CPU Overheating and PC Shutting Down after HDD Replacement

[gurmansach44]

I had a hard drive with a bad sector that I had to replace, so after doing that I noticed that my computer was getting very hot while playing games that would run much cooler before. I thought it was nothing, so I brushed it off until one day my computer shutdown in the middle of a session from overheating. The first thing I did was check Piriform's Speccy ( I already had it downloaded from before because of the bad hard drive) for the actual temperature. It read 22-30 C at startup, and slowly climbed up to 45-50 C, where before it would have stayed at most at 30 C on idle. Then, when playing games it would quickly spike to 70 C, keep spiking all the way to around 83 C, and eventually shut off the PC (I think the max was 85 C). So I first took off the CPU cooler, cleaned all the thermal paste from the CPU and cooler, and reapplied more, then reseated the cooler. Tested it, no changes. Then, I cleaned out any dust I could find in the case, without change. The last thing I tried was switching the sata III port the new HDD was using, since I saw that on another forum post, and still nothing. I'm out of ideas at this point, so any would help. Thanks in advance.

Specs:
Motherboard: MSI 760GMA-P34 (FX)
CPU: AMD FX 8320 (stock clock speeds)
CPU Cooler: Cooler Master Hyper T2
Ram: Crucial Ballistix Sport 8GB Kit (4GBx2) DDR3 1600
GPU: Gigabyte GTX 960 Windforce OC 4GB
PSU: EVGA Bronze 500W 80+
Storage: WD Blue 1TB 5400 rpm 3.5 inch

 

[BFG-9000]

Does the cooler's fan spin? Do the top of its heatpipes get hot?

You have a 125w chip and a cooler rated for only 130w TDP so everything must be perfect for it to work.

 

[gurmansach44]

Does the cooler's fan spin? Do the top of its heatpipes get hot?

You have a 125w chip and a cooler rated for only 130w TDP so everything must be perfect for it to work.

Yes. The cooler's fans are always spinning, and I felt the heat pipes when I opened the pc, they were hot.

 

[gurmansach44]

I opened Fortnite as a test to see the temps on the cpu, and simply having the game open made them gradually climb to 80C from 40C. Within about 5 seconds of closing the game the temperature went down to 40C again without any interruption.

 

[gurmansach44]

Anyone going to reply?

 

[Karadjgne]

Don't bump, ppl have lives so it might take a minute to get a reply.

I'm going to assume you had the latest bios revision from before the hdd swap. Since you only list 1 form of storage, I'm also going to assume that you had to reinstall windows. Did you also go back to the motherboard website and reinstall all of the motherboard drivers? Windows drivers by nature are very generic, and often not worth much as mobo drivers are specific to the mobo components, especially stuff like USB, Sata etc.

 

[gurmansach44]

Don't bump, ppl have lives so it might take a minute to get a reply.

I'm going to assume you had the latest bios revision from before the hdd swap. Since you only list 1 form of storage, I'm also going to assume that you had to reinstall windows. Did you also go back to the motherboard website and reinstall all of the motherboard drivers? Windows drivers by nature are very generic, and often not worth much as mobo drivers are specific to the mobo components, especially stuff like USB, Sata etc.

Sorry about that, I'm just frustrated because I haven't really had a time where my pc actually worked like it should. Anyway, when I got the new drive, I used an external drive and made an image of the old HDD on that, then used a winpe session on a usb drive with macrium reflect to copy the image onto the new drive. I made an exact copy of the old drive, so there shouldn't be any problems there.

 

[Karadjgne]

Ok. Well first, speccy is a great little program. For Intel cpu's. For Amd FX, it sucks. Pretty much every program has issues with Amd FX other than Amd Overdrive or MSI Afterburner so I'm a little unsure of the validity of the results.

As is, the 8320 is a 125w cpu and most of the 760g chipsets don't support 125w cpus, they'll only support upto 95w cpus. If you look up the specs, while the bios actually does have support for the 125w FX, every single one has the notation of 'not supported by the manufacfurer', which is not the case for the 95w cpus such as the 6300 or 8320E/8370E.

