Ryzen 5 3600 Crashing During Stress Test

[ConnorBCFC]

Hi.

built a brand new pc today with a Ryzen 5 3600. Due to budget issues, it is using the Stealth cooler which i know isn’t great. I’ll be looking to upgrade this soon. I did use Arctic MX-4 paste to replace the stock paste, which I thought would help things.

I’ve had to lock the frequency at 3.6GHz to avoid stress test crashing. However, even at that frequency, the computer still crashes under Aida64 stress test.

the temps sit at around 75C before the pc freezes and just reboots.

is there anything I can do to avoid this happening? I know stress tests aren’t realistic, but surely the pc shouldn’t be crashing after 5 minutes?

would massively appreciate any help. thanks!

*EDIT* I was just surfing the internet, no stress test open and the PC crashed again.

 

[ConnorBCFC]

You need to give more information on your system

What motherboard
What ram
What drives
What GPU
What PSU
What settings in bios are you running
What voltages are you using for Vcore Vsoc etc
What are speed and timings you using on the RAM and what voltages are you putting through those

have you memtested prior to installing windows

are you installing new windows or just plugging in an existing system

what settings have you used for AIDA stress

Have you been running Hardware Monitoring to see what is going on during idle and stress test

have you got any crash message from windows

your stability issue is highly generic can be any one or multiple issues with hardware.

etc etc.

CPU: Ryzen 5 3600 @ 3.6GHz - Core voltage set to auto
MOBO: MSI B450M MORTAR MAX
RAM: Corsair Vengeance LPX DDR4-3200 16-18-18-36
SSD: Crucial P2 NVME M.2
GPU: XFX RX 580 XXX GTS
PSU: SEASONIC Focus GX-650 80+ Gold

Have mem tested after installing Windows. All clear no errors.
No errors from windows, just a straight up reboot.

 

[ConnorBCFC]

Sounds like a RAM issue to me. What have you got?

Corsair Vengeance LPX DDR4-3200 16-18-18-36. Have tested using memtest86 but no errors.

 

[ConnorBCFC]

RAM faulty? Use different slots, and test individual modules on their own.

Have already memtested but no errors.

 

[ConnorBCFC]

Update.

I’ve just tried resetting the BIOS to defaults. The PC restarted after about 2 minutes. I’ve now got the Windows 10 ‘Choose an option’ menu to troubleshoot.

 

[ConnorBCFC]

Could be a faulty PSU, does it crash if you just load the GPU?

I’ve just tried stressing only the GPU in Aida64.

the PC instantly went very laggy and difficult to use (expected I suppose). Temps have hit 78C at highest. It’s been running for 7 minutes so far with no crashes, which is longer than it’s managed when stress testing the CPU, FPU, Cache and System Memory.

I did notice the actual GPU usage kept jumping around. Usually anywhere between 65% and 100% but occasionally as low as 3% for about a second. Is this something I should be concerned about?

*EDIT* still going 15 minutes in. The SSD is also being used 100%. Is this normal?

*EDIT 2* the SSD usage has now gone down to 1% and stayed there for a while. Weird?

 

[ConnorBCFC]

Are you on a bios later than 7B89v24 as that says "Improved NVMe device compatibility".

Sorry for being dumb- how do I check that?

 

[ConnorBCFC]

Quickest way is to go into the BIOS, and on the main screen it should give the version number.

BIOS version is E7B89AMS.270. Do I need to update it? If so how?

BTW, after resetting the BIOS settings and just locking the CPU frequency to 3.6GHz with no XMP profile I can run CPU and GPU stress tests with no restarts as of yet.

 

[ConnorBCFC]

Aside from any potential BIOS issues, it's quite important to make sure Windows has done all its updates before using the system in anger.

AMD driver packages can crash on install with out of date Windows versions, for example.

Running without XMP is a pretty big performance hit. We should be able to fix that, with a bit of luck. You really do want XMP...

Yeah definitely want XMP. I’ve just turned it on and now running stress tests again. Fingers crossed.

just don’t want the issue to come up again in the future.

 

[ConnorBCFC]

Have you installed the Chipset drivers?

Not yet. Thanks for reminding me. Will do that after these stress tests.

 

[ConnorBCFC]

The BIOS you have looks pretty recent to me. If there's a BIOS date as well, that would confirm, as the version numbers don't quite match those on the website.

Will double check after stress tests. Thanks.

