Is X570 Chipset overheating a possible cause of NVMe's dropping from the system?

[A Computer Guy]

For the past few days I've suddenly starting having my 2nd and 3rd NVMe drives disconnect in Windows 10. The cause was a bit of a mystery. Both drives are 970 Evo Plus 2TB on the 2nd and 3rd NVMe slot of my X570 Taichi. One is fairly used (27TB written) and the other is lightly used (3.5TB written). I'm not RAID-ing them in any way.

Symptoms

• File copy fails, Windows says something to the effect of device disconnected.
• Applications using the disk might stop responding or fail operations. (VMWare, Handbrake)
• Bluetooth (per the system is actually a USB device) periodically disconnects/reconnects. (discovered this via USB device viewer after being bothered by dinging sounds every few minutes)

Errors

• event id 1, WHEA-Logger
  • A fatal hardware error has occurred. A record describing the condition is contained in the data section of this event.
• event id 11, stornvme
  • The driver detected a controller error on \Device\RaidPort3.
  • The driver detected a controller error on \Device\RaidPort4.

Warnings

• event id 50, ntfs (ntfs)
  • {Delayed Write Failed} Windows was unable to save all the data for the file
• event id 51, disk
  • An error was detected on device \Device\Harddisk1\DR1 during a paging operation.
  • An error was detected on device \Device\Harddisk0\DR0 during a paging operation.
• event id 129, stornvme
  • Reset to device, \Device\RaidPort3, was issued.
  • Reset to device, \Device\RaidPort4, was issued.
• event id 140, Ntfs (Microsoft-Windows-Ntfs)
  • The system failed to flush data to the transaction log. Corruption may occur in VolumeId: D:, DeviceName: \Device\HarddiskVolume4.

After taking the system apart I noticed the chipset heatsink was loose and tightened the rear screws for the heatsink ensuring good mounting pressure with the thermal pad. I suspect the X570 Chipset might have been overheating although the sensor usually reports the chipset temp is usually between 50c-58c.

 

[Yraggul666]

HELL->o there!
I had a similar issue with my MSI X570 Ace (the first one, the one with a fan on the chipset) and my AData SX6000Pro 1TB NVMe, and rarely with my WD850N 1TB NVMe; it wouldn't show up in the system or Bios sometimes after a restart or power-off.
Never had any WHEA errors tho, it just didn't show up in system or Bios and i had to restart and/or power-off again to make it show up again; once it showed up copying and everything else was ok.
Thing is that mobo is at this moment in a friend's system, running perfectly fine, no hiccups, rock solid and everything.
I'm not sure if it's the chipset or smth else like the controller maybe missing somethin. Don't know, just talking jibberish probably. Keep us posted tho.

 

[INSTG8R]

The chipset temps are definitely good I’m on X570S and I peak at around 61C but around the same average as you.
seeing as they are Samsung drives have you tried installing the Samsung NVME driver? I mean it’s a long shot but no hurt in trying it

 

[Ferrum Master]

Try reseating things. CPU and drives.

Do look for firmware updates for the drives.

X570 is known to be picky about drives sometimes. There are no hard conclusions, some makers just silently update firmwares, WD Black 850 being the most popular one having problems with X570.

 

[A Computer Guy]

The chipset temps are definitely good I’m on X570S and I peak at around 61C but around the same average as you.

It may have gotten over 65c before I tightened it back down but thought this chipset was good to 95c - not that I would want to run it that high.

seeing as they are Samsung drives have you tried installing the Samsung NVME driver? I mean it’s a long shot but no hurt in trying it

Samsung has driver for 970 Evo/plus but newer drives only use the standard driver. Had been using it for awhile until I found out I can't restore system image onto a different drive because of it when I upgraded the boot drive to 980 Pro from 970 Evo and tested the image restore. So I just use the Microsoft Standard NVMe driver these days and avoid any NVMe that needs specific drivers so system recovery will be successful.

 

[teck]

It could be anything, the PSU, the MB, the CPU, the RAM, the X570 chipset. It could be even the NVMe drive itself.
It will be hard to detect the issue without changing out some components.

I have X570 Aorus Master and have 2 x Samsung 970 Evo Plus in the first and second port. I constantly get overheating errors from the drives in CrystamDiskInfo 60-70 degrees, but never got any disconnects.
The X570 chipset is around 54 degrees. The chipset fan speed is around 1850 RPM.

Did you ever got any more disconnects after you tighten the chipset fan? What CPU and RAM do you have? What bios are you running?