You have also added a 130w cooler to a 125w TDP cpu. This will be an issue. TDP is Thermal Design Power set by Amd under use of Amd standard applications, which equate to @ nominal pc usage, basically light office work or web surfing and watching YouTube. It's by no means the maximum power the cpu can obtain, which is Peak power. Peak is @1.5x - 2x TDP depending on the cpu, and in your case, it's closer to 2x TDP. So figure on gaming usage being closer to 70% of 250w (that's about 175w) on a 130w cooler and thats a problem. You are going to see high temps no matter what on any game that pushes more than a couple of threads.

Pretty much in a nutshell, you've got too much cpu for the rest of the pc to handle heatwise, the board doesn't have the heatsinking and capacity to do the job right, the cpu cooler is vastly too small for a 125w Vishera, it's more compatible with the 95w versions at stock clocks or mild OC where vcore is under stock usage.

 

[gurmansach44]

Ok. Well first, speccy is a great little program. For Intel cpu's. For Amd FX, it sucks. Pretty much every program has issues with Amd FX other than Amd Overdrive or MSI Afterburner so I'm a little unsure of the validity of the results.

As is, the 8320 is a 125w cpu and most of the 760g chipsets don't support 125w cpus, they'll only support upto 95w cpus. If you look up the specs, while the bios actually does have support for the 125w FX, every single one has the notation of 'not supported by the manufacfurer', which is not the case for the 95w cpus such as the 6300 or 8320E/8370E.

You have also added a 130w cooler to a 125w TDP cpu. This will be an issue. TDP is Thermal Design Power set by Amd under use of Amd standard applications, which equate to @ nominal pc usage, basically light office work or web surfing and watching YouTube. It's by no means the maximum power the cpu can obtain, which is Peak power. Peak is @1.5x - 2x TDP depending on the cpu, and in your case, it's closer to 2x TDP. So figure on gaming usage being closer to 70% of 250w (that's about 175w) on a 130w cooler and thats a problem. You are going to see high temps no matter what on any game that pushes more than a couple of threads.

Pretty much in a nutshell, you've got too much cpu for the rest of the pc to handle heatwise, the board doesn't have the heatsinking and capacity to do the job right, the cpu cooler is vastly too small for a 125w Vishera, it's more compatible with the 95w versions at stock clocks or mild OC where vcore is under stock usage.

I understand this, but I've had this pc since summer of 2016 and this problem only came up when I changed the hard drive, so I don't think it's any coincidence that my computer suddenly won't be able to handle the heat. Yes, it did get hot before, but only read 70 C in speccy. Also, I'll download overdrive and see what the temps are.

 

[Karadjgne]

Just remember, Overdrive works backwards. It reads the thermal limit of the cpu, and just how close you are to it. So a reading of 30 would be great, you are 30°C away from max, but a reading of 20 would be worse, being only 20°C away from max. Just don't get stuck on exact numbers, the FX does not have the same thermal sensors on the cores as Intel, so relies more on package temps than actual core temps.

 

[gurmansach44]

So when it says the thermal margin is 45 C it is that much away from max?
Also, if it was that, that would be telling me the max was 75 C and it has gone at least 10 above that.
Update: The thermal margin went into negatives when I opened a game, I checked speccy and it was at 80 again, so I think it is pretty accurate (and both of them are showing that my cpu is getting way too hot).

