Computer not resuming correctly

[mikeymikec]

Spec:

Intel Core i5-4690k
ASUS Z97 PRO GAMER (latest BIOS)
8GB DDR3-1600 dual-channel
ASUS GeForce GTX 750Ti
Pioneer DVDRW
LiteON DVDRW
Corsair VX450W
Samsung 840 PRO 256GB
WD Black 1TB
Win7 64 SP1 (up-to-date)

The CPU and motherboard are new, everything else was re-used when I did a recent platform upgrade. The PSU is the oldest part at 5 years old.

The only problem I'm experiencing is with coming out of sleep mode. Maybe one in eight times, it will do one of the following:

1 - PC power LED comes on, fans spin up, but monitor doesn't show anything on screen and caps/num lock lights non-responsive. After using the reset button, the system will come on as normal and complain about an unexpected shutdown. Sometimes 0x00000116 is recorded.

2 - Yesterday it came out of sleep mode fine except for a small black box on the welcome/password screen, and YoutTube clips were not rendering normally (the kind of weird repaint effect when the codec is catching up with the feed, but a permanent effect). Rebooting fixed it.

3 - Today it did scenario 1 without recording a BSOD.

After the previous BSOD (and before scenario 2 occurred) a couple of days ago I updated the graphics driver to the latest version. This morning (before scenario 3) I downgraded the graphics driver down to 347.88 because another user on the nvidia forum reported similar issues and that fixed it for them.

I'm pondering buying a new PSU but considering that the PSU handles everything else I throw at it, it seems odd that this is the only symptom of its failure.

 

[redzo]

This may be related to the infamous haswell c7 sleep state.
http://www.corsair.com/en-us/blog/2013/may/haswell-compatibility-with-corsair-power-supplies
I can disable C7 CPU power state on my gigabyte mobo. Asus should have exposed this setting on your mobo also. I would suggest experimenting with this setting(disable it).
I don't think that it is worth purchasing another PSU if this ends up being your issue.

 

[JeffMD]

I am not inclined to point at the psu either. I am not familiar with the haswell bug, but I believe a driver bug can also cause this. A driver being told to come out of sleep may be hanging the system. It could also be the sign of a dying motherboard. I had an evga nForce 680i SLI that started power cycling over and over when I attempted to wake it out of sleep.

 

[mikeymikec]

There are two settings in the BIOS under CPU advanced power management configuration that I think are relevant and I've changed:

"CPU C7 Report" - I've set this to disabled
"Package C-states support" - I've set this to C6

The board is only a month old so I hope it's not failing

Thanks for the info, I'll give it a try.

 

[Steltek]

Might not be a bad idea to run an SFC /scannow session from an elevated command prompt to ensure you don't have any corrupted system files.

 

[mikeymikec]

Might not be a bad idea to run an SFC /scannow session from an elevated command prompt to ensure you don't have any corrupted system files.

No corruptions found.

 

[mikeymikec]

Since the C7 change, the only incident I've experienced so far is that when it came out of sleep mode this morning, the display took longer (about 10 seconds longer) to initialise and a display driver reset was noted in the log (along with a 'video hardware error' in Reliability Monitor).

One other thing I noticed is that before resuming, I noticed that the num lock light was off, which I think I've seen on every occasion where it's had trouble resuming.

This 'video hardware error' business is a symptom I should have mentioned before: It's not in common with every problematic resume scenario (maybe 50% of them), but it is happening.

- edit - I personally think the 'video hardware error' business and problems coming out of standby are connected (partly because the BSODs point at graphics mostly AFAIK). After a bit of reading up I've made two changes for the time being, one was to disable PCIE link state power management (Power setting in Windows) and also to set the BIOS from Auto to PEG (the setting that tells it where to find the primary graphics device).

I've considered that it may be a graphics hardware problem, but it seems a bit implausible that a graphics card is only faulty in one way, ie. only coming out of standby. IMO the remaining possibilities are power and to do with the BIOS.

 

[mikeymikec]

I'm unsure how to proceed from this point.

Since my last post, there have been only two occurrences of undesirable behaviour:

19/10/2015: The computer resumed from S3 successfully, but hung about two minutes later while I was browsing FB in Firefox. (NB: FF is extremely reliable on my computer, I might get an FF crash once (or something similarly undesirable) every six months, maybe less)

Approx 30/10/2015: The computer failed to resume from S3 correctly, but the symptoms were different. The power LED and fans came on, but no keyboard response or display. I pressed the reset button and the PC resumed from the backup hibernate file that is generated for the hybrid sleep feature. Windows was unaware of anything unusual occurring, nothing dodgy in the logs, and so nothing on the following graph.

Here's how the reliability graph looks:

Things have apparently improved (before the C7 change I was getting resume problems every other day, noted on the graph where there are two X's in the same column), but I don't think the problem has been completely nailed yet. The last X on the right was a video hardware error (26/10) that didn't present any symptoms that I noticed. The one before it was the 19/10 incident.

Just before the most recent event I was on the verge of doing things like reinstating 4GB RAM I took out before I started this thread (attempting to troubleshoot this problem). I guess I'll leave things as-is for a bit longer and see where things go from here. Personally I suspect that the PSU is starting to struggle, but if so it would be the first of its model that I've sold to die to my knowledge. I guess it's probably had to work harder than most of the other VX450W's I've sold though.

Perhaps it's wishful thinking but given that the most recent incident was different in its symptoms, perhaps it's not the same problem as I was originally experiencing. Dunno.

 

[mikeymikec]

07/11 - failed to come out of S3 (standard symptom, power LED and fans on, no keyboard or display response).

I responded by swapping out the PSU for a spare (CX430) I've recently received as a result of an unrelated RMA. I also reinstated the original 4GB RAM alongside the 8GB RAM I've been using.

08/11 - The 'resume successful then hang a couple of minutes later' problem reoccurred. I responded by removing the 8GB pair and just running with the 4GB pair.

I also found a setting in the BIOS labelled 'VRM spread spectrum' with a description that it is supposed to increase system stability. I thought I'd give it a try, so that's enabled now.

I'm thinking that most of the symptoms pointed to the graphics card (with the exception of the C7 suggestion, I think that was relevant and has certainly reduced the frequency of resume issues). I have an effective enough spare in case it comes to that, but I think I'm going to have a wonderful time trying to RMA it, if it is the cause of the problem.