RAM not recognized on hot reboots after suspend/resume — X870 Tomahawk WiFi

S

sorn.deni154e02db

New member
Joined
Feb 15, 2026
Messages
13
The title. The mainboard is an MAG X870 Tomahawk WiFi.

If I suspend and resume (sometimes it takes a few cycles to trigger this) and then do a hot reboot, I get greeted by a message saying that my CPU or RAM configuration has changed. When I enter the UEFI settings, I see that (usually) only half of the RAM (one 32GB stick) is detected—sometimes even less (once, the UEFI detected only 8GB). Simply shutting down and cold booting fixes the issue. So far I have tested with MCR disabled and the issue persists. Still have to check the Power Down, it's possible I left it enabled.

Has anyone else experienced this or something similar?
 
Sort by date Sort by votes
Are the memory modules listed in the Memory compatibility list of the mainboard?
How's the speed of the memory being set?
Has mainboard's BIOS updated to the latest 7E51v1A71 to check?
 
Upvote 0 Downvote
I also have this issue on the X870 Tomahawk WiFi (non-E version) since I bought it a year ago. I'm currently on 7E51v1A68, but every bios I tried before that did not resolve the problem. The issue is tied to letting the PC Sleep. It does seem like the PC needs to be a sleep for more than a few seconds (like maybe 30 minutes) before the issues reveals itself.

I have two symptoms:
- USB ports devices may not be detected after waking. This happens on Linux AND Windows operating systems. It doesn't matter if I use EXPO or Default memory timings. Unplugging the impacted USB device and plugging it back in resolves that issue.
- A reboot of the PC (after waking from a Sleep, mentioned above) is guaranteed to report that the CPU or RAM has changed. When I go into the BIOS, one of the RAM slots will show a fraction of the memory it should have (around 2GB instead of 16GB). It's usually channel A or channel B, but I've seen both show only 2GB (total of 4GB). I used to have some BIOS screenshots of this, but can't seem to find them.

Observation: To avoid the CPU/RAM changed message is to always power down completely instead of using reboot if the PC had recently gone to sleep. It's annoying but prevents having to deal with the BIOS.

I have not yet tried the most recent BIOS version, but I'm not expecting any behavior change.

Main Board: MSI MAG X870 Tomahawk WIFI
Bios Version: 7E51v1A68 (but happened on all previous versions I tried)
Video Card: Sapphire 7800XT Pulse
PSU brand and model: Seasonic Prime GX-1000
12v rail Rating of listed PSU: 996W
CPU: Ryzen 9700X
Memory: Kingston (2x16GB) 32GB - KF560C30BBEK2-32
SSD/ HDD: SN850X (1TB), KC3000 (1TB), A400 (960GB) SSD
CPU COOLER: Thermalright Phantom Spirit 120 SE
 
Upvote 0 Downvote
I've also been experiencing this but only since switching to Linux. I was running Windows 11 previously and never saw the issue at all. I checked the RAM compatibility list and my RAM is definitely listed.

I was running 7E51v1A50 but when this started cropping up I updated to 7E51v1A71 to see if that would resolve the issue - it didn't. I've just been fully powering down my PC instead of rebooting to get around it, but I would prefer not to have to do that.
 
Upvote 0 Downvote
jackmranso15a802e8 said:
I've also been experiencing this but only since switching to Linux. I was running Windows 11 previously and never saw the issue at all. I checked the RAM compatibility list and my RAM is definitely listed.

I was running 7E51v1A50 but when this started cropping up I updated to 7E51v1A71 to see if that would resolve the issue - it didn't. I've just been fully powering down my PC instead of rebooting to get around it, but I would prefer not to have to do that.
Click to expand...
Thanks for the reply. Unfortunately I forgot I posted this here, and just started a new thread complaining about the same issue. Interesting because I also use Linux, however I have experienced the issue with Windows as well. My guess is, in your case it either started happening bit later because something degraded, or you haven't immediately noticed it. Similar like my previous issue with MSI AM5 board (b650 Tomahawk Wifi) which started with long boot times, then progressed to computer not being able to boot/post at all, this also started similarly. At first every say third (after suspend) reboot would result in RAM not being detected, but now it's almost every boot I think. I mean I did recently reboot after suspending because I forgot and it worked that, but that's rare. I was also suspecting the issue is related to MCR, but I have tried with MCR off, and it might have decreased the chance of the issue happening, but it didn't solve it.
 