 

[Karadjgne]

Technically, the FX have a core max of 62°C, but that's impossible to guage as there's no sensor strip on the cores as such, and with the ambient heat travels to the core surfaces where there are strips, getting 72°C is more normal, but I've seen plenty of ppl take the FX upto 80ish as max. Temps on them have always been a bone of contention. This is why Overdrive is so good at what it does. Having a thermal margin of 45 really means the cpu is well under what it considers max (it's an algorithm mostly, but set by Amd as firmware not by any program) so is doing just fine. You'll know you are in trouble if you hit single digits and should definitely look for a better cooling solution if OD is consistently reading below 20 when you are gaming or other heavy cpu usage. If you hit 0 thermal margin, you are at Max according to the cpu and can expect throttling as a minimum result, or even shutdown. As far as temps go, you are thinking in terms of Intel, you shouldn't. It's meaningless, only the thermal margin and your relative closeness to 0 means anything. You have an AMD cpu, so any relationship to Intel temps and the above or below 70°C safety line are moot. As I said earlier, the exact number is not as important as what the number is. 40,45,50 are basically the same, long way away from thermal margin, 25,20,15 are awfully close.
It might be that the cpu has decided that 0 thermal margin is set at 100, so 45 would be roughly 55°C in Intel terms, or 0 tm could be 62°C in which case 45 would be 17°C in Intel terms. This is why Intel temps don't apply, like 75°C, as you don't know exactly what 0 thermal margin is, only the cpu does. All that's important is you have 45°C to spare until you have reached max.

 

[gurmansach44]

Technically, the FX have a core max of 62°C, but that's impossible to guage as there's no sensor strip on the cores as such, and with the ambient heat travels to the core surfaces where there are strips, getting 72°C is more normal, but I've seen plenty of ppl take the FX upto 80ish as max. Temps on them have always been a bone of contention. This is why Overdrive is so good at what it does. Having a thermal margin of 45 really means the cpu is well under what it considers max (it's an algorithm mostly, but set by Amd as firmware not by any program) so is doing just fine. You'll know you are in trouble if you hit single digits and should definitely look for a better cooling solution if OD is consistently reading below 20 when you are gaming or other heavy cpu usage. If you hit 0 thermal margin, you are at Max according to the cpu and can expect throttling as a minimum result, or even shutdown. As far as temps go, you are thinking in terms of Intel, you shouldn't. It's meaningless, only the thermal margin and your relative closeness to 0 means anything. You have an AMD cpu, so any relationship to Intel temps and the above or below 70°C safety line are moot. As I said earlier, the exact number is not as important as what the number is. 40,45,50 are basically the same, long way away from thermal margin, 25,20,15 are awfully close.

I got to -9 C running a light game (Microsoft FSX). Closed it before the pc shut down again.

 

[Karadjgne]

Yep. Definitely need to do something about the cooling. It could be that the cooler isn't clicked in fully on one pin, could be the paste isn't seated properly and fully covering the cores on the ihs or could be that after the swap, a windows update actually unparked all your cores fully and now you are using all 8 to some degree where prior you were only using 4. But one thing for sure, a 130w CM hyper T2 is not enough cooler for a fully functional 125w FX Vishera under decent loads.

 

[gurmansach44]

Yep. Definitely need to do something about the cooling. It could be that the cooler isn't clicked in fully on one pin, could be the paste isn't seated properly and fully covering the cores on the ihs or could be that after the swap, a windows update actually unparked all your cores fully and now you are using all 8 to some degree where prior you were only using 4. But one thing for sure, a 130w CM hyper T2 is not enough cooler for a fully functional 125w FX Vishera under decent loads.

Again, the cooler isn't the problem, since I've had this setup for nearly 2 years now without an issue, up until I swapped the hdd. Also, I've already reseated the cooler and replaced the thermal paste after cleaning it thoroughly off the cpu and the cooler, so that shouldn't be an issue either. I also cleaned out any dust I could find in and around the cooler, so that should not be a problem. If a windows update unparked all my cores, how would I go about fixing it?

 

[Karadjgne]

It is fixed. Core parking was an eco thing amd tried out, parking the cores means they don't get used at all. Unparked you can use all the cores.

 

[mdd1963]

If the cooler's fan appears to be working, it is either now not locked down/secured properly, or, the thermal interface material perhaps has solidified /dried up and perhaps is now marginal...

