GA-H87-D3H: Blue screen STOP errors, possibly related to Windows 7 Sleep

[Steve_Ell]

Hi,

I'm having problems with blue STOP screen errors, which as far as I can tell only happen after the machine has been in sleep.

There are a variety of STOP errors, and these have been happening since the machine was built in March this year, although they seem to be becoming more frequent.

I have recently updated all motherboard drivers from the Gigabyte website, and updated the BIOS to the latest available – to F9 from F6.

Memtest 86 was run overnight when the system was built, and I ran it again for 8 hours/10 full passes on 24/10/2014, no errors were found.  The memory was chosen from the Gigabyte "Compatibility list".

I've seen another post on this forum where the suggestion was to underclock the memory or disable C-states in the BIOS, although no specific mention of sleep was made:

http://forum.giga-byte.co.uk/index.php/topic,13468.0.html

The poster says they underclocked the memory but do not say if they altered the C-state settings.

Does anyone know if this is common to Haswell systems, or is an issue specific to this board?

Any help would be greatly appreciated!

Thanks.


PS.  I've done quite a lot of troubleshooting at the Windows level, following advice on the Microsoft forum, so you can read in more detail what I've tried already:

http://answers.microsoft.com/en-us/windows/forum/windows_7-system/blue-screen-stop-errors-possibly-related-to/7324428e-4621-46fd-b119-9d5bb3b03677?page=1


System:  Windows 7 64-bit, Intel Core i7 4770S, Gigabyte GA-H87-D3H motherboard revision 1.0, Crucial Ballistix Sport 2 x 8GB RAM, Pioneer BDC-207 Blu-Ray/DVD drive, Samsung EVO 840 120GB SSD (Windows C: drive), Samsung EVO 250 GB SSD, Samsung F3 1TB HDD, WD 800AAJS HDD, Seasonic G-550 PSU.



List of STOP errors:
-------------------

10 Apr 2014 - Page fault in non-paged area, 50 - in ntfs.sys


20th May - Kernel Stack in Page Error, 77


25th May - Stop error 24 in ntfs.sys

-suggests chkdsk /f ; also "dump size too small"

- chkdsk fixes errors: free space marked as allocated in (1) bit map (2) MFT volume


1st June - Driver Power State Failure - after installing Automap 4.8 (Novation keyboard software - uninstalled again!)


16 Jun - Memory Management, 1A - while playing mp3 in VLC


21 Aug -  Memory Management, 1A


15 Sep - Stop error 24


5 Oct -   Memory Management, 1A - while playing mp3 in VLC

- ran chkdsk afterwards, fixed errors


7 Oct - Memory Management, 1A



---- drivers updated -----



13 Oct - bad_pool_caller, C2 - just out of sleep, in Firefox, going to first new page


15 Oct - Memory Management, 1A - 13:50 , less than 30 seconds out of sleep, had checked Windows Updates and pressed "Install".


--- BIOS update ----


19 Oct - memory management, 1A - 15:00 .  Had been using sleep (ie not shut down) for 48 hrs since BIOS update.  Out of sleep 5-10 mins, writing email in Thunderbird.


20 Oct - Memory management, 1A - 14:20.  Just out of sleep - less than 2 mins?  Was opening VM in VirtualBox. 


21 Oct - Page fault in nonpaged area, 50 - 14:50. Firefox plugin container crashed coming out of sleep, was restarting when bluescreened. Had been shut down overnight.  No minidump.


---- enabled Driver Verifier


22 Oct - STOP error 24, Ntfs.sys - 18:20, was restarting after coming out of sleep.  No minidump.

27 Oct - Special Pool Detected Memory Corruption, C1 - just out of sleep, reading an email.

31 Oct - Memory Management, 1A - about 4 minutes out of sleep, using Firefox.

[Steve_Ell]

P.S.  The title of this thread should begin "GA-H87-D3H", sorry about that.  I don't know how I managed to delete the first bit, but I can't find a way to edit it now.  Perhaps a moderator could correct this? Thanks.

[Steve_Ell]

Update:  I've tried underclocking the memory to 1066 Mhz, as did the poster linked to above.  This hasn't resolved my blue screen errors.

I'll report back if I find a solution!

[autotech]

Try disabling sleep state by telling windows to always keep hard drive on.  High performance then choose when computer goes to sleep. Just turn off monitor after a period of time and see if this makes a difference before you disable them in bios.

[dmdilks]

Have you tried different memory. What is the speed of the and how many sticks? I have seen where memory just doesn't run on some boards. I have had that do that a lot. Some of the that series boards did like 4 sticks of memory.

You running a video card or through the CPU. The only reason about the video card is that you are only running a 550w PSU. If you are running with no video card you should be alright.

[Steve_Ell]

Try disabling sleep state by telling windows to always keep hard drive on.  High performance then choose when computer goes to sleep. Just turn off monitor after a period of time and see if this makes a difference before you disable them in bios.

autotech,

thanks for the suggestion!  So, you mean that that's a good way to test the c-states separate from sleep mode, right? 

