PC sometimes won't wake up from sleep mode, red + yellow lights on MB

H

hupimies2151a02d0

New member
Joined
Sep 22, 2024
Messages
6
Board: X670E Tomahawk WIFI
BIOS: 7E12v1E
GPU: AsRock RX 7900 XT Phantom Gaming OC
CPU: Ryzen 7 7800X3D
Cooler: Noctua NH-D15
RAM: 2x16GB Corsair Vengeance 6000MHz
SSD: Samsung 2TB 990 PRO
PSU: Corsair 850W RM850x
SYS: Windows 11 Home x64 Ver. 10.0.22631

About 7/10 times the pc wakes up normally from sleep mode, during which red and yellow light on motherboard (CPU, RAM) flash. 3/10 times the lights light up for about 30 seconds, after which they turn off about the same time. After this only black screen, USB devices get power (leds on mouse light up, not on keyboard) but no other connection. Have waited for over 30 minutes, no change. Always gets fixed by forced shutoff via holding down the power button, after this boots up to saved state (previous applications open etc.) No problems on regular startup.

Worked normally for about a year, problem started a few months ago suddenly. Didn't disconnect or open up the PC / download any software / do anything else notable which could have caused system change. All temps during benchmarks normal and no problems during running. PC has never crashed etc.

List of things that have not fixed the problem:

Updated BIOS
Reset BIOS settings through BIOS menu, haven't cleared CMOS with jumping
Updated GPU drivers
Updated Windows from 10 to 11
Reset Windows power settings through troubleshooting
Turned off fast startup
Unplugged and reconnected PSU (also removed and reconnected GPU, didn't try to boot without it)
RAM are in slots A2 and B2 per the manual, as they have always been

Not a huge problem, but very annoying since I primarily keep PC in sleep mode. Also have to think that constant forced shutoff doesn't do the PC any favors. Any thoughts on possible fixes? Could clearing CMOS have any other effect than resetting through the BIOS menu does?
 
Sort by date Sort by votes
Michele P said:
I forgot to say that everybody should also disable the windows fast start option (no reason to keep that activated unless you have the OS installed on a mechanical drive, and we live in 2025)
Otherwise this sleep bug will affect the system even more often.
And linux user should stay onlts kernel versions, preferably 6.6 instead of 6.11/12

The problem for me resides in Agesa, in particular the way it manages:
- memory context restore
- power down
- expo
- cpu power states and eventual cpu protections added for safety measures.

Many of those things were not properly functional in the early stages of agesa, this is also why the oldest bioses do not exhibit the problem.
The specific sleep patch added in amd chipset drivers was maybe targeted to older agesa versions… while the new ones are broken again.

Release after release, we can now have reliable fast memory trainings and a cpu that operates in a safer way (limited soc voltages for example, but also other less publicly known hardware safety instructions and security mitigations)

All good things, but not without side effects apparently, like this sleep bug we are talking about (and maybe the difficulties to restart when expo is enabled)

agesa must work in armony with the amd chipset drivers, given there’s nothing else interfering. Yes, other devices, relative drivers, and even microsoft updates, can easily break such a fragile feature.

Said that it’s obvious the bug will affect almost every manufacturer, maybe some more than others. But it’s also clear that diagnosing a problem with so many variables it’s not easy. Those MSI support answers are understandable. On my x670 ace I don’t use sleep regularly, but I can’t replicate the issue either.

Idk if opening tickets is a good idea.
Maybe mass report the bug on AMD subreddit?
Replying “me too” to every other report you find in different places?
Let the issue have more public visibility, outside the specific vendor channels
Click to expand...
I would like to add that Fast Startup reduces the size of the hiberfil.sys file when using Shut Down.

The article states:
Hibernation files are used for hybrid sleep, fast startup, and standard hibernation. There are two types, differentiated by size, a full and reduced size hibernation file. Only fast startup can use a reduced hibernation file.

See https://learn.microsoft.com/en-us/windows/win32/power/system-power-states for details.

Fast Startup is only implemented when a user shuts down Windows. It will log-off the user(s), closing any programs they have running. It will store the running state of the OS in the hiberfil.sys file. This allows a memory image of a previously loaded OS to be reinstated, whereas booting from scratch requires loading of the OS piece by piece and with much slower mechanical storage each seek/read operation can take a relatively long time. Fast Startup take much less time as it reads from a single file that is stored in contiguous blocks that can be retrieved rapidly.
 
