Question Is my CPU, GPU, PSU or OC causing my PC to freeze under load?

Jul 25, 2021
13
0
10
Hi guys,

First post, so go easy!

I have been having an issue the last few months, whereby during heavy load (i.e. at graphically intensive points of certain games - Death Stranding, CyberPunk 2077) my PC locks, screen freezes, becomes totally unresponsive and all connected devices (bluetooth, Xbox controller) become disconnected. The only course of action at this point is to force a restart by holding down the power button on the PC, and the freezing seems to become more likely to happen the sooner I play the game again. I can consistently replicate it by stress-testing the PC under Heaven Benchmark and Prime95.

I've struggled to identify the cause of the freezing, because it first started happening after I replaced my CPU, CPU cooler, motherboard, RAM, GPU and SSD at the same time. I've also started overclocking; I must admit I'm not very sure what I'm doing with the overclocking but I deliberately only made small changes in line with the guidance available online.

My spec is:
CPU: AMD Ryzen 7 3800X (1 year)
CPU cooler: Corsair Hydro 100i RGB (6 months)
MB: ROG STRIX B250-F GAMING (1 year)
SSD: Samsung 970 EVO 1TB (1 year)
GPU: GTX 2070 (1 year)
RAM: 32GB (1 year)
PSU: Corsair CX750 (7 years old)

As you can see the PSU is the oldest component, but I'm not sure how to tell if it is at fault for the freezing. I couldn't find anything in the Event Viewer that coincided with the crash, and I don't really know how else to identify the cause.

Many thanks for any advice you can give!

Oli
 
Hi guys,

First post, so go easy!

I have been having an issue the last few months, whereby during heavy load (i.e. at graphically intensive points of certain games - Death Stranding, CyberPunk 2077) my PC locks, screen freezes, becomes totally unresponsive and all connected devices (bluetooth, Xbox controller) become disconnected. The only course of action at this point is to force a restart by holding down the power button on the PC, and the freezing seems to become more likely to happen the sooner I play the game again. I can consistently replicate it by stress-testing the PC under Heaven Benchmark and Prime95.

I've struggled to identify the cause of the freezing, because it first started happening after I replaced my CPU, CPU cooler, motherboard, RAM, GPU and SSD at the same time. I've also started overclocking; I must admit I'm not very sure what I'm doing with the overclocking but I deliberately only made small changes in line with the guidance available online.

My spec is:
CPU: AMD Ryzen 7 3800X (1 year)
CPU cooler: Corsair Hydro 100i RGB (6 months)
MB: ROG STRIX B250-F GAMING (1 year)
SSD: Samsung 970 EVO 1TB (1 year)
GPU: GTX 2070 (1 year)
RAM: 32GB (1 year)
PSU: Corsair CX750 (7 years old)

As you can see the PSU is the oldest component, but I'm not sure how to tell if it is at fault for the freezing. I couldn't find anything in the Event Viewer that coincided with the crash, and I don't really know how else to identify the cause.

Many thanks for any advice you can give!

Oli
7 years old that should be the old green label power supply it was garbage when they made it.
EDIT green label. https://www.google.com/imgres?imgur...hUKEwjC5fP_tP_xAhVMf6wKHfYTAHAQMygCegUIARCFAg


You should also set everything back to factory defaults because it could be a bad overclock. You can enable DOCP for the memory but leave the CPU and GPU stock.

If it still fails when you run the stress test then set the memory back to factory settings no DOCP and try it again but I think the PSU is the problem.
 
Last edited:
...

My spec is:
CPU: AMD Ryzen 7 3800X (1 year)
...
MB: ROG STRIX B250-F GAMING (1 year)
...

There's your problem.... that motherboard is LGA-1151, Ryzen 7 3800X CPU is AM-4socket.... 🙄

Could you really mean a ROG STRIX B350F? or 450F? or 550F? hopefully so.

If you haven't lately, update BIOS and reset CMOS. Don't try any "overclocking" and run full stock for a while. Overclocking 3rd gen is pointless anyway, and frequently winds up killing performance much as yours is doing.
 
Hey guys, thanks so much for the quick replies.

Just returned OCs to stock and will test and monitor any performance changes. To clarify, I was OCing my GPU not the CPU.