 I'm currently running the machine 24/7, with sleep set to "never" in Power Management.  It's on "balanced" at the moment, so that will still be using the c-states?  I thought I should do this to be sure the system is stable when it doesn't go into sleep mode.  If it's still unstable there may be some more fundamental issue.

[Steve_Ell]

Have you tried different memory. What is the speed of the and how many sticks? I have seen where memory just doesn't run on some boards. I have had that do that a lot. Some of the that series boards did like 4 sticks of memory.

You running a video card or through the CPU. The only reason about the video card is that you are only running a 550w PSU. If you are running with no video card you should be alright.

dmdilks,

No, I haven't tried different memory yet. I have two 8GB sticks - I did check them against the Gigabyte compatibility list, so they're tested for this particular board – I'll add the details to my signature for reference.

Yes, I'm using the integrated graphics, so the PSU should be more than adequate.  (I used a PSU calculator when putting the system together).

Thanks for the input!

[dmdilks]

Yes, I'm using the integrated graphics, so the PSU should be more than adequate.  (I used a PSU calculator when putting the system together).

Thanks for the input!

I just ask that to make sure that people wouldn't say that was the problem. Even with a video card you should have enough amps to run one.
You could go in and look at "change advanced power settings" and look in there. It could be the hard that set to go to sleep too. You have to play around with some of the settings in there.

[autotech]

If you have hard drive set to never sleep then yes it will tell you if there is a problem with sleep states. If it doesn't blue screen now with hard drive set as always on then it narrows it down to the sleep states.

I have seen c3 cause some problems with different PSU's. If it goes ok thru windows then you can try just disabling c3 sleep state and see if that corrects it.

[Steve_Ell]

Thank you both for your input!

It was definitely worth checking the PSU was adequate.  It's also on Seasonic's Haswell compatible list, so should be okay with the CPU power saving stuff.

I checked the hard drive sleep, and it is currently disabled (unless the whole PC sleeps, obviously).



Update:

So, since Dec 8th, I've been running the PC 24 hours a day, without using sleep mode, and there have been no BSOD or system crashes!  (Firefox did go a bit funny one time, but I'm not blaming that on the hardware!)

I've restarted a few times for other reasons, but the machine has been on for periods of 36 hours solid with no sign of instability.

So, I'm pretty sure my crashes and BSODs are to do with sleep mode.

This is quite different from what I've been reading about people with similar Z87/H87 boards, who have found that underclocking the memory or disabling the C-states has helped their problems.  Their systems have been unstable generally, with no mention of sleep mode, for example:

http://forum.corsair.com/v3/showthread.php?t=119923

Next, I'm going to enable hard drive sleep, while still leaving the PC on all the time, and see if that triggers the problems.  I'll report back!

[Steve_Ell]

Update:

Since Dec 20th, I've had hard drive sleep enabled – no crashes or BSOD.  So I'm pretty sure now that hard drive sleep isn't a problem.

I did try using sleep mode again on Dec 26th, to see if the previous problems reoccurred – immediately upon resuming from sleep mode, Windows Explorer crashed and the PC needed a restart.


So, my questions are:

- can I now assume the memory is ok?  Or should I still try different memory?

- can I now assume the c-states are ok, as they're being used when the PC is idle?  It's in "Balanced" mode in Power Management, and according to CPU-Z it clocks down to 800MHz.

Thanks!

[autotech]

I would say memory is fine.  Sometimes sleep state C3 causes problems.

[Steve_Ell]

I would say memory is fine.  Sometimes sleep state C3 causes problems.

Yes, I think the memory is fine too (hurrah!)

And given that I only get BSODs and crashes after using sleep mode, is there any point disabling c3 (etc) in the BIOS, as I wouldn't then be able to use sleep mode anyway?

At this point, I'm quite happy leaving the PC on all day, as it only uses 20W or so when idle.  (I turn the monitor off when I'm away from it).

[Steve_Ell]

I would say memory is fine.  Sometimes sleep state C3 causes problems.

Update:

I tried disabling C3 and C6/C7 CPU states in the BIOS - still got blue screen on resuming from sleep.

At this point, I'm happy to declare the hardware is probably ok!

Thanks to everyone here for their help.

There may yet be a Windows solution, I'll take the issue back to the Windows forums (link at the top of this thread), but I'm just happy that the system is stable in normal use.

[Tiger]

BLS8G3D1609DS1S00 is a single DRAM. It is not a matched pair. Using two is not guaranteed for compatibility.

If you bought a matched pair, BLS2KIT8G3D1609DS1S00, you would have a kit guaranteed to work together.

Using a single kit is recommended. Mixing two singles may result in the inability of the memory to operate at rated specifications. I believe someone already suggested running the memory at a lower clock speed which is a work around for running mixed memory.