Last edited:
Upvote 0 Downvote
As a note for those who may come across this forum and have issues with S3 Sleep.

There are many topics regarding S3 suspend behaviour that may simply be a configuration issue.

1. S3 fails to resume using keyboard/mouse. This is likely due to USB wake-up events being disabled in the BIOS settings.

2. Failure to stay in S3 sleep, waking immediately or unexpectedly.
This can be caused by Wake-on-LAN being enabled in the BIOS settings or a task in the TaskScheduler.

The following PowerShell command can be used to list all tasks that are allowed to wake the computer:

Get-ScheduledTask | where {$_.Settings.WakeToRun}

These tasks can then be disabled completely or have the option disabled (look in the Conditions tab) using the TaskScheduler.

As a personal note, if you have cats that are drawn to resting on the keyboard (kittens seem to love being the centre of attention) you may want to adjust the Windows settings for the keyboard/mouse in Device Manager so that only the mouse (ironically) can wake the computer. Note that there can be several keyboard and mouse devices listed as a keyboard/mouse may have dual functionality. The best method is to disable the 'allow this device wake the computer' for all but one and proceed to eliminate those that cause the PC to wake.
 
Last edited:
Upvote 0 Downvote
and even though fast startup is a different thing from S3 sleep, it's always better to have windows booted in a clean state
not "resumed", because it's what fast start up does, it resumes windows every time to turn off/on the system (you need to "force" a restart to have a clean boot)

it's kind of anachronistic to use a feature born in the era of Windows 7/8 (?) when mechanical drives were still a thing as OS drives
in 2025 everybody installs the OS on an nvme, or ssd at least, this feature adds little to no improvements at the cost of high unpredictability

You boot windows, that is already resumed from the beginning because fast start up, then put the OS in sleep mode and hope that everything works smooth... maybe it's asking a bit too much

The main problem it's not system waking up unexpectedly, but that doesn't wake up properly and you need to power cycle the pc
Many reasons can cause to fail a wake up from S3 sleep
We can speculate about AGESA/AMD being the main offender (95% it's their fault), but there are many other causes, that's why it's a very difficult bug to diagnose
- a particular device, usually connected to usb, just disconnecting the device may solve the problem
- a poor written or bugged driver, this happens a lot, typically coming from gpu to lan drivers (including, nvidia and amd gpu drivers, in my long time experience I've seen many specific gpu drivers causing sleep to stop working)
- windows updates
- running softwares
- and who knows what else, from cheap PSUs to your astrological sign (I've been always very unlucky with this feature, probably for a bad ascendant XD)
 
Last edited:
Upvote 0 Downvote
@Michele P Your are correct that any savvy user will be using NVME or SSD as a Boot drive. However, mechanical drives are still the most economical in terms of price per GB. I have not stated that I use a mechanical drive for booting.

Fast Startup only functions after a Shut Down ( Fast Startup invokes the S4 power state but logs-off the users) event which is not identical to the S4 Hibernation or S3 Sleep states.

You mention "unpredictability". The majority of PC users may assume that a PC is incapable of making data errors and (thanks to technology) for the most part this true.

However, Enterprise solutions use ECC RAM and specific storage solutions to detect and correct errors as there is no inherent safety in the process of copying data from one place to another.

I could not afford 16TB of NVME storage as a backup media, no matter that it would save much time.

Here is a portion of the article: https://learn.microsoft.com/en-us/windows/win32/power/system-power-states
When using fast startup, the system appears to the user as though a full shutdown (S5) has occurred, even though the system has actually gone through S4. This includes how the system responds to device wake alarms. Fast startup logs off user sessions, but the contents of kernel (session 0) are written to hard disk. This enables faster boot.

You state that I have missed some reasons that cause a PC to fail when resuming. My post has stated that a specific issue is present that occurs long before those elements even factor into the problem I and others have. I have added additional information that some might find useful in diagnosing an issue that prevents sleep resuming when using the keyboard/mouse and something that may prevent it staying in the sleep state.

So, a debug code of 11 prevails any process that that would eventually lead to the processing of the S3 Boot Script. If it stops here then the mission is no go.

What you are referring to is a much later failure of the S3 resume process that would cause Windows to fail resuming. This should be indicated with either an automatic reboot, BSoD or a debug display code of E8-EF.

Also, note that the S3 resume process actually utilises the previous (successful) boot events to initialise the reinstatement of the system state.