Upvote 0 Downvote
jf.led.projec159302e2 said:
I also have this issue on the X870 Tomahawk WiFi (non-E version) since I bought it a year ago. I'm currently on 7E51v1A68, but every bios I tried before that did not resolve the problem. The issue is tied to letting the PC Sleep. It does seem like the PC needs to be a sleep for more than a few seconds (like maybe 30 minutes) before the issues reveals itself.

I have two symptoms:
- USB ports devices may not be detected after waking. This happens on Linux AND Windows operating systems. It doesn't matter if I use EXPO or Default memory timings. Unplugging the impacted USB device and plugging it back in resolves that issue.
- A reboot of the PC (after waking from a Sleep, mentioned above) is guaranteed to report that the CPU or RAM has changed. When I go into the BIOS, one of the RAM slots will show a fraction of the memory it should have (around 2GB instead of 16GB). It's usually channel A or channel B, but I've seen both show only 2GB (total of 4GB). I used to have some BIOS screenshots of this, but can't seem to find them.

Observation: To avoid the CPU/RAM changed message is to always power down completely instead of using reboot if the PC had recently gone to sleep. It's annoying but prevents having to deal with the BIOS.

I have not yet tried the most recent BIOS version, but I'm not expecting any behavior change.

Main Board: MSI MAG X870 Tomahawk WIFI
Bios Version: 7E51v1A68 (but happened on all previous versions I tried)
Video Card: Sapphire 7800XT Pulse
PSU brand and model: Seasonic Prime GX-1000
12v rail Rating of listed PSU: 996W
CPU: Ryzen 9700X
Memory: Kingston (2x16GB) 32GB - KF560C30BBEK2-32
SSD/ HDD: SN850X (1TB), KC3000 (1TB), A400 (960GB) SSD
CPU COOLER: Thermalright Phantom Spirit 120 SE
Click to expand...
Thanks. Yeah, I updated to the latest BIOS (7E51v1A82) and it didn’t help. I didn’t notice any USB issues on this board, but I did have that exact problem on a B650 Tomahawk before. That one also had issues waking from suspend, but it was a different one (wouldn’t wake up at all. This was possibly caused by a faulty 7900x or its MCU, which I have successfully RMAd).

I did run into a different USB. In BIOS/UEFI I had the Wifi/Bluetooth option set to “Only Bluetooth” mode (Bluetooth part is USB). That worked fine until (I assume) the other (RAM) issue caused a partial BIOS reset, which left the Bluetooth device stuck in some weird state.

I was getting long boot times because the kernel was waiting for an unknown USB device that would appear and disappear. Fix was just going back into BIOS and re-selecting “Only Bluetooth” (any other option would’ve worked as well).
 
Upvote 0 Downvote
Hi everyone,

I’m having a strange issue with my MSI MAG X870 Tomahawk WiFi and I would like to know if anyone else has seen something similar.

The problem seems to happen after Linux tries to suspend and fails or wakes back up immediately. The system makes an attempt to enter sleep, then comes back, but it does not behave correctly afterwards. Sometimes it becomes unstable or partially frozen. If I then press the reset button or do a hot reboot, on the next POST I get a black BIOS/firmware screen saying something like the hardware has changed and I need to enter BIOS with F1.

When I enter the BIOS, one of my 16 GB RAM modules is sometimes detected incorrectly as:

  • Unknown
  • 2 GB instead of 16 GB
After another reboot, the same module may be detected correctly again as 16 GB.

At first I thought it could be a bad DIMM or a bad RAM slot, but I swapped the two RAM modules between slots and the issue does not seem to consistently follow the same slot or the same memory bank. Also, I ran Memtest+ open-source and it did not report any memory errors.

The important detail is that I have not seen this happen under Windows. It seems related to Linux suspend/resume or a failed suspend followed by a hot reboot/reset.