@Zerk2012 yeah that's the one. I don't really know how to measure performance of a PSU, to me either it powers everything consistently or it doesn't. It's always done that and only now am I questioning it. Would this or this be a suitable upgrade? Like you say, it seems I'm due an upgrade anyway, so maybe I just bite the bullet. What would a 'bad overclock' mean here? I have OC'd past the limit of the hardware? Or is it bad luck? Is it likely to cause damage?

@drea.drechsler - good spot. It's ROG STRIX B450-F, apologies for that. My CPU definitely fits and runs well for most daily tasks and games. I am not too confident updating BIOS, I've heard I should be very careful. Is that likely to help / be worth the risk?

Thanks again both!

Oli
 
... I am not too confident updating BIOS, I've heard I should be very careful. Is that likely to help / be worth the risk?
...
It fully depends on the BIOS version you're on, of course. But there have been significant changes to BIOS' as the platform has matured with stability improvements along with performance improvements. So it could make a difference.

Also, it's not so very risky if you do the update using the utility that's in the BIOS. It won't start the update unless it verifies an un-corrupted BIOS file that's correct for your board is being used. Just follow instructions in your manual. The most likely cause of failure would be a power outage during the update so don't start if storms are in your area...or until you set your cat out of the room.

Also ensure your chipset drivers are up to date. Get them from the AMD support web site, not the mobo mfr's site. That way there's no chance of also getting the bloatware they like to bundle with it.
 
It fully depends on the BIOS version you're on, of course. But there have been significant changes to BIOS' as the platform has matured with stability improvements along with performance improvements. So it could make a difference.

Also, it's not so very risky if you do the update using the utility that's in the BIOS. It won't start the update unless it verifies an un-corrupted BIOS file that's correct for your board is being used. Just follow instructions in your manual. The most likely cause of failure would be a power outage during the update so don't start if storms are in your area...or until you set your cat out of the room.

Also ensure your chipset drivers are up to date. Get them from the AMD support web site, not the mobo mfr's site. That way there's no chance of also getting the bloatware they like to bundle with it.
Thank you. Just updated chipset drivers, never thought to do that before. Will give BIOS update a go using the utility. Should I expect to reinstall Windows when I do it?
 
No, Windows re-install is not needed. The only thing you should do is reset CMOS after completing it. That's to be sure all settings are in correct defaults for the new BIOS.
OK BIOS updated via internet to a more recent version. It did not detect the most recent version available online, but it's certainly more current than the previous version. I tried to update via USB drive but got a "the selected file is not a proper BIOS file" error...
 
Managed to update it to the latest version, and remove all OCs. The issue persists, this time it even happened on Rocket League (much less intensive game).

Bitten the bullet and gone for a Corsair RM850x to beef up the PSU.
 
No, Windows re-install is not needed.
....
Bitten the bullet and gone for a Corsair RM850x to beef up the PSU.
Probably overkill as a 650W unit would have sufficed and still leave headroom to upgrade to a 6800XT or 3080TI class GPU. But the Corsair RM(x) series are good PSU's, I've an RM550x and an RM650x. Cool running and quiet even under heavy loads, with the fan spinning.

Have you tried un-installing the GPU drivers with DDU and then re-installing?

Also, from command prompt with admin rights do a system file check by issueing: SFC /SCANNOW
 
Last edited:
Probably overkill as a 650W unit would have sufficed and still leave headroom to upgrade to a 6800XT or 3080TI class GPU. But the Corsair RM(x) series are good PSU's, I've an RM550x and an RM650x. Cool running and quiet even under heavy loads, with the fan spinning.

Have you tried un-installing the GPU drivers with DDU and then re-installing?

Also, from command prompt with admin rights do a system file check by issueing: SFC /SCANNOW
Yeah, just want to ensure there's a lot of breathing space for future upgrades. Let's hope it does the trick.

I seem to be advised to run SFC /SCANNOW for most of my PC issues! :sweatsmile: will give it a go and also try reinstalling the GPU drivers. Thank you.
 
....
I seem to be advised to run SFC /SCANNOW for most of my PC issues! :sweatsmile: will give it a go and also try reinstalling the GPU drivers. Thank you.
It seemsWindows fairly easily gets it's system files corrupted. This helps find the inegrity violations and fix it up. I run it after every Windows update and it finds and fixes something probably 1/2 the time. I've even run it after fresh installs and it almost always finds integrity violations to fix.

But if you've not uninstalled video drivers with DDU you definitely should give it go. It's not that hard, just follow the instructions.
 