Also, MSI have already confirmed that they have confirmed the issue in testing.
 
Last edited:
Upvote 0 Downvote
HoovaDevil said:
I have not stated that I use a mechanical drive for booting. You mention "unpredictability". Do you suggest that we should all be building systems with only NVME and ECC RAM?
There is no inherent safety in the process of reading/writing data. This is a reason for enterprise devices using ECC RAM to detect errors.
Click to expand...

no, unpredictability is not referred to mechanical drives

A windows booted with fast start up enabled is more unpredictable than a fresh booted one, especially if you already know that the system exhibits problems in a "resumed" state (this is the topic of the thread)
Even if the resume is a little different, why introducing additional failure points?


You are free to use whatever you want but I'm pretty sure 99.9% of people with windows 10/11 are not using a mechanical drive anymore as their main OS boot drive, you don't need to speed too much for a small recent ssd/nvme.
There's difference between storage and OS drive, I still use mechanical drives in my small home NAS (even my little raspberry pi has its OS installed on a nvme drive) , but I think we can agree that is stupid to install a recent windows on a mechanical disk.
Fast start up was introduced in that period when people transitioned from mechanical to the first slow SSDs for BOOT drives
This is why I suggest to just disable an old feature meant specifically for old windows on old disks

You are adding a resume process (windows fast start up) just before another "different" and bugged resume (the S3 sleep/wake up issue of the topic): I see the first one redundant, as well as possible causes of further unpredictability

EDIT: I'm pretty sure MSI has been able to replicate the problem, cause it's very common on very different AM5 boards... whether they can fix something or not it's another matter of discussion, because they can't clearly touch AGESA.
 
Last edited:
Upvote 0 Downvote
@Michele P I think you have missed the point.
1. Fast Startup is only active when choosing Shut Down.
2. We use NVME/SSD drives because they are faster, not because rotating magnetic storage is less reliable.
3. You say "I think we can agree that is stupid to install a recent windows on a mechanical disk" but it is exactly that technology that allowed decades of computing. The technology has improved and so we now have NAND flash devices that exceed the older tech at the cost of longevity. You might be surprised with the latest HDD technology.
4. Storage is storage. It does not matter which physical medium you use. Again, the world relied on rotating magnetic storage devices for decades without major issues.
5. Progress is an funny thing. We used NVME drives to decrease BOOT times only to find that DDR5 'training' took longer.
6. You state "because they can't clearly touch AGESA", which implies that any issue that they find will be ignored by AMD. You suggest that board partners cannot relay issues to AMD with any hope of resolution. I would counter that with this: Any underlying issue with a BIOS could affect the future stability of any variation of the board in the future. It is not advantageous to avoid fixing a bug in the hope that it will not bite you later.
 
Last edited:
Upvote 0 Downvote
are you trolling or what?

HoovaDevil said:
1. Fast Startup is only active when choosing Shut Down.
Click to expand...

Michele P said:
because it's what fast start up does, it resumes windows every time to turn off/on the system (you need to "force" a restart to have a clean boot)
Click to expand...

I know exactly what it is, and that's different from the S3

The point of the thread is:

the system fails to wake up from sleep

How can I trust the fast start up feature if I already know the pc has troubles resuming properly from a power saving state?

Let's say you use fast start up option (windows does that by default)

- turn off
- turn on (it's not a clean boot, system is in a resumed state), FIRST RESUME
- system "apparently" functional, you don't restart
- put system to sleep
- wake up the system, SECOND RESUME... bugged, power cycle to recover

If I already know the second resume from sleep S3 is problematic, even if it is different, I find it hard to trust the first resume.
Who tells me that the system wasn't already bugged since the first resume? like added latency or any other more or less hidden problem.
If I'm trying to diagnose the problem, I try to eliminate the first resume as it could be the first point of failure (the unpredictability I was talking before)
Probably it won't fix anything, but it's something everybody should try in this situation, and maybe keep it disabled isn't a bad idea after all.

Then, I don't care if you still use magnetic tape as a boot drive, an intelligent person has to evaluate PRO and CONS of a feature. Do I need it or not?
I'm pretty sure with a recent SSD/NVME everybody can get the same boot times when the feature is turned off.
So why keep that option ON and have a "dirty" OS every time the system is turned on? I prefer to have a clean OS environment every time I press the power on button instead.
I can't deny the feature can be useful on slower disks, but nowadays it's a rarity.
 