Hardware:

  • Motherboard: MSI MAG X870 Tomahawk WiFi
  • CPU: AMD Ryzen 7 9800X3D
  • GPU: ASUS TUF RTX 4080
  • RAM: 2 x 16 GB DDR5 — Crucial Pro RAM CP2K16G64C38U5B with EXPO2 Profile (6000 mhz)
  • PSU: Corsair RM1000x
  • Monitor: ASUS VG28UQL1A, 4K / 144 Hz, HDMI 2.1
  • USB devices connected: keyboard KROM Kernel TKL, Logitech G502 Hero, Sound Blaster X3, eDNI reader card
Software / OS:
  • Fedora 44
  • GNOME / Wayland: yes
  • NVIDIA driver from RPM Fusion: 595.58.03
  • Installed NVIDIA packages include:
    • xorg-x11-drv-nvidia-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-power-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-libs-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-cuda-libs-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-kmodsrc-595.58.03-1.fc44.x86_64
Power / suspend checks:
/sys/power/mem_sleep shows:
s2idle [deep]

So the system is currently using deep.
The NVIDIA suspend/resume/hibernate services are installed and enabled:

nvidia-suspend.service: enabled
nvidia-resume.service: enabled
nvidia-hibernate.service: enabled

They show as inactive/dead when checked manually, which I believe is normal because they only run during suspend/resume.

GNOME automatic sleep on AC is disabled:

gsettings get org.gnome.settings-daemon.plugins.power sleep-inactive-ac-type
'nothing'

BIOS / memory settings:

  • Current BIOS version: 7E51v1A82
  • EXPO: Profile 2 (6000mhz)
  • Power Down Enable: disabled
What I have tried so far:
  • Swapped both RAM modules between slots
  • Ran Memtest+ with no errors found
  • Confirmed Windows does not seem to trigger the issue
  • Confirmed NVIDIA power package is installed
  • Confirmed NVIDIA suspend/resume/hibernate services are enabled
  • Confirmed Linux is using deep sleep mode
  • Disabled GNOME automatic suspend on AC
Possible pattern:

The issue seems to be:

Linux failed suspend / resume
↓
system becomes unstable or does not fully suspend
↓
hot reboot or reset
↓
BIOS reports hardware change
↓
one RAM module may appear as Unknown / 2 GB
↓
after another reboot, RAM is detected correctly again

This makes me suspect a BIOS/AGESA/platform firmware issue related to suspend/resume or warm reboot memory training, rather than a physically bad RAM module.

I have also seen users suggest testing:

usbcore.autosuspend=-1

in case the issue is related to USB/xHCI autosuspend during Linux sleep, but I have not fully tested that yet.

Has anyone else with the MAG X870 Tomahawk WiFi seen RAM being misdetected after Linux suspend/resume or a hot reboot?

Any advice on BIOS versions, AGESA versions, Memory Context Restore, Power Down Enable, EXPO stability, or Linux suspend settings would be appreciated.

Thanks.
 
Last edited:
Upvote 0 Downvote
davidmorenomagi162602ec said:
Hi everyone,

I’m having a strange issue with my MSI MAG X870 Tomahawk WiFi and I would like to know if anyone else has seen something similar.

The problem seems to happen after Linux tries to suspend and fails or wakes back up immediately. The system makes an attempt to enter sleep, then comes back, but it does not behave correctly afterwards. Sometimes it becomes unstable or partially frozen. If I then press the reset button or do a hot reboot, on the next POST I get a black BIOS/firmware screen saying something like the hardware has changed and I need to enter BIOS with F1.

When I enter the BIOS, one of my 16 GB RAM modules is sometimes detected incorrectly as:

  • Unknown
  • 2 GB instead of 16 GB
After another reboot, the same module may be detected correctly again as 16 GB.

At first I thought it could be a bad DIMM or a bad RAM slot, but I swapped the two RAM modules between slots and the issue does not seem to consistently follow the same slot or the same memory bank. Also, I ran Memtest+ open-source and it did not report any memory errors.

The important detail is that I have not seen this happen under Windows. It seems related to Linux suspend/resume or a failed suspend followed by a hot reboot/reset.

Hardware:

  • Motherboard: MSI MAG X870 Tomahawk WiFi
  • CPU: AMD Ryzen 7 9800X3D
  • GPU: ASUS TUF RTX 4080
  • RAM: 2 x 16 GB DDR5 — Crucial Pro RAM CP2K16G64C38U5B with EXPO2 Profile (6000 mhz)
  • PSU: Corsair RM1000x
  • Monitor: ASUS VG28UQL1A, 4K / 144 Hz, HDMI 2.1
  • USB devices connected: keyboard KROM Kernel TKL, Logitech G502 Hero, Sound Blaster X3, eDNI reader card
Software / OS:
  • Fedora 44
  • GNOME / Wayland: yes
  • NVIDIA driver from RPM Fusion: 595.58.03
  • Installed NVIDIA packages include:
    • xorg-x11-drv-nvidia-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-power-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-libs-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-cuda-libs-595.58.03-1.fc44.x86_64
    • xorg-x11-drv-nvidia-kmodsrc-595.58.03-1.fc44.x86_64
Power / suspend checks:
/sys/power/mem_sleep shows:
s2idle [deep]

So the system is currently using deep.
The NVIDIA suspend/resume/hibernate services are installed and enabled:

nvidia-suspend.service: enabled
nvidia-resume.service: enabled
nvidia-hibernate.service: enabled

They show as inactive/dead when checked manually, which I believe is normal because they only run during suspend/resume.