 

[ConnorBCFC]

Grab the latest ones from here. Could well be whats causing your crashes.

https://www.amd.com/en/support/chipsets/amd-socket-am4/b450

Thanks, much appreciated. These tests are going much better now. Would crash around 4mins 30 in previously, now been running for 10 mins plus and all good.

 

[ConnorBCFC]

Two othe


I've done a bit of digging- there's only one newer version, and it's a Beta, so you're basically up to date.
Maybe some settings need tweaking, but if you're running, that's a start.

Really appreciate the help. Thanks!

 

[ConnorBCFC]

Hi all.

Thanks for the responses. For the time being, I appear to have fixed the issue. Updating the chipset drivers, locking the CPU at 3.6GHz and setting the heatsink to 100% at all times has definitely helped.

Will give Prime95 a go and planning to benchmark and stress the GPU later. However, it does pass an Aida64 stress test just fine.

Will update later.

 

[ConnorBCFC]

Hi again.

All was great yesterday, no problems whatsoever. Ran CPU stress tests for several hours, tested some games, ran Heaven for the GPU and no issues.

however, this morning I launched the PC, opened Aida64, and before I could even click start the PC rebooted itself.

it hasn’t happened again since. Should I be concerned? Reminder that this PC isn’t for me.

 

[ConnorBCFC]

Yes, you should be concerned, that isn't normal behaviour. If you can't have the system running normally without restricting the clock speed, or limiting it in any other way there is a fault.

Try running a live USB, like Linux Mint etc to see if that behaves when the system is set up 'normally' and if not you've ruled out the SSD, and the O/S install. It sounds very RAM/PSU, with an outside chance of it being the motherboard.

Thanks. I’ll be honest, I’ve got no idea what I’m doing with Linux. Is there any other way I can test the SSD for faults?

I’ve already tested the RAM and it seems fine. I suppose it could be PSU, but again how do I test that?

 

[ConnorBCFC]

Well this is the crux of the problem when you are building machine without spare parts and tools to test with, if you find a faulty part it is hard to locate it and verify you've found the issue.

I'd normally offer to look at it if you live local, as I am happy to help other forum users out, but you'd not be able to come in due to virus restrictions still in place.

I don't think you really need any experience with linux at all, since Mint works just like Windows on the surface. All you do is create a bootable USB stick, and it will run the O/S from that, you can then connect to your network and download Prime95 etc. then run it.

I’ll try and watch a few videos and give it a go. The problem is these crashes seem not to happen very often, so I might not be able to get it to crash with Linux.

 

[ConnorBCFC]

Well this is the crux of the problem when you are building machine without spare parts and tools to test with, if you find a faulty part it is hard to locate it and verify you've found the issue.

I'd normally offer to look at it if you live local, as I am happy to help other forum users out, but you'd not be able to come in due to virus restrictions still in place.

I don't think you really need any experience with linux at all, since Mint works just like Windows on the surface. All you do is create a bootable USB stick, and it will run the O/S from that, you can then connect to your network and download Prime95 etc. then run it.

Have tried to use CrystalDiskInfo and CrystalDiskMark and all seems okay.

I’m getting similar speeds each time I run it so no signs of inconsistency. I do have my own personal system I could test the power supply on, but it’s a lot beefier than this one and I’m not sure 650W will be enough.

 

[ConnorBCFC]

I sounds like motherboard, PSU, potentially RAM. If new SSD it is very rarely a problem.

Don’t know if you had formatted and reinstalled windows. If the system was not stable at the time of Install it will always be problematic.

anyway it is not common to have multiple components broken as they are all new parts.

something isn’t right. If you got a spare PSU you can certainly test it to rule out PSU issue.

if SMART info is not giving you anything on the SSD then it is highly unlikely the SSD is at fault.

I would focus on RAM, CPU and Motherboard.

with ram you can test individual sticks at a time running single channel and also different dim slots.

I was thinking the same. I have memtested the RAM though and it came back fine.

I’ll try a fresh windows install. Do I need to chose the option to completely clean the drive or can I keep installed files and games etc?

 

[ConnorBCFC]

I've seen something elsewhere about a 3600 system which kept crashing- turns out the mobo was undervolting the CPU at times, causing it to trip over.
Might be worth trying mid-level load-line calibration (LLC) in BIOS. What that wouldn't do is help your temps, but stress-testing with a stock cooler is always going to be right on the edge anyway.

Okay, I’m in the BIOS atm. There’s 2 settings:

• CPU Loadline Calibration Control
• CPU NB Loadline Calibration Control

Each option has the choice of either Auto or 8 different modes, just called Mode 1-Mode 8.

Which should I be changing and what to?