I am running F34 bios version, I cannot update, as soon as I update the bios to anything newer I get random BSOD. It seems my CPU or RAM becomes unstable with newer version of bioses. I have everything on auto, no OC, no PBO. Some people are recommending to change my old 3800X from 2019.

 

[Drezina007]

I am owner of MSI x570s Tomahawk wifi and have not had problems with SDD drives yet. I have Samsung SSD SM961, Micron 2200S, both NVMe drives, 3 SATA drives, 1 SATA SSD, and 2 other spindle drives. Running dual boot setup with both Windows and Linux and I have not had any file corruptions. On linux I use filesystems with both data and file checksuming so I would have noticed if something is wrong. Also had Crucial P3 4TB PCIe M.2 and no problems.

However this is my 3rd motherboard with AM4 socket and 3rd different AM4 CPU and what I learnt is I must use just only 2 memory modules which are listed on compability list. Because you also mentioned USB disconnections so this is not about SSD drives and in my opinion symptom of memory instability.

Also make sure you have heatsinks on your SSD and try to avoid having GPU blowing hot air on them (impossible on some mobos).

So try to run your setup with only 1 RAM module and do some file operations and to avoid CPU related issues set it to ECO mode in BIOS.

 

[Bill_Bright]

After taking the system apart I noticed the chipset heatsink was loose and tightened the rear screws for the heatsink ensuring good mounting pressure with the thermal pad. I suspect the X570 Chipset might have been overheating although the sensor usually reports the chipset temp is usually between 50c-58c.

I am not a proponent of regularly replacing TIM (thermal interface material). In reality, it rarely EVER needs replacing, contrary to much advice often seen on this and other sites. The true facts are TIM will easily last 10, 15 years or even longer, if left undisturbed. Even if it dries out, the solids left behind are still doing their job of filling the microscopic pits and valleys in the mating surfaces, thus preventing insulating air from getting in there.

Note the ONLY reason TIM comes in a liquid form is simply so it can be squeezed out of the tube, or scooped out of the tub, then evenly spread across the device.

Yes, a fresh new layer of TIM might yield a few (typically <5) degrees improvement, but the fact is, if you "NEED" those few degrees to prevent crossing thermal protection thresholds, you were too close to those thresholds in the first place. In other words, you have other, more critical cooling issues that need addressing first.

I note there is not a single TIM maker, CPU/GPU maker, motherboard maker, computer maker, or heatsink maker that recommends regular replacement of TIM. Not one!

HOWEVER the TIM should ALWAYS be replaced if the cured bond between the mating surfaces has been broken. Once properly applied, then broken, insulating air will be allowed to get in there. Worse, bolder size dust particles (microscopically speaking) and other potential contaminants will be carried in there with that insulating air. Not good.

We don't know if the rear screws where never properly tightened, or if they worked loose due to fan/drive motor vibrations, heavy foot-falls in your room, sonic booms or whatever. In any case, it appears the heatsink was either never properly mounted in the first place, preventing a good bond, or the bond broke later on. Regardless, IMO, because it was loose, the TIM needs to be replaced.

Therefore, I recommend removing the heatsink, thoroughly cleaning the mating surfaces of old TIM (I use 91-93% isopropyl alcohol and a clean, microfiber cloth), applying a proper layer of new TIM, then remount the cooler, ensuring this time, all mounting screws are [evenly] tight and secure.

Just be sure take the necessary ESD prevention precautions; unplug from the wall, touch bare metal of case interior before reaching in to discharge any static in your body.

 

[A Computer Guy]

I am not a proponent of regularly replacing TIM (thermal interface material). In reality, it rarely EVER needs replacing, contrary to much advice often seen on this and other sites. The true facts are TIM will easily last 10, 15 years or even longer, if left undisturbed. Even if it dries out, the solids left behind are still doing their job of filling the microscopic pits and valleys in the mating surfaces, thus preventing insulating air from getting in there.

I did have some CPU water block maintenance as well as temps during this time of storage misbehavior were 15c higher than usual warranting some inspection. For some reason the TIM was discolored and I wonder if the thermal properties of the TIM were failing after seeing this. One of the dark spots you can see is above the IO die although it could be nothing of consequence.

Note the ONLY reason TIM comes in a liquid form is simply so it can be squeezed out of the tube, or scooped out of the tub, then evenly spread across the device.

Yes, a fresh new layer of TIM might yield a few (typically <5) degrees improvement, but the fact is, if you "NEED" those few degrees to prevent crossing thermal protection thresholds, you were too close to those thresholds in the first place. In other words, you have other, more critical cooling issues that need addressing first.

I note there is not a single TIM maker, CPU/GPU maker, motherboard maker, computer maker, or heatsink maker that recommends regular replacement of TIM. Not one!

