvidia 7900GT. With Beta 2 and RC1 Sleep worked perfect, I couldn’t be more thrilled. With both RC2 and RTM the machine goes into sleep (the power led blinks), it powers back up after moving the mouse, but the screen stays off and the hard drive led stays on solid. A reboot will bring it back to life (restoring from the hibernate file on disk), usually, although I’ve had a couple instances where it refused to boot and had to run the repair utilities.
So much for energy savings!
Thanks for the info, Ahti! I’ll see what I can do about fitting that in 🙂
@Ed34: Yep. With a Pentium D and a nVidia 7900GT, you really do need S3 suspend whenever you can, that stuff doesn’t run off of hydrogen! But with Vista, to suspend, you have to waste more time than to shutdown. And it wasn’t like that in the beta! :@
If DWM dies, try ‘net stop uxsms’ and then ‘net start uxsms’ from an elevated cmd prompt. Manually running dwm.exe won’t work because Vista launches it with said service.
Of course these features work on W2K and XP. They work because the hardware was developed with those operating systems and ships with those operating systems. No manufacturer is going to ship a computer with an operating system that isn’t compatible with the hardware.
That’s why there’s this two-month gap between Vista RTM and public availability. The hardware manufacturers need time to get these bugs worked out with their hardware. Rest assured that companies like Dell and HP have been scrambling to fix all of the little inconsistencies so that when they ship Vista, everything (including hibernate) will work perfectly.
Vista isn’t done yet. You spent a whole page groaning about a single bug in an unfinished application. RTM means Release To Manufacturing, not Release to the Public. This period between RTM and public release was added specifically to allow things like this to be fixed before the product gets to you. And your uninformed complaining is the perfect example of why they limit this release to developers only.
I’m a developer. It’s really simply: if the drivers are broke, don’t ship them with the Operating System.
The drivers that shipped with RTM in Nov. are the same that will ship with Vista (in the box) in Jan. Only OEM machines will have custom drivers, and that’s not what this article is about.
My Mac goes to sleep flawlessly. Could this be a case of the 25+ year old BIOS holding back PCs?
*cough* dropped EFI support *cough*
Of course not, why would you ever imagine that Microsoft’s refusal to leave BIOS could have anything to do with this kind of inherent system instability?
Oops! I forgot the /sarcasm tag!
Even with buggy BIOSes full of out-of-specifications ACPI instructions designed for M$ Windows, it’s harder to believe that this so expected windows version fails to do it’s job when deep slipping or hibertating!
It looks like M$ standards never lasts too long.. You will see..
It’s not sleeping. It’s calculating how much money Microsoft made off you for your poor choice of Windows and Office software.
I can verify everything stated in the article is exactly what’s happening on my Asus A8N32-SLI based system. S3 suspend works great on this system under XP. S3 worked great in the Vista beta. But Vista’s S3 using ‘out of the box’ drivers got badly broken in the RTM.
The primary question in my mind is WHERE ARE THE NVIDIA PLATFORM DRIVERS FOR VISTA RTM??!! I check for them almost daily and not only is there no sign of them, there’s almost no discussion going on anywhere on the internet about the lack of nvidia vista RTM drivers. What’s up with that??
Until I can get the basics like this and PGP support working properly, I have to relegate Vista to getting run occasionally in a VM or multiboot.
Vista IS Windows ME, The Sequel.
Only it’s worse.
It always scares me when developers confuse technical things. Like the difference between “inherit” and “inherent”.
Oh, and this light blue on light gray textarea isn’t doing much for me, either. : \
LoL..
You caught my ?slipping” and “sleeping” misuse of words. Despite that, I think that windows sucks idea still remains there.
PS: This WYSIWYG html editor sucks too! It could have syntax highlighting to prevent scary confusions.
Thanks for installing a pirated version of Windows Vista, my friends.
You know the ‘billgates release’? We made a tiny one-byte change in shell32.dll on the ‘leaked’ version, acutally we made several (to determine WHO leaked it), but the final internet-released version, it was just one change, a call to an invalid instruction by adjusting jump offset, it wouldn’t affect anything unless some special condition triggered it.