Last edited:
Upvote 0 Downvote
As I have received confirmation of the issue from MSI, I will update here if/when there is anything worth noting.

UPDATE: MSI have NOT found the problem in testing.
MSI sent a message that suggested I try removing any monitoring software which did not make sense if they had found the problem in testing.
They state that this claim "may be an input error". I am unsure if that also holds true for the statement "We will also continue to pay attention to this issue" which also appeared in the original message.
 
Last edited:
Upvote 0 Downvote
Hello,

First of all, please excuse my English.

I came across this topic because I have a similar problem with my computer. My computer sometimes fails to wake up from sleep mode. This issue started for me over two months ago when I had a problem with one of the MSI Center versions.

The version I had contained a bug that prevented user cooling scenarios from loading at computer startup. I uninstalled that version using the MSI Center cleaning tool and installed a previous one – version 2.0.35.0. After that, the problems with waking up from sleep mode began.

Even with the newest version of MSI Center, the sleep issue still occurs – sometimes the computer wakes up, and sometimes it doesn't. For over a year (since I completed my PC build), I had zero problems with waking up from sleep until I encountered the issue with the MSI Center versions.

Uninstalling MSI Center does not solve the problem for me.

My configuration is a 7600x CPU, a B650 Tomahawk motherboard, and Kingston Fury Beast 6000MHz RAM running on EXPO. Memory Context Restore is enabled, and Power Down has been tested with both Enabled and Auto settings. I have the newest BIOS and AMD drivers installed. I also have PBO Enabled/Curve Optimizer.
Disabling EXPO mode seems to help; the computer woke up without a problem several times in a row but i have nottested it long-term yet.
This setup worked perfectly fine for over a year. The only thing I changed was the MSI Center version, so perhaps it has something to do with it?
 
Upvote 0 Downvote
Radkovich said:
Hello,

First of all, please excuse my English.

I came across this topic because I have a similar problem with my computer. My computer sometimes fails to wake up from sleep mode. This issue started for me over two months ago when I had a problem with one of the MSI Center versions.

The version I had contained a bug that prevented user cooling scenarios from loading at computer startup. I uninstalled that version using the MSI Center cleaning tool and installed a previous one – version 2.0.35.0. After that, the problems with waking up from sleep mode began.

Even with the newest version of MSI Center, the sleep issue still occurs – sometimes the computer wakes up, and sometimes it doesn't. For over a year (since I completed my PC build), I had zero problems with waking up from sleep until I encountered the issue with the MSI Center versions.

Uninstalling MSI Center does not solve the problem for me.

My configuration is a 7600x CPU, a B650 Tomahawk motherboard, and Kingston Fury Beast 6000MHz RAM running on EXPO. Memory Context Restore is enabled, and Power Down has been tested with both Enabled and Auto settings. I have the newest BIOS and AMD drivers installed. I also have PBO Enabled/Curve Optimizer.
Disabling EXPO mode seems to help; the computer woke up without a problem several times in a row but i have nottested it long-term yet.
This setup worked perfectly fine for over a year. The only thing I changed was the MSI Center version, so perhaps it has something to do with it?
Click to expand...

I don't use MSI Center. It's AGESA/BIOS related.
I asked my work colleague to test wake up from sleep on his recently build computer for his son. He tried it two times. One success and one failure. AM5 MSI motherboard.
 