HOWEVER the TIM should ALWAYS be replaced if the cured bond between the mating surfaces has been broken. Once properly applied, then broken, insulating air will be allowed to get in there. Worse, bolder size dust particles (microscopically speaking) and other potential contaminants will be carried in there with that insulating air. Not good.

We don't know if the rear screws where never properly tightened, or if they worked loose due to fan/drive motor vibrations, heavy foot-falls in your room, sonic booms or whatever. In any case, it appears the heatsink was either never properly mounted in the first place, preventing a good bond, or the bond broke later on. Regardless, IMO, because it was loose, the TIM needs to be replaced.

Yea I'm not sure either although it was working for 2yrs 24/7 without incident but obviously if it's not secured properly it can break the bond at any time. I'll have to add check back plate screw to my list of things to do when getting a new motherboard. Temperature stability probably had been an advantage up until this point being on 24/7. You can add vibration from sump pump to your list. Unfortunately my PC is now located above my basement sump pump for the past year (on top of enclosed shelf) .

Therefore, I recommend removing the heatsink, thoroughly cleaning the mating surfaces of old TIM (I use 91-93% isopropyl alcohol and a clean, microfiber cloth), applying a proper layer of new TIM, then remount the cooler, ensuring this time, all mounting screws are [evenly] tight and secure.

Any recommendations for 1mm thermal pad? When I inspected the original thermal pad it was not dry but heat over the past 2yrs seem to have taken its toll on it for sure.

Just be sure take the necessary ESD prevention precautions; unplug from the wall, touch bare metal of case interior before reaching in to discharge any static in your body.

Also washing hands before working on the PC is a good way to help reduce ESD issues.

 

[Bill_Bright]

For some reason the TIM was discolored and I wonder if the thermal properties of the TIM were failing after seeing this.

Some TIM will discolor after a few heat/cool cycles. That is normal and has no significant impact on cooling properties AS LONG AS the discoloration is not due to oxidation (indicating that air somehow got in there) or some contaminate that got in there during an improper application or if the cured bond was later broken.

Note it is not easy for the cured bond to break. It typically happens due to abuse such as rough handling of the computer during transport (or cleaning), or, sadly and ironically, by users twisting the cooler too hard to see if the cooler is loose, breaking the bond in the process.

I don't see how washing the hands helps with ESD - especially since drying the hands by rubbing them in a towel, or via an air dryer can certainly reintroduce many 1000s of volts of static again without the person even realizing it. However, washing will certainly remove excess skin oils and other contaminates that might get in there. So yeah, a clean environment is always better.

 

[DaemonForce]

The X570 chipset temps should not be an issue.
If you want to test that idea, move the disk to the top slot.
I did that and haven't seen any real world difference.

 

[A Computer Guy]

It could be anything, the PSU, the MB, the CPU, the RAM, the X570 chipset. It could be even the NVMe drive itself.
It will be hard to detect the issue without changing out some components.

970 evo swapped to another PC they appear to be working fine so I think I can rule out the drives themselves. Copy tests were failing quickly within a 15 seconds at the height of the problem. In the other motherboard hundreds of 1TB copied back and forth no problem.

I have X570 Aorus Master and have 2 x Samsung 970 Evo Plus in the first and second port. I constantly get overheating errors from the drives in CrystamDiskInfo 60-70 degrees, but never got any disconnects.

My drives typically operate all in the mid 40's according to Samsung Magician however HWiNFO64 reports all 3 temps for each drive and the highest temp is about inline with what you have reported during heavy activity but I haven't gotten any temperature related errors or throttling that I am aware of prior to the incident.

The X570 chipset is around 54 degrees. The chipset fan speed is around 1850 RPM.

Same but my fan never engages. I'm not sure what the critical point is where it should turn on but I have tested that it works using manual control and the fan doesn't really effect the outcome when the temperature seems to be in the typical range (mid 50's) for the x570 chipset.

Did you ever got any more disconnects after you tighten the chipset fan?

So far things are looking good now after firmly securing the chipset heatsink. Usually I only change one thing at a time but I needed to address some minor debris in my water block and repasted it at the same time.

What CPU and RAM do you have?

see my pc specs link in sidebar. 5950x, Crucial ECC 128GB. So far flawless in operation.

What bios are you running?

P5.50

I am running F34 bios version, I cannot update, as soon as I update the bios to anything newer I get random BSOD. It seems my CPU or RAM becomes unstable with newer version of bioses. I have everything on auto, no OC, no PBO. Some people are recommending to change my old 3800X from 2019.