In the future, please try to avoid using pirated software ^_^
Sincerely yours,
Microsoft
*Editor’s note*
This isn’t Microsoft, and he’s kidding. At least, I hope this really isn’t true :@
Doesn’t matter to us, we have direct access to these builds from Microsoft, and it doesn’t affect the integrity of our tests
um hello0000
sleep has been broken in xp forever, sometimes it works sometimes it dont, been this way forever! sheesh…
You can add unexplained screen resolution changes to your list.
I have Vista RC2 installed on my Thinkpad T42p, and after waking deep sleep, sometimes the screen resolution spontaniously changes to 800×600 for no reason!! Arrghh!!
I thought I’d weigh in in support of the author on this one.
I’ve been running RC2 on my Asus A6Ja (Centrino Duo platform, 4.2 Windows Vista rating) notebook since it was released and I’ve had a number of problems with both sleep and hibernate. Originally, the problem was that the wireless networking wouldn’t re-acquire my access point. As of late however, I get the solid HDD light, no action scenario when coming out of sleep or hibernation.
Compared to Windows XP and Windows 2000 (both of which I’ve used for years on various platforms), Windows 2000 had the most robust hibernation and sleep of the three! I had very very few problems with hibernation under XP (on the same notebook even) but sleep didn’t seem to work properly all the time.
So, count me as another user with Vista hibernate and sleep problems, desperately looking for a cure!
I have to confirm that for some machines power management is really a problem in Vista. My main notebook (a very recent Acer) won’t sleep (it shuts the screen off but the cooler remains on at maximum speed, tha case becomes hot and just a forced powerdown will stop it). It won’t resume from hybernate. The display remains black after a screen switch off caused by power management policy. Basically nothing power-related works as expected with ALL drivers comming from Windows itself or Windows Update, a fresh install and all the default settings. XP works perfectly for me while no matter how entusiastic I may be with Vista, I cannot use it. I read this tale over and over again and understand this is a real problem. It is absolutelly the oposite of what MS states about fantastic improvements of the power management in Vista. How unfortunate.
This is a general comment to the author(s), skip if looking for Vista specifics.
Dear Mr. Guru, thank you for this heads up. Most computer ?enthusiasts?, well, they would eat there own young. A few years back in one of the ACM journals, there was an article on a first C++ programming assignment. There was a huge response on how flawed it was. Talk about over reacting. I liked this article. I thought your last article about Firefox was pretty much ^&*(hit. If anyone else gets down this far in the comments, I would take not of this and hope that my machine(s) did not have this problem and not be surprised if it did. FYI I think Mr Lopat makes a few good points, but perhaps he is being a bit too harsh. But take his points to heart. Be specific, precise and accurate. Do not generalize. You may have done a lot of work, however, with millions in the user base, odds are you won’t cover everything. For example look at the title of this article. What you can say is ” We have done xxx amount of work , with these machines, and this does not work” or something to that effect. Your mileage may vary.
12.6 seconds to boot your operating system! For the love of all that is good an holy! 12.6 seconds is just too long. I expect the 6 second boot that Gates promised. In that extra 6.6 seconds I could… uh… well… hmmm… scratch something I guess.
Drivers do matter- I have almost all the same issues after HP installs a new driver for one of my cards occasionally. I fall back to the old driver and everything is fine; sound returns, it will awaken normally, everything loads.
Get a mac!
Hey thanks for the feedback, Gary.
You have to understand the reason for the disparity this and our Firefox article. One is an article reporting the results of a study concerning an aspect of an operating system, the other is more of an editorial-type article, the sole purpose of which is to raise concern regarding something tangible. We didn’t anyone to go You know what, they’re right, because we were expressing our fear for something that looked inevitable, and we wanted to say You know what, be careful, this might not turn out so good. Hopefully the devs at Mozilla took note.