Yeah, just want to ensure there's a lot of breathing space for future upgrades. Let's hope it does the trick.

I seem to be advised to run SFC /SCANNOW for most of my PC issues! :sweatsmile: will give it a go and also try reinstalling the GPU drivers. Thank you.

You should run DDU [Display Driver Uninstaller, a free program, just google it] in safe mode to remove video drivers.

As to purchasing a Corsair RM850x, yes it is a bit overkill, however it should last you through your next video card upgrade and the RM series is much better than the old CX you replaced. So, don't feel bad.
 
Managed to update it to the latest version, and remove all OCs. The issue persists, this time it even happened on Rocket League (much less intensive game).

Bitten the bullet and gone for a Corsair RM850x to beef up the PSU.
Make sure you post back with the results when you get the new power supply. I expect to hear all good now!

EDIT I did post that as a cure on the first response to your post.
 
@drea.drechsler that makes sense then, good to know - it did indeed fix some corrupted files this time. Thank you for all the advice.
@Danra yes, that's what Drea has been recommending, I'm a little nervous about doing that but will give it a go if no success with the new PSU. Exactly what I was after, so hopefully a good investment.
@Zerk2012 will do, fingers crossed! Yes, I had suspected it may be the cure but of course wanted to explore less expensive suggestions first.
 
Damnit. The crash just occurred. Performance is very good aside from that.

One thing to note is that if I restart the PC immediately, it's prone to happen again even during low load.
 
Damnit. The crash just occurred. Performance is very good aside from that.

One thing to note is that if I restart the PC immediately, it's prone to happen again even during low load.
Looking back I see you have 32Gb RAM...how many sticks? 4 sticks can be more difficult to get stable, try just two for a while. Then the other two.

Another thing is increase voltage on the RAM. DDR4 is good up to 1.5V but operating over1.4 isn't recommended for 24/7 use.
 
From what i know, since upgrading to 4x memory my negative -10 all cores on my cpu started to show way more unstability. Even games started giving me fatal errors. If you want better performance of the cpu only enable PBO and nothing else. The overclocking isn't worth much imo, if you would like better temps lower the PTT or run the cpu on eco mode.
 
Looking back I see you have 32Gb RAM...how many sticks? 4 sticks can be more difficult to get stable, try just two for a while. Then the other two.

Another thing is increase voltage on the RAM. DDR4 is good up to 1.5V but operating over1.4 isn't recommended for 24/7 use.
Yeah it's 4 x 8gb (Corsair Vengeance RGB PRO DDR4 3200 MHz). Will try 2 at a time soon. I had no idea running 4 could be cause stability problems 🤔

Noticed that Memory was running at 2133MHz, so have now overclocked to 3200MHz - performance is better but the issue persists. My BIOS didn't allow me to change voltage, it was visible but greyed out.

Also worth noting I'm now able to run Prime 95 and Heaven Benchmark and have it not crash. Before I started this thread, this would systematically cause the crash, so that's some progress!
 
...My BIOS didn't allow me to change voltage, it was visible but greyed out.
....
I don't know of any good reason that would be. Not on that board. It would be a problem in my mind as adjusting DRAM voltage is essential to stabilizing extreme memory configurations which a 4 DIMM array most definitely is. You realy should try running on two sticks or maybe even one, swapping them out to identify if you have one bad DDR4 stick.
 
Last edited:
I don't know of any good reason that would be. Not on that board. It would be a problem in my mind as adjusting DRAM voltage is essential to stabilizing extreme memory configurations which a 4 DIMM array most definitely is. You realy should try running on two sticks or maybe even one, swapping them out to identify if you have one bad DDR4 stick.
Will do, thank you. Appreciate all the advice!
 
Apologies for going quiet here, I did recently try with 2/4 sticks of RAM and sadly still experienced the issue, I still need to try the other 2/4 sticks as a next test. You really notice the performance dip when downgrading to 16gb.

In the last few months I have obviously gone through various graphics card driver updates and even updating my BIOS and an upgrade to Windows 11, none of them have fixed the issue. There are still no suspicious events in the Event Viewer or any other clue as to the cause, so I'm stumped.

Going back to the RAM voltage, @drea.drechsler, what would you recommend? You said don't go over 1.4, but also said 1.5 should be fine. So should I go for 1.4? I feel like there's something more major that we're missing, though - I don't see why running at the default RAM voltage would be the cause of fatal system errors while gaming.

Many thanks,
Oli