Yea I don't bother with PBO anymore other than to reduce power consumption. I'm not using PBO for awhile. You can see the origin of this story here... https://www.techpowerup.com/forums/...eated-something-or-busted-custom-loop.297316/

The X570 chipset temps should not be an issue.
If you want to test that idea, move the disk to the top slot.
I did that and haven't seen any real world difference.

View attachment 356378

The top slot has had no failures and is the hottest drive too of the bunch. (980 pro vs 970 evo plus)

additional info

On thing that changed about the PC setup during the first incident was I connected an AOC USB laptop monitor to the PC. I don't necessarily think it's related but the power requirement on the back say 5v at 2amp (it uses a double USB connector and I connected it to my USB3 ports). If the USB drew too much current, could that could damage the motherboard or chip I/O die? The monitor worked flawlessly so I think the usage of it was simply coincidental to the NVMe problem and the NVMe problem continued after the monitor was removed. I may have to keep in mind to use a powered USB hub for such large devices.

However this is my 3rd motherboard with AM4 socket and 3rd different AM4 CPU and what I learnt is I must use just only 2 memory modules which are listed on compability list. Because you also mentioned USB disconnections so this is not about SSD drives and in my opinion symptom of memory instability.

Unfortunately I don't have (haven't found) an I/O map for my motherboard to confirm but I think the Bluetooth USB is connected to the chipset so all the failures NVMe + Bluetooth seem to be connected with the chipset. All other NVMe & USB devices were working uninterrupted when observing via USB Device Monitor.

I don't see how washing the hands helps with ESD - especially since drying the hands by rubbing them in a towel, or via an air dryer can certainly reintroduce many 1000s of volts of static again without the person even realizing it. However, washing will certainly remove excess skin oils and other contaminates that might get in there. So yeah, a clean environment is always better.

Sufficient moisture in the hands helps prevent static discharge or as someone once told me. I've never had an ESD incident when washing hands before a build or maintenance although I have no hard evidence on hand to support the assertion.

 

[Drezina007]

128GB ECC RAM? It should be at least 4 modules of RAM. I would try just only 1 memory stick and retry tests.
BTW I am suffering bluetooth disconnectins as-well but between Smartphone and earphones in some areas sometimes. Changing earphones did not help so I would not use it as serious issue in your case.
What nvme slot is directly cpu attached? Have you tried this one?

 

[Bill_Bright]

Sufficient moisture in the hands helps prevent static discharge or as someone once told me.

Ummm, no!!! You don't want moisture on your hands or anywhere near electronics (live or not).

What you were probably told is high humidity "air" suppresses and helps prevent the build up of static. Low humidity environments (like the Arizona desert) can result is extreme levels of static. Furnace systems in the winter can really dry out the air (and skin) too and create high-static environments.

If you can walk across a carpeted room and get zapped on a doorknob (or the dog's nose), you might consider getting a humidifier in the room, or added to your furnace.

This is why environmentally controlled computer/server rooms don't just control the temperature in the room, they control the humidity too. And in fact, they will actually add moisture to (humidify) the air when needed. The ideal humidity in server rooms and data centers is ~50%. When I lived and worked in Arizona and New Mexico, we even "issued" spray bottles of water (with a few drops of fabric softener) for those in carpeted offices to help keep workers from zapping computers. At least until anti-static carpet became commonplace.

 

[A Computer Guy]

128GB ECC RAM? It should be at least 4 modules of RAM. I would try just only 1 memory stick and retry tests.
BTW I am suffering bluetooth disconnectins as-well but between Smartphone and earphones in some areas sometimes. Changing earphones did not help so I would not use it as serious issue in your case.
What nvme slot is directly cpu attached? Have you tried this one?

Yes that's right 4 modules. Note this ram config has been in place and working since sometime around Nov 2023 so it's been in place for a long time already. The top NVMe slot is direct to CPU. That is working fine all tests on that slot are working fine with all drives.

So far since securing the chipset heatsink the situation has improved 100%. More testing to occur over the next two days so I can test at the hottest part of the day. I'll have to check periodically to see if those heatsink screws loosen over time perhaps due to vibration or temperature changes. It could just be they were never secured properly from the factory and worked their way further out over time.

 

[Bill_Bright]

so I can test at the hottest part of the day. I'll have to check periodically to see if those heatsink screws loosen over time

Well, of course the hottest part of the day is only an issue if the room temp is not controlled and ambient temps rise significantly with the outdoor temps. Load on the system is typically more influential. Either way, do keep close eye on the temps - and those screws.

