[SOLVED] CPU keeps overheating, causing crashes

[Clive Staples]

Hey folks,

A bit over a year ago I bought an MSI Aegis RS series desktop because it came with an RTX 3080. At that point you couldn't get an RTX 3080 anywhere for love or money. Historically I've built my own rigs from the ground up, but this had decent specs out of the box--and that 3080. It's a mid-tower case though.

I immediately added an NVME drive (Samsung 970 Plus) and set that as the boot device/Windows volume. I also replaced the 16 GB of RAM with 32 GB of Corsair memory. I made both those changes within 2-3 weeks of receiving the PC. Everything has been running wonderfully since then. There haven't been any other hardware changes.

But in the past two-three weeks, I've been getting crashes or summary power shut downs, increasing in frequency. Either the system hangs and is unresponsive to all input (except the case power switch), or else the system literally powers itself down. This even happens when I leave it on overnight - I guess the Windows backup process or a scheduled virus scan must be pegging it.

When I'm watching, I've noticed that when this happens it's usually under load and the fans are going faster than usual. So I stared running HWMON and realized that the CPU cores are idling ok at around 41 to 44 degrees Celsius, but under any kind of load--from a relatively light-weight 3D game to a complex Word doc, or rendering video--they flare up and hover between 80 and 100. During video rendering or gaming, all cores routinely report 95 to 100 degrees Celsius.

I've tried contact MSI's support since it's still under warranty. Their initial response before I noticed the cooling issue was "RMA It". But that doesn't work for me because I rely on this thing for my livelihood (to include rendering videos). I can't just send it off for a couple of weeks and wait for them to maybe fix or replace it a month or two later. I've asked them to help me troubleshoot the cooling issue and so far I haven't heard back.

I've never overclocked anything before so I don't have a lot of experience fighting with cooling issues. My last couple of builds have used Corsair liquid cooling solutions (H70 and H105 I think). This thing has a proprietary cooling loop - looks like about a 120 mm radiator. Again, it was just fine for a year, but something has apparently gone sideways.

I've enabled "Smart Fan Mode" in the BIOS for all case fans--not sure why it was off, but I guess because of noise--and now the fans spin up like mad when it heats up, but they're not getting the job done.

I did notice (and remove) a fair amount of dust built up on the front panel air filter, but cleaning that was unsurprisingly not the solution here.

I don't know what the normal RPM should be for the "pump fan" listed in the BIOS hardware monitoring, but I couldn't help noticing it seems to be running about 5x the RPMs of the other fans.

Any advice would be appreciated, thanks!

---
Intel Core i7 10700K
Windows 10 64-bit Home
MB: MSI Z490-A PRO (MS-7C75)
RAM: 32 GB (8 GB x 4 modules) Corsair Vengeance LPX (DDR4 2933 (PC4 23400 )
Graphics: MSI RTX 3080
PSU: MSI Mag A750 GF
---
Three front panel intake fans (120mm);
One top exhaust fan (120mm);
One radiator exhaust fan (120 mm);
Some type of OEM liquid cooling with ~120 mm radiator

 

[CountMike]

Hey folks,

A bit over a year ago I bought an MSI Aegis RS series desktop because it came with an RTX 3080. At that point you couldn't get an RTX 3080 anywhere for love or money. Historically I've built my own rigs from the ground up, but this had decent specs out of the box--and that 3080. It's a mid-tower case though.

I immediately added an NVME drive (Samsung 970 Plus) and set that as the boot device/Windows volume. I also replaced the 16 GB of RAM with 32 GB of Corsair memory. I made both those changes within 2-3 weeks of receiving the PC. Everything has been running wonderfully since then. There haven't been any other hardware changes.

But in the past two-three weeks, I've been getting crashes or summary power shut downs, increasing in frequency. Either the system hangs and is unresponsive to all input (except the case power switch), or else the system literally powers itself down. This even happens when I leave it on overnight - I guess the Windows backup process or a scheduled virus scan must be pegging it.

When I'm watching, I've noticed that when this happens it's usually under load and the fans are going faster than usual. So I stared running HWMON and realized that the CPU cores are idling ok at around 41 to 44 degrees Celsius, but under any kind of load--from a relatively light-weight 3D game to a complex Word doc, or rendering video--they flare up and hover between 80 and 100. During video rendering or gaming, all cores routinely report 95 to 100 degrees Celsius.

I've tried contact MSI's support since it's still under warranty. Their initial response before I noticed the cooling issue was "RMA It". But that doesn't work for me because I rely on this thing for my livelihood (to include rendering videos). I can't just send it off for a couple of weeks and wait for them to maybe fix or replace it a month or two later. I've asked them to help me troubleshoot the cooling issue and so far I haven't heard back.

I've never overclocked anything before so I don't have a lot of experience fighting with cooling issues. My last couple of builds have used Corsair liquid cooling solutions (H70 and H105 I think). This thing has a proprietary cooling loop - looks like about a 120 mm radiator. Again, it was just fine for a year, but something has apparently gone sideways.

I've enabled "Smart Fan Mode" in the BIOS for all case fans--not sure why it was off, but I guess because of noise--and now the fans spin up like mad when it heats up, but they're not getting the job done.

I did notice (and remove) a fair amount of dust built up on the front panel air filter, but cleaning that was unsurprisingly not the solution here.