GNOME automatic sleep on AC is disabled:

gsettings get org.gnome.settings-daemon.plugins.power sleep-inactive-ac-type
'nothing'

BIOS / memory settings:

  • Current BIOS version: 7E51v1A82
  • EXPO: Profile 2 (6000mhz)
  • Power Down Enable: disabled
What I have tried so far:
  • Swapped both RAM modules between slots
  • Ran Memtest+ with no errors found
  • Confirmed Windows does not seem to trigger the issue
  • Confirmed NVIDIA power package is installed
  • Confirmed NVIDIA suspend/resume/hibernate services are enabled
  • Confirmed Linux is using deep sleep mode
  • Disabled GNOME automatic suspend on AC
Possible pattern:

The issue seems to be:

Linux failed suspend / resume
↓
system becomes unstable or does not fully suspend
↓
hot reboot or reset
↓
BIOS reports hardware change
↓
one RAM module may appear as Unknown / 2 GB
↓
after another reboot, RAM is detected correctly again

This makes me suspect a BIOS/AGESA/platform firmware issue related to suspend/resume or warm reboot memory training, rather than a physically bad RAM module.

I have also seen users suggest testing:

usbcore.autosuspend=-1

in case the issue is related to USB/xHCI autosuspend during Linux sleep, but I have not fully tested that yet.

Has anyone else with the MAG X870 Tomahawk WiFi seen RAM being misdetected after Linux suspend/resume or a hot reboot?

Any advice on BIOS versions, AGESA versions, Memory Context Restore, Power Down Enable, EXPO stability, or Linux suspend settings would be appreciated.

Thanks.
Click to expand...
This issue is very similar to what I described. It could even be the same issue, because suspend/resume problems can be caused by the system configuration. I have two Linux distros on this machine, Ubuntu and CachyOS, and I haven’t touched the GRUB kernel parameters, systemd, or anything else that affects suspend/resume and Nvidia, and I have no suspend/resume issues.

The problem is that it’s really hard to know whether this is caused by the mainboard or, say, the processor’s MCU. I now have two AM5 systems, mainly so I can check and test components.

If you can, try booting with a different CPU and see whether the issue persists. If you can replicate it, it’s probably the motherboard; otherwise, it’s probably the CPU. There’s a small chance it’s RAM or something else, but IMO it’s a small one. You can also try disabling memory context restore, to see if it helps.
 
Upvote 0 Downvote
I just noticed that you seem to be using X11. Over the last few years, Nvidia has been focused on Wayland AFAIK. It’s possible they barely test X11 anymore. If you’re using X11, try Wayland and see if it makes any difference.
 
Upvote 0 Downvote
Not sure why you have so many posts, makes it hard for someone to get an answer with disjointed conversations. I too am experiencing this issue with the my x870 tomahawk and Bazzite, but works fine on Windows.

I solved one issue of it immediately resuming from sleep by disabling GPP1, similar to the issue with the gigabyte board but that used GPP0. With nvidia fixing Linux sleep issues in their drivers, I thought I’d finally primarily boot Bazzite. However, I’m still running into the issue you described here with the change in memory configuration intermittently after a reboot. I reproduce the issue slightly different than you.

I reproduce the issue by either letting the computer sleep for a while, then restarting it. Or I sleep and restart it multiple times until it triggers the configuration change. I’ve been troubleshooting this randomly over a couple years hoping it would get fixed.

I have 32GB of memory 2x16GB CL30 Corsair memory with XMP profiles because none with EXPO were available at that time. I thought maybe that was my issue, but the XMP profile, the AMD OPP profile, or manually setting the memory timings eventually still result in the configuration change. I quadruple checked my memory was in the right slots. I’ve updated to the latest BIOS. I would love to switch to Linux on this machine but cannot because of this issue. I am kind of fed up with this board.
 
