Question Newer AMD System Randomly Crashing

[rkoenn]

I built a friend an AMD Ryaen 2700 CPU, AsRock B450 motherboard, 2 8GB Viper memory sticks, an XPG 512 GB NVME SSD, and an XFX RX-560 video card in early 2020. He used it fine except for one time when it refused to boot and didn't recognize the NVME SSD. I told him to remove it and clean the contacts and after that it ran and has always booted ever since. However a couple of months ago he started getting random crashes. On some crashes he got a Windows error message that indicating an AMD driver timeout problem. He gets these randomly a couple of times a day and ALWAYS when on the internet browsing. Two weeks ago I did a clean install of Windows with the latest build and had no problems doing that at all. I then ran a stress test for an hour and it passed. I also ran a heavy 3D graphics test which ran and I verified the video card cooling fan cycled when the card got hot. On one of those I got a crash with an error message. He bought a second identical card, which he has returned, and the symptoms were the same. When I installed the clean windows I also installed AMD's latest video and motherboard drivers. Does anyone have thoughts on what is causing this and what might be done to stop it? I am wondering if the video driver updates may have started it and an earlier driver might stop it although that seems to be going backwards. Any help is greatly appreciated.

 

[Belegost]

I would highly recommend running memtest86+, I ran into similar crashes that seemed to point at the video driver, but, after a good week of debugging, an overnight pass of memtest found a set of bad addresses on one stick of RAM. After replacing that things stabilized fully, I can only suspect the video driver was pointed to because it was reading bad data from those addresses.

 

[rkoenn]

I would highly recommend running memtest86+, I ran into similar crashes that seemed to point at the video driver, but, after a good week of debugging, an overnight pass of memtest found a set of bad addresses on one stick of RAM. After replacing that things stabilized fully, I can only suspect the video driver was pointed to because it was reading bad data from those addresses.

Thanks, that has definitely crossed my mind but I think he used the machine removing one stick at a time and still had the problem. I just asked him to try again as a memory stick problem would definitely cause such problems. I built a high end system for a friend some years back with 8 sticks of RAM. Suddenly after 3 or so years it started crashing. It was used to do heavy duty engineering number crunching and would run a simulation overnight due to the length of the run. One day he came in and it was crashed. It started repeating and it hadn't before. I tested memory two sticks at a time with memtest and finally on the last two it found a problem. Then I isolated it to one stick. The memory had a lifetime warranty so whoever it was, a big RAM maker, replaced it for free. I should have run this when I had the machine a couple of weeks ago.

 

[CuriousMike]

Power supply ?

 

[rkoenn]

Power supply ?

It could cause some type problems but my experience is that they usually fail or don't fail. A flaky voltage would cause problems but I don't think they would manifest in a driver time out error message. Thanks for thinking of it. I've built 500+ computers starting around 1985 and power supplies are the number one failure but never this way. Number two is DVD drives. Thinking back I have rarely ever had a MB or video card fail, maybe once or twice. I have had a very rare memory stick failure but typically those are repeatable and easy to figure out but not all have been.

 

[CuriousMike]

Take his video card and memory and jamb them in your system and run it for a day.
See if it fails the same way.

 

[itsmydamnation]

I had what i thought was video card issues , ended up being needing more agressive CPU vdroop / load line calibration. CPU was fine on CPU only heavy benchmarks prime small FTT etc , Only when using GPU did it happen.

its easy and worth a shot

 

[Leeea]

if it is over clocked, un-overclock it

if it is xmped (which is overclocking fyi), un-xmp it.

sounds like a ram problem, memtest86 it overnight.

 

[rkoenn]

if it is over clocked, un-overclock it

if it is xmped (which is overclocking fyi), un-xmp it.

sounds like a ram problem, memtest86 it overnight.

As far as I know it is not overclocked. It is a very low stress machine just used by an older gentleman to basically do some paperwork and browse the internet. My biggest problem is I used to live one mile from him and working on it was easy. These days since I moved we are 140 miles apart and he isn't too hardware savvy so I have been trying remote troubleshooting except for when he brought it over 3 weeks back. Thanks for those suggestions. I believe he ran memtest after I told him to try that with no problems. I have resolved issues twice on high end systems I built that had a memory stick, one of eight, go bad after 3 or years and Memtest isolated it to one stick.

 

[GrumpyMan]

Since a video card never runs at the same wattage, when gaming the more power it requires I could see it trigger or just stress a bad power supply with bad capacitors. I would try a different power supply to at least rule that out as a possible cause in your power delivery system on your mobo as a start in the diagnosing process. Also check out the caps on your mobo, rare, but you never know.

 

[rkoenn]

Since a video card never runs at the same wattage, when gaming the more power it requires I could see it trigger or just stress a bad power supply with bad capacitors. I would try a different power supply to at least rule that out as a possible cause in your power delivery system on your mobo as a start in the diagnosing process. Also check out the caps on your mobo, rare, but you never know.

Over my 30+ years building and servicing computers I don't recall ever having a power supply cause spurious problems although I know it could and respect your suggestion. They are the number one failure component from my experience with CD players number two. But it has always been a case of complete failure or no problem. That said it is something I will try if he sends me the box to troubleshoot. My first thoughts are maybe a BIOS update but research has not shown that is a complaint from owners of this MB. I am really stumped but will continue to troubleshoot it. Also he is not a gamer and this always seem to happen when on the internet browsing, I am wondering about possibly a network driver problem.

 

[kschendel]

Rubber chicken waving: is power supply idle current control set to typical current idle in the BIOS? If not, try that.