More Z97X-UD3H-BK Sleep issues

Associate
Joined
7 Mar 2005
Posts
1,650
Location
Wiltshire
Due to some Corsair RAM being RMA'd I thought I'd try some slightly faster Kingston HyperX Fury 1866MMHz on my both PCs (both as per spec) in the meantime.

The RAM itself is fine, but frequently and randomly the machines will blue screen but only when resuming from sleep. This issue never occurred with my previous Corsair 1600MHz memory. The BSOD error is always system_service_exception, but Win debug analysis shows that each crash is caused by a different driver, which points back to the RAM/IMC in my view.

Here's what I've tried so far with no success:

- RAM has been tested with a minimum of 8 passes in MemTest86+
- Enabled/disabled "C" states
- Reinstalled Windows
- Updated Chipset drivers
- BIOS updated from F6 to F7e
- Increased IMC voltage
- Enabled Enhance Memory stability in bios
- Disabled fast boot
- Removed CPU overclock
- As the symptoms occur on both PCs this rules out a faulty component

I have absolutely no other issues at all and every test I throw at the computers is fine, for whatever reason it just doesn't like resuming from sleep with this memory running at it's default 1866 settings. My theory is that there's a short period during the resume process when insufficient voltage is applied to either the memory or the IMC, and that at all other times the supply is fine.

To be honest I think that the answer is just to stop using sleep, but as it worked with my previous memory - and on all previous PCs I've had - I don't see why I should have to do so.

I'd really appreciate any advice or further suggestions.
 
Thanks wazza, yes it's the latest F7e beta bios, the issue remains the same with F6 too.

It's 2 x 8Gb sticks, what would you recommend as an IMC bump, +0.15v to System Agent, and Analog/Digi I/O?
 
I don't believe it's possible to tell on Gigabyte Z97 boards. The BIOS only shows the value as an offset, not what it's from.

I've also been unable to find anything that can monitor it, none of HWINFO HWmonitor nor even Gigabyte's own tools.

The Gigabyte o/c guides I've found recommend setting these values to between +0.15v and +0.3v but these all imply that's for much faster RAM and that 1866 should need no IMC bump - which is sort of backed up by mine being stable in all other tasks. I guess it's possible an IMC bump may help the memory get over the initial hump when resuming.
 
try it and see,it still might be a bios bug

but touch more imc usually helps

unplug any usb peripherals aswell apart from mouse/kb see if that's affecting it
 
Cheers wazza - I am trying the IMC voltage bump now and will see how it goes. I'll move on to USB after that, I do have an Xbox 360 controller and external USB3 drive plugged in that can be removed to leave just mouse and keyboard.
 
Adding IMC voltage has not resolved the issue, nor has removing superfluous USB devices.

I think the next few things to try - following on from the list of things I've already tried in the OP - are:

1 - Run the RAM at 1600Mhz
2 - Increase command rate from 1T to 2T
3 - Increase memory voltage from default 1.5v

Anything else I should be considering?
 
Neither down clocking to 1600MHz nor a command rate of 2T has resolved the issue. I'm now trying walking up the voltages on one machine, and have put different memory in the second machine which is now working perfectly.

As Kingston's memory selector recommended this RAM for my motherboards, I'll contact them next for advice.
 
I contacted Kingston last week and since some initial back and forth for further info, I've had radio silence from them.

Since then I've tried the following all without success:

Downclocked RAM to 1600MHz
2T instead of 1T command rate
Enabled/Disabled memory training
Disabled/enabled Intel Dynamic Storage accelerator
RAM in different slots
Increased CPU vcore
Increased DDR voltage
Set LLC to Turbo
Enabed/Disabled all power saving options in BIOS
Disabled VT-d and virtualisation
Moved SSDs to different SATA ports
Disabled AHCI link Power Management
Manually set RAM timings
Borrowed a friend's Vengeance Pro 1866 C9 memory - same result in both boards.

As Wazza mentioned earlier I believe this must be a BIOS bug that is only triggered when resuming from sleep with memory clocked higher than 1600Mhz. I'll approach Gigabyte next for their advice, but if anyone has any suggestions in the meantime I'd still appreciate it.
 
Gigabyte have not been much help as yet, they've had me remove the GPUs, replace the hard disk and further update chipset drivers all without success.

Does anyone have memory running at > 1600MHz with this board and happen to use sleep?
 
Only because I leave it on when I eat dinner. But you make a valid point that it's only a difference of 10-15 seconds.
 
Why shame? It's not about startup speed as such, it's just very convenient to be able to pickup from where you left off.
 
Bickaxe - any chance you can confirm which Intel RST/AHCI driver you are using? I'd previously tried 13.2.4 and 13.6.0 (latest) with no luck. I am now using 13.1.0 as per Gigabyte's driver page and I think it may have resolved the issue, but will need a couple more days of stability to be sure.
 
Last edited:
Spoke too soon, wasn't that either. I've now tried a list of literally 30 different things, and it has to be some form of hardware incompatibility between the motherboard and something else.

The only thing that's changed since my previously perfect Sandy builds are the boards and processors.

Edit: I should have added that occasionally the BSODs occur after sleep with the 1600MHz Vengeance also, it just seems that 1866+ memory exacerbates the problem.
 
Last edited:
Back
Top Bottom