Upvote 1 Downvote
HoovaDevil said:
I have X670E Carbon Wifi with the same issue.
When I built it in May 2024 it would resume from sleep with EXPO enabled.
I decided to upgrade to AM5/X670 quite late because I though all of the bugs and issues would be solved.
However, it eventually started to BSoD at some point. This might be hours or just 20mins.
It took a while before I realised it only occurred after waking from S3.
I updated to BIOS 1J1 and although it would no longer BSoD it would sometimes fail to resume.
It can sometimes resume 1 or 2 times but then fails again.
When it fails the last debug code is 11 (00 can show for 10-20 secs) and the CPU LED (red) is lit. It does not progress and the reset button has to be used.
I have contacted MSI and performed testing, including complete (clean) installation of Windows 11 and updated all drivers but the issue persists.
I have also tested with Ubuntu 24.04.1 both Live USB and a clean install. The issue persists.
MSI suggest that it is a problem with the OS, drivers or a CPU/socker fault (bent pins or fracture) but the system has been stable using BIOS 1J1 since July 2024.
I have also tested legacy BIOS versions and the only one that allows S3 sleep to resume correctly every time is v120.
The fact that the debug code stops at an early stage and does not progress indicates that the other phases required for S3 resume are not being executed.
Of course, if the CPU LED is to be believed it has either failed or is missing and so S3 resume cannot continue.
Has anyone in this thread had success with any 600 series chipset board?
How many of you have opened a ticket? What was the support response?
It appears that MSI does not accept that this is an issue that requires their attention.
I would like to hear what others think and perhaps we can persuade MSI that it does warrant investigation.
Click to expand...
I have the same issue on x870E MSI edge ti which is a brand new motherboard. MSI also tried to blame motherboard and windows except I can totally disprove that because I bought two separate x870E Edge ti's and also tried 3 different sets of RAM sticks, and two versions of windows. 11 23h2 and 11 24h2. Im lucky enough to be able to have enough money to order 4 motherboards at once, so I also swapped to an asus x870A and the issue was instantaneously gone. Switching back to the x870E edge ti and the issue is back. They even tried to blame my power supply. Im running a 7800X3d. a lot of people running 9800X3D's dont seem to have the problem it seems the board is optimized for the new CPU's
 
Upvote 0 Downvote
My own issue with S3 Resume issue has been resolved.

The motherboard is an MSI X670E Carbon Wifi (manufactured December 2023) that came with BIOS E7D70AMS.1G1 (2024/4/15).
The CPU is an AMD 7900X.

I purchased all of the components (PSU, CPU, RAM, NVMe and motherboard) from a single vendor and they allowed all of them to be returned under a single RMA.

They found that the MSI X670E Carbon Wifi and AMD 7900X were at fault.

It could be that I was unlucky to have received 2 faulty components but I leave you to draw your own conclusions.

A few weeks after completing the system it started to BSoD at various times. This could be after many hours of use or sometimes only 20 minutes.
It took a while for me to realise that the BSoD only occurred after resuming from S3 Sleep.

Shortly after contacting MSI (it took 7 days to receive a reply), BIOS E7D70AMS.1J1 was posted (2024/07/10) so I flashed it. It did not fix the issue.

The system was now failing to resume from S3 Sleep. The debug code 11 (or 00) was shown and did not progress.

When testing multiple S3 Sleep/Wake cycles it might resume correctly one time (very rarely) and remain stable for the duration of 8-12 hours afterwards so
BIOS E7D70AMS.1J1 affected some change.

Despite the S3 issue the system was otherwise 100% stable and showed no signs of any hardware fault when running MemTest86 or Prime95.
Note that the system was able to resume from S4 Hibernation correctly.

With the replacement components the system now resumes from S3 Sleep without issue.

If you have the same issue and are still in warranty I recommend that you consider raising an RMA for your parts.
 
Upvote 0 Downvote
Months later, I find this through a google search as my board is exhibiting the same issue. This is actually ridiculous and I might as well just change out the board.
 
Upvote 0 Downvote
Hi all,

Another one by here with the same issue. Currently I'm cheking different options, settings etc... with he MSI support in order to let them know is a problem with the BIOS firmware and nothing with the RAM, CPU or similar. I tested with one RAM module removing one of them, and nothing change. Tested with EXPO profile, resume from sleep show 11 Debug display, red led and black screen without resume properly. Remove EXPO and in some cases 1, or 2 times it works, 3erd usually fail. Tried with EXPO and the last BIOS release available doesn't work. Even tested with " memory context restore (BIOS>Overclocking) and Power Down Enable option( BIOS>Settings>AMD Overclocking)" and no way. It's clear a fail of the firmware of the last BIOS. I will keep testing more thing, but I know what will be the result.
 
Upvote 0 Downvote
This is definitely an MSI BIOS issue. Unfortunately I have to tell you what my ultimate fix was. I sold my PC with the MSI B650 Tomahawk Board and built a new one with ASRock B650 Steel Legend with a DDR5 Kingston Fury Beast 6000 CL30 RAM for the same price. All the problems I had with the RAM compatibility, long boot times etc. are GONE. The cold start boot times from ASRock are just awesome. I got sick of weeks experimenting with the junk MSI BIOS software, playing with the voltages and settings in the MSI overclocking or the AMD overclocking sections in the BIOS. Sorry MSI, but I'm done with your products.
 
Upvote 1 Downvote
Back
Top