Upvote 0 Downvote
psychoundergroundegg said:
Not sure why you have so many posts, makes it hard for someone to get an answer with disjointed conversations. I too am experiencing this issue with the my x870 tomahawk and Bazzite, but works fine on Windows.

I solved one issue of it immediately resuming from sleep by disabling GPP1, similar to the issue with the gigabyte board but that used GPP0. With nvidia fixing Linux sleep issues in their drivers, I thought I’d finally primarily boot Bazzite. However, I’m still running into the issue you described here with the change in memory configuration intermittently after a reboot. I reproduce the issue slightly different than you.

I reproduce the issue by either letting the computer sleep for a while, then restarting it. Or I sleep and restart it multiple times until it triggers the configuration change. I’ve been troubleshooting this randomly over a couple years hoping it would get fixed.

I have 32GB of memory 2x16GB CL30 Corsair memory with XMP profiles because none with EXPO were available at that time. I thought maybe that was my issue, but the XMP profile, the AMD OPP profile, or manually setting the memory timings eventually still result in the configuration change. I quadruple checked my memory was in the right slots. I’ve updated to the latest BIOS. I would love to switch to Linux on this machine but cannot because of this issue. I am kind of fed up with this board.
Click to expand...
Thanks, this is very useful. Your case sounds almost identical to mine: X870 Tomahawk, Linux affected, Windows fine, and memory configuration change after suspend followed by reboot.

In my case the problem also seems to appear after a failed or unstable Linux suspend/resume and then a hot reboot/reset. Memtest+ reports no errors and the issue does not seem to consistently follow the same DIMM or slot.

The GPP1 detail is very interesting. I have not tried disabling GPP1 yet. Do you know exactly which device/function GPP1 controls on this board? I am a bit concerned it may disable some PCIe/M.2/USB functionality depending on the board mapping.

At this point this really looks like a board firmware/AGESA/Linux suspend-resume/warm reboot issue rather than bad RAM.
 
Upvote 0 Downvote
davidmorenomagi162602ec said:
Thanks, this is very useful. Your case sounds almost identical to mine: X870 Tomahawk, Linux affected, Windows fine, and memory configuration change after suspend followed by reboot.

In my case the problem also seems to appear after a failed or unstable Linux suspend/resume and then a hot reboot/reset. Memtest+ reports no errors and the issue does not seem to consistently follow the same DIMM or slot.

The GPP1 detail is very interesting. I have not tried disabling GPP1 yet. Do you know exactly which device/function GPP1 controls on this board? I am a bit concerned it may disable some PCIe/M.2/USB functionality depending on the board mapping.

At this point this really looks like a board firmware/AGESA/Linux suspend-resume/warm reboot issue rather than bad RAM.
Click to expand...
To be clear you wouldn’t be disabling devices, just disabling them from waking the system. See the discussion here https://github.com/ublue-os/bazzite/issues/1928
 
Upvote 0 Downvote
A little update or big depending, as I’m still testing. I noticed the system always does a warm reboot, as there is never an audible click on reboot. I was trying to find a way to make the system do a cold reboot.

I came across this post
https://www.reddit.com/r/Bazzite/comments/1q03hiq
but was trying a bunch of stuff before this, and eventually came back to it. Apparently, reset instructions are handed off to the ACPI system. This means we can configure the Linux kernel parameters to do a full power-off reboot. In Bazzite, the kernel reboot parameter is configured to do this with “rpm-ostree kargs --append='reboot=pci'” The “cold” setting for this parameter is unnecessary as it is likely default and didn’t do anything when I set it by itself, as it was mentioned in the URL.

After rebooting a couple times so the kernel parameters can take effect, I heard an audible click of a full power off. The system came back on and my memory modules which both had reverted to 2GB (during all these resets) now showed 16GB each.

This has appeared to fix the issue, as I haven’t been able to reproduce it yet. However, it makes me wonder why this doesn’t happen on Windows. Maybe unrelated, but I have also enable FCH Spread Spectrum which may reduce interference.

As for your issue with your system crashing or having a black screen waking from sleep, NVIDIA has fixed this in their Linux 595.58.03
drivers see
https://www.reddit.com/r/Bazzite/comments/1s2mk4d
so you may want to install a newer drivers as I’m no longer having that issue on Bazzite.

Hope this is clear and helps. Seems like the Linux kernel might need a fix to handle this how Windows does.
 
Upvote 0 Downvote
Back
Top