Bought the components for this rig in August 2019 for my mother in law to use at work. Her typical workload is word processing / web stuff.
Worked well for a few months, now gets BSoD crashes (IRQL, Store Exception, DPC Watchdog, and others). Crashes usually happen when idle or under low load.
First few months: fine.
Last couple of months: one-two idle crashes per week.
Now: daily idle crashes.
Crashes after 1-2 hours of idle typically. Sometimes the crashes are not BSoD, sometimes they are to BIOS.
Ryzen 3600 + Stock Cooler
MSI B450 Tomahawk
Corsair Vengeance LPX 16 GB (2x8) 3200 x 2 (32 GB total)
EVGA 650 GQ 80+ Gold 650W
Samsung 970 EVO 1 TB SSD NVMe M.2
Gigabyte GTX 1050 Ti Windford OC 4GB GDDR5 PCI-E
Steps I've taken:
Crystal Disk says M2 is good, so does Samsung NVMe utility
Ran memtest, no errors
CPU heat is fine according to CoreTemp
Can run Prime95 no problem (though I only let it go for about an hour)
Various DISM and system file checker stuff
Windows and all drivers fully updated, no apparent errors, no weirdness in Device Manager
Downgraded BIOS to a June or July (I forget) 2019 build, ab rather than abba
Manually set CPU voltage to 1.35v
Clean install of Windows (first try was crashing a lot; second try much less frequent crashes)
Newest drivers, AMD chipset drivers
Would post minidump file but sometimes not a BSoD and a new install so not sure if it exists yet (going to go back to her office tonight to try to fix, will post a minidump if one was generated.
Why would it work well for a few months and then start dying?
Chances i should just RMA this CPU?
I'm out of my depths here. I'm not a true enthusiast. I can put together a system and get it past POST and install Windows, but after that I really start struggling. So please assume that I have the capabilities of an intelligent child in the replies.
First thing I would do is disable "Fast Startup" in Windows 10. Google how. It caches old data and can often be the source of instability. Then re-install the latest chipset driver.
Honestly it is almost never the CPU that goes bad, not that it can't be, but it is rare. Usually however it is far more likely a power supply, motherboard or a memory stick has gone bad.
I would load the latest bios which should be 7C02v36. It was released the end of last month. It runs stable for me and I also have a 3600 and B450 tomahawk. I don't use the stock cooler as I felt it ran too warm with it. I use a 212 Evo.
The 7c02v35 bios from last November was also very stable for me. Make sure to reset the bios defaults after installing the bios. Leave everything at defaults until you know it is stable. I would load Ryzen Master. You can check that your temps are not an issue.
Make sure your Ram is on the QVL list for your board. If not then your board may not be setting the correct timings. Contact the Ram's maker and ask what timings should be used on your setup and you can manually set them if needed.
If that does not help then try running one stick of ram at a time and see if that helps the stability. You can isolate a bad stick like this.
If you then still have issues you might want to talk to MSi about what is going on and see if they have advice.
perhaps you have received some update for Windows 10, many people have the same issue.
"Picked up by the always-excellent BleepingComputer, Microsoft’s new KB4556799 Windows 10 update is causing a myriad of problems for users, including Blue Screen of Death (BSOD) crashes, deleted data, performance issues, broken audio and more. And users are not happy."
Same issue here, ram and psu checked.
Memtest coulnd't pass rolled back Bios before 188.8.131.52abba and then memtest was succesfull but still windows crash at idle.
I tried a lot of thing, manual overclocking, enable amd cool n quiet, reinstalling windows, trying linux, trying with another ram stick, etc. Windows log a Kernel critical error 41.
I think I am going to RMA the cpu
I have a 3600+B450 pro carbon AC MSI + seasonics gold 650 PSU
Hi, it may be possible your particular setup needs a little more SOC voltage. At least in my case, i had to take mine up to around 1.13, as shown in Ryzen Master as "VDDCR SOC". Take a look at this:
Of course the post is from someone with a Asrock, but i also have that setting in my Gigabyte board, and maybe something similar in your bios. Of course this still means something isn't really right somewhere, but it will compensate and let you be stable, if that is the issue.
Yes it is what I think also. I just tried to increase Vcore but it didn't help.
I could try that but I think that I am going to return it. For long term, I don't know if tweaking the cpu just to get it working is reliable.
Very bad silicon lottery perhaps ?
had the same fukn issue for a month , even replaced almost all parts and still happend, at the end i randomly clicked oc mode in the ryzen master software ´( after i tried literaly every fix ) and it worked lol now my cpu core voltage is at 1.09 at permanantly and no crashes