 

[A Computer Guy]

Well, of course the hottest part of the day is only an issue if the room temp is not controlled and ambient temps rise significantly with the outdoor temps. Load on the system is typically more influential. Either way, do keep close eye on the temps - and those screws.

So far chipset is staying below 60c in successful 200GB transfers in various combinations between all drives in all 3 M.2 positions including an Optane I put in PCIe x16 slot 2. All drives are peaking in the mid 50's with the 970's staying just below 50c. Ambient is around 32c currently.

 

[Bill_Bright]

Ambient is around 32c currently.

Whoa! 32°C (89.6°F) is pretty warm for a room temp. That would be uncomfortable for me - though perhaps I am just spoiled. If it is not yet the hottest part of the day, and if these temps are normal for this time of year, I might consider a small window AC for that room.

I personally would not like my chipset temps to be that high in low demand scenarios and simply transferring files is NOT a very demanding task. I will assume your case interior is free of heat trapping dust so that leaves me to wonder about your case cooling? Too often I have seen case cooling suffer when users implement alternative cooling solutions for their CPU. Even with excellent CPU, graphics, and VRM cooling, there still needs to be a sufficient amount of cool air flowing through the case for other heat sensitive components and devices, such as the chipset and drives. You might want to add another case fan, or replace current fans with higher capacity fans.

That said, no fan solution will cool an inanimate (not living) object cooler than the ambient air. So now we might be back to adding a small window AC for that room.

 

[Drezina007]

You could also help it with silicon glue to secure the heatsink properly. I would use paper to create a barrier in size of heatsink and height of gap between mobo and heatsink, paint it to black from outside and stick it to mobo with eg super glue. And when it is dry pour silicon glue between this barrier and the chipset but not on it. Dry it overnight and then put thermal compound on chipset and last layer of silicon and stick heatsink on it. Precise and time-consuming job but it will pay off I think because the extra silicon will not only secure the heatsink but also will help transferring more heat.

 

[A Computer Guy]

Whoa! 32°C (89.6°F) is pretty warm for a room temp. That would be uncomfortable for me - though perhaps I am just spoiled. If it is not yet the hottest part of the day, and if these temps are normal for this time of year, I might consider a small window AC for that room.

I personally would not like my chipset temps to be that high in low demand scenarios and simply transferring files is NOT a very demanding task. I will assume your case interior is free of heat trapping dust so that leaves me to wonder about your case cooling? Too often I have seen case cooling suffer when users implement alternative cooling solutions for their CPU. Even with excellent CPU, graphics, and VRM cooling, there still needs to be a sufficient amount of cool air flowing through the case for other heat sensitive components and devices, such as the chipset and drives. You might want to add another case fan, or replace current fans with higher capacity fans.

That said, no fan solution will cool an inanimate (not living) object cooler than the ambient air. So now we might be back to adding a small window AC for that room.

Wify kicked me out of the AC room this year so I'm dealing with higher PC temps and as much ice water as I can stand.

At the moment case is open cooling while testing.

 

[Bill_Bright]

Wify kicked me out of the AC room this year so I'm dealing with higher PC temps and as much ice water as I can stand.

Well, I got rid of my wife so I get to rule the roost again! Though she might say she got rid of me! At least we are still good friends.

At the moment case is open cooling while testing.

Okay but understand that is not normally good for cooling. The side panel works to create a slight positive pressure inside the case and channel/funnel the air flow over the sensitive devices and "through" the case on its desired path. Removing that panel and there's no pressure and no flow over the sensitive devices.

For that reason, when removing the side panel, I typically recommend blasting a desk fan in there to compensate.

 

[A Computer Guy]

Okay but understand that is not normally good for cooling. The side panel works to create a slight positive pressure inside the case and channel/funnel the air flow over the sensitive devices and "through" the case on its desired path. Removing that panel and there's no pressure and no flow over the sensitive devices.

My box fan is tilted next to me and doing double duty cooling my person and the open case.

 

[thesmokingman]

Drives die, even nvme. I had great luck with Seagate SSDs, every one I bought have died randomly which makes it a fabulous hit rate. My Sammy SSD are still kicking!

 

[A Computer Guy]

Drives die, even nvme. I had great luck with Seagate SSDs, every one I bought have died randomly which makes it a fabulous hit rate. My Sammy SSD are still kicking!

I had a Seagate Firecuda 530 at one point and it just would not work in the Taichi after some incident occurred after installing Seagate's SSD tools. It just simply refused to recognize the device in UEFI afterwards.

 

[mclaren85]

I had the exact problem when playing NFS Heat (pirate version). I have two nvme, and when I play nfs heat, one nvme always disconnects.