Removing sink and carefully applying a ultra thin layer, and making sure the cooler is properly locked down might eliminate both possibilities.

 

[gurmansach44]

If the cooler's fan appears to be working, it is either now not locked down/secured properly, or, the thermal interface material perhaps has solidified /dried up and perhaps is now marginal...

Removing sink and carefully applying a ultra thin layer, and making sure the cooler is properly locked down might eliminate both possibilities.

Again, I've already done this.

 

[gurmansach44]

It is fixed. Core parking was an eco thing amd tried out, parking the cores means they don't get used at all. Unparked you can use all the cores.

If that's the case, what's causing it to overheat?

 

[Karadjgne]

Loose or even busted plastic pin on the heatsink, maybe just not clicked in all the way after the repaste. Paste badly spreading or loose cooler not applying enough pressure to spread the paste sufficiently and evenly. Malware. Bad driver allocation like sometimes nvidia drivers will foul up the fonts section making for 25-50% cpu usage at all times. Changes to startup programs making for looping drivers, causing excessive cpu usage. Microcode fixes due to Spectre and Meltdown changing bios values resulting in over volting. All cores now fully unparked and used vrs prior parked cores keeping temps lower under nominal usage. Accidentally hitting the 'OC now' button whole inside the case, causing software OC to raise buss speeds, and vcore to match.

There's half a hundred different ways to get the cpu running hotter, there's no easy fix without eliminating possibilities first. You say you repasted, check the cooler/paste first. Just because you did do it, doesn't mean something is not quite right. Not hard to bust one of those plastic pin tabs or just as easy is to think it's fully seated but 1 pin wasn't quite, and slipped back out. AMD has a bigger core under the ihs, so more care is needed to make sure paste is fully covering the lid. An easy way to check is using Hwmonitor, it'll pull all 8 core temps (the number isn't important, what is is all the numbers should be relatively close)

You can park cores in windows power settings. But parking 2 cores means you effectively have an fx6300 with the fx8320 name.

 

[gurmansach44]

Loose or even busted plastic pin on the heatsink, maybe just not clicked in all the way after the repaste. Paste badly spreading or loose cooler not applying enough pressure to spread the paste sufficiently and evenly. Malware. Bad driver allocation like sometimes nvidia drivers will foul up the fonts section making for 25-50% cpu usage at all times. Changes to startup programs making for looping drivers, causing excessive cpu usage. Microcode fixes due to Spectre and Meltdown changing bios values resulting in over volting. All cores now fully unparked and used vrs prior parked cores keeping temps lower under nominal usage. Accidentally hitting the 'OC now' button whole inside the case, causing software OC to raise buss speeds, and vcore to match.

There's half a hundred different ways to get the cpu running hotter, there's no easy fix without eliminating possibilities first. You say you repasted, check the cooler/paste first. Just because you did do it, doesn't mean something is not quite right. Not hard to bust one of those plastic pin tabs or just as easy is to think it's fully seated but 1 pin wasn't quite, and slipped back out. AMD has a bigger core under the ihs, so more care is needed to make sure paste is fully covering the lid. An easy way to check is using Hwmonitor, it'll pull all 8 core temps (the number isn't important, what is is all the numbers should be relatively close)

You can park cores in windows power settings. But parking 2 cores means you effectively have an fx6300 with the fx8320 name.

Okay, so I checked task manager to see cpu usage, and the most I've seen are spikes of 10%, and it stays at 1% most of the time on idle. In games it goes to 30-50%, which I think is normal. I noticed that sometimes the clock speed spikes to 3700mhz (max) and temp spikes to 40-50 C on idle, and then goes back down to 1400mhz and around 20 C instantly, is that normal? When replacing hard drive, nothing was moved but the drive, and temps went up, but they stayed the same as after hdd replacement when I reseated the cooler and respread the thermal paste, so it can't be either of those. There is no malware as far as I know, since nothing is using over 10% cpu on idle, and usually 50% max, and nothing noticeable changed after meltdown.