Anyway, NeoSmart Technologies is nothing without feedback from the readers and contributions from the members, glad to hear from you 🙂
i am just a random yahoo with no clue, but one thing i know:
if it is from MS then it is destined to suck big time in some or a lot of aspects.
and that’s final!
i have no hope that MS will ever produce something truly good.
Soyo SY-P4I865PE Plus DRAGON 2 motherboard
Intel Pentium 4 3.20 GHZ HT
2x1024MB pc3200 PNY RAM
Windows Vista Home Premium
PNY GeForce 7800 GS 8x AGP, 256MB (97.46 ForceWare)
PCI Creative X-Fi XtremeMusic
LG Flatron L1920P lcd monitor
My computer shuts down when going into sleep mode, but I get no video when resuming along with a single long bios beep. I am unable to do anything but hit the reset button. Vista then shows resuming but screen goes black with a flashing cursor and some junk on the screen. I then have to hit reset again and I am able to boot properly. Any ideas if this has to do with the nvidia drivers or some bios problem? Not sure what a single long bios post beep means. I have Pheonix Award Bios. Let me know please if anyone has any ideas. In the meantime I have disable all sleep options. Thank you!
mhajii210
February 14th, 2007 – Using the RTM version of Vista on my Gateway MP8708 laptop, and STILL no relief for the Sleep/Nightmare problem. I experience several symptoms: GUI freeze on resume though mouse still moves, or no network connectivity with the LAN LED solid but the Network Manager indicating no connection and then a BSOD after 4-5 minutes.
I strongly suspect that the problem is in my LAN driver, but Gateway says it’s compatible. I’ve installed all of the other Vista driver updates from Gateway. Is everyone still having this problem, or is the public version different from the “RTM” version that I got from MSDN in December? All of the sources are quiet regarding this problem, and I’m starting to wonder if it was fixed. Ack.
Sorry Scott, it was never fixed.
You can check if it is your LAN driver easily by disabling the onboard network crad from the BIOS, booting into Vista, and seeing if that fixes your problem.
Actually, you can even disable it from the device manager, it should have the same effect.
Good luck with it, whichever way you choose.
A few days ago, I bought a Velocity Micro E6400 machine at Bestbuy for about $1,400 with Nvidia 7600GS I think and it either did not sleep or hibernate or if it did, it would not come out. I called Velocity tech support and after two hours, he told me it was probably a bad motherboard. I returned it and bought the next Veloicyt Micro model up at Bestbuy for about $2,000 but had the so called “Geek Squad” check this new model before I left the store. They had the same problem with this model as well. So I returned it. I doubt any of the Velocitiy Micro computers will do sleep mode or the power saving mode. It was very disappointing.
My trackpad will not wakeup from sleep or hibernate in Vista (full version). I’m using bootcamp on a Macbook Pro. Everything else works, but the mouse doesn’t work at all when I sleep or hibernate.
Anyone have a way to reinitialize the mouse upon wakeup?
–Neal
Neal, you either have to delete and reinstall the trackpad from Device Manager, or just reboot. It sucks, I know.
Computer Guru– thanks for the heads-up. What a time consuming fix… especially since I have to do this all without a mouse, takes forever with the keyboard.
I think there was a solution out there… something about reinitializing the mouse each time it wakes-up. But I forgot the blog and can’t find it now.
Neal :appl:
Well Neal, if you ever do find that blog again, do post back, ‘k?
Thanks! 🙂
Most Odd…
Vista Ultimate on a Presario 1516 4 years old Laptop with P4 2GH, 512MB and radeon IGP 340 64Mb shared memory. I install in it to have a first look before use Vista on my Bussines.
All working right, suspend and hibernate if I use the desktop button, but if I suspend the system by closing the lid the screen don’t resume.
Any sense on this?
Continued….
My Vista is not Rtm but final version bought at a store.
My Conclusion: It’s not a driver problem, nor a Bios problem, it is just some kind of bug.
Well, RTM means final version.
Anyway, I think you need to install the latest ATi Drivers to take care of your problem:
http://neosmart.net/blog/2007/final-ati-drivers-for-windows-vista-rtm/