I don't know what the normal RPM should be for the "pump fan" listed in the BIOS hardware monitoring, but I couldn't help noticing it seems to be running about 5x the RPMs of the other fans.

Any advice would be appreciated, thanks!

---
Intel Core i7 10700K
Windows 10 64-bit Home
MB: MSI Z490-A PRO (MS-7C75)
RAM: 32 GB (8 GB x 4 modules) Corsair Vengeance LPX (DDR4 2933 (PC4 23400 )
Graphics: MSI RTX 3080
PSU: MSI Mag A750 GF
---
Three front panel intake fans (120mm);
One top exhaust fan (120mm);
One radiator exhaust fan (120 mm);
Some type of OEM liquid cooling with ~120 mm radiator

Single 120mm AIO is just too small for that CPU , not better that most 120 air coolers and maybe barely better than OEM. Don't let 125W TDP fool you, that's just at idle or low usage, at load it may be even double. Looks like they cheaped out on cooling.
You may also want to check for CPU/core voltages. Many a BIOS is set to too high in the name of stability.

 

[KyaraM]

Get a new cooler. Either an AIO 240mm or 360mm in the front sucking air in (or top blowing out, usually slightly worse cooling), or a big air cooler like a Noctua NH-D15 chromaxx.black or a BeQuiet! Dark Rock 4 Pro. The air coolers would likely be cheaper and should do the trick imho. Be aware of RAM clearance issues and case size with the Noctua, though.

 

[Phaaze88]

Someone else had a cooling thread regarding one of those Msi prebuilts.
The reason for the overheating is due to the AIO being a MAG Coreliquid. These units have a problem with early clogging caused by contaminated coolant.
Msi is working on replacing them: https://www.tomshardware.com/news/m...40r-360r-aio-coolers-due-to-sediment-build-up


Basically, you need a new cpu cooler.

 

[Kona45primo]

Not sure how deep you want to dive into the solution but taking a look at the case of your MSI it looks like it's quite restrictive. Getting a new case like a Lian Li Mesh II Performance will give you plenty of options for a large cpu cooler or larger AIO.

Had a friend who purchased an HP Omen & he was running into issues regarding temps with his stock build, 1st we upgraded his CPU cooler to a Noctua U12A, that helped. Next issue was his new 3070 was too large, so he upgraded his case to a Mesh II.

The temp drop from the new cooler was about 10-15 degrees under load with an additional 10 degrees with the new case. The new case also reduced his GPU temps significantly. Plus side is when he's ready for a full system upgrade he's just going to swap back to the stock case & sell his old system and rebuild in the Mesh II.

I also upgraded to the Mesh II from a very restrictive case filled with fans... again I saw a 15 degree temp reduction going from an old properly working H60 120mm AIO to a AS500 plus. other PC component temps dropped significantly as well.

 

[Clive Staples]

Thanks folks. Here's a follow-up question. I've purchased a new CPU cooler (CORSAIR - iCUE H100i ELITE CAPELLIX) and I'm ready to install it.

Almost.

The backplate from the factory-installed CPU cooler looks like this:

I guess the three bolts or screws in the middle are actually the CPU mounting bracket backplate - is that right? So even if I had a tool to remove them I wouldn't want to.

But the cooler backplate seems to be stuck to the mobo using adhesive, which seems normal enough. Any suggestions on how to get it off of there?

I don't have a hairdryer to heat up the adhesive, and it's on there quite snug.

Also, is there any chance I could use that backplate with the new cooler? The threads on the Corsair backplate are adjustable, and I think they can match the configuration of the stock backplate. So maybe the new cooler would work with that backplate?

Is it a bad idea to try that?

Thanks again for your help.

 

[CountMike]

Thanks folks. Here's a follow-up question. I've purchased a new CPU cooler (CORSAIR - iCUE H100i ELITE CAPELLIX) and I'm ready to install it.

Almost.

The backplate from the factory-installed CPU cooler looks like this:

I guess the three bolts or screws in the middle are actually the CPU mounting bracket backplate - is that right? So even if I had a tool to remove them I wouldn't want to.

But the cooler backplate seems to be stuck to the mobo using adhesive, which seems normal enough. Any suggestions on how to get it off of there?

I don't have a hairdryer to heat up the adhesive, and it's on there quite snug.

Also, is there any chance I could use that backplate with the new cooler? The threads on the Corsair backplate are adjustable, and I think they can match the configuration of the stock backplate. So maybe the new cooler would work with that backplate?

Is it a bad idea to try that?

Thanks again for your help.

Try one screw, if hole and treads are same than no problem, Don't even have to take whole cooler off.

 

[Clive Staples]

Thanks for the advice, and thanks again to everyone who posted here.

The screws matched the original backplate holes, so I was able to install the new CPU cooler successfully. The CPU temps are now idling around 31 to 34 degrees, and peaking around 54 - 55 degrees under some basic load. This will hopefully resolve the crashing issue and avoid making a very small coaster of the CPU.

The silver lining here is that the system could benefit from the improved cooling of the new 240mm radiator anyway. Even when the previous 120mm was working correctly, it didn't cool things down this much. It had been configured to exhaust air through the radiator, but Corsair suggested configuring the new radiator fans for intake, so I went with that.

In place of the old radiator and fan on the back panel, I've relocated a 120mm maglev fan that I had previously mounted to the top of the case. It's the only exhaust fan.

Anyway, TL DR everything seems fine now. Thanks again.