 

[Karadjgne]

Task manager usage: perfectly normal.
Idle usage temps: perfectly normal, when background stuff opens up initially it sets a high cpu usage for a second, then settles down, so temporary spikes are perfectly normal.

Nothing noticeable changed after Meltdown : meaning something unnoticed did change. I'd check the bios, and reset it to optimal default values. After that, I'd run Asus RealBench and Overdrive and monitor the cpu checking the thermal margin. If the cpu can handle that and stay stable, I'd start with a vcore of 1.35v and cpu-z and try again, in cpu-z it'll show your running vcore during stress, and if it's (should be) lower, then lower the vcore closer to that value. So if vcore is 1.118v then lower vcore from 1.35 to 1.24v. Try again and keep doing that gradually until you notice cpu-z vcore showing very little movement down. Then start lowering vcore by very small amounts, 1.24 to 1.238 etc until you loose stability, then bump the vcore back up a notch or 2. That'll drop cpu temps by as much as 10-15°C under loads.

Welcome to overclocking.

 

[gurmansach44]

Task manager usage: perfectly normal.
Idle usage temps: perfectly normal, when background stuff opens up initially it sets a high cpu usage for a second, then settles down, so temporary spikes are perfectly normal.

Nothing noticeable changed after Meltdown : meaning something unnoticed did change. I'd check the bios, and reset it to optimal default values. After that, I'd run Asus RealBench and Overdrive and monitor the cpu checking the thermal margin. If the cpu can handle that and stay stable, I'd start with a vcore of 1.35v and cpu-z and try again, in cpu-z it'll show your running vcore during stress, and if it's (should be) lower, then lower the vcore closer to that value. So if vcore is 1.118v then lower vcore from 1.35 to 1.24v. Try again and keep doing that gradually until you notice cpu-z vcore showing very little movement down. Then start lowering vcore by very small amounts, 1.24 to 1.238 etc until you loose stability, then bump the vcore back up a notch or 2. That'll drop cpu temps by as much as 10-15°C under loads.

Welcome to overclocking.

That's a lot to take at once, I'll try my best to follow. Thank you so much for all the help.

 

[Karadjgne]

It's commonly referred to as undervolting, but for all intents and purposes it's overclocking 101. The procedure is identical to what OC'rs do just without changing the multipliers/buss. Both amd and Intel deliberately set stock voltages high, every cpu responds slightly differently, so to cover the stability of any cpu, the vcore is jacked up. By undervolting, you are reducing the overhead voltage of the cpu, which lowers the cpu operating temp, upto a point where the cpu decides enough is enough, and you get bsod or freezes or lockups etc. That's instability. Simple fix is to add back a small amount of vcore until you are over what the cpu demands, but as low as the cpu can handle.

With your sub-standard cooling, anything that'll lower cpu temps is a bonus.

 

[gurmansach44]

So, I don't know how, but I went into power settings and made it so the max cpu usage is 80%. Whether it is an incorrect reading, or some miracle happened, my cpu (according to amd overdrive) will not not surpass ~30 C thermal margin. I just didn't want to take chances with changing the voltage and accidentally changing something I shouldn't have, since this is my first homemade pc and I definitely do not have the capability of replacing any parts or trashing anything in the first place. I don't really trust these results, but I'll keep testing to see if anything happens again, and if they were faulty readings. Although I'm a bit worried on what happened when I turned my computer on today. It took really long to boot, and didn't even manage to make it to the windows logo. So, I pressed ctrl alt delete, still nothing. Reset button, still nothing. Held down the power button for a good minute. Nothing. In the end I didn't see anything else, so I flicked the power switch off and back on, and turned on the pc successfully (this session is where I made the change in control panel from 100% max usage to 80% max usage). Should I be concerned of this, and do you think the temps after me changing the settings are true? Again, thanks for all the help.