So, long story short, my pc build which has been mostly rock solid for years has now started to crash doing just simple stuff like listening to music or watching videos.I thought the problem was my rather old Radeon R9 290x, but I already tried an old Nvidia 560 Ti and it also crashed.
I checked the event log and saw a display driver crash, but I don’t know if that really tells you anything, since the pc crashes on the other graphics card too. It crashes in a weird way, no bluescreen or anything. On the AMD graphics card, it will still keep playing my music for a bit even after the screen freezes and goes black, but the music stops after a minute or so. Interestingly, on the Nvidia graphics it just hard crashes into a black screen and reboots, still no bluescreens.
I’ve already tried reinstalling WIndows, reinstalling graphics drivers, cleaning the pc and reseating the graphics card. So far, nothing has worked! And I’m nearly out of brainpower to solve this issue.
My PSU is also just a couple years old, Seasonic Focus+ Platinum 750w, so I don’t think that it would be the issue, could be wrong tho. Any suggestions are very welcome at this point!
My full specs: Ryzen 5 3600, Asus x370 Prime Pro motherboard, 16Gigs Gskill 3000mhz RAM, Radeon R9 290x, Seasonic Focus+ Platinum 750w, 3 SSD’s (Samsung 970 EVO, 860 EVO, 840 EVO), SoundBlasterX AE-5 Plus soundcard
RAM? You don’t mention reseating the RAM, or running a memory test overnight. Once I had success cleaning the RAM contacts with a pencil eraser (carefully ensuring no rubber bits were left on the RAM). If the RAM is not all on one stick, you could try running with one stick removed.
Memtest86 is indeed a good idea. Another avenue to test is using Linux. Boot a LiveCD* (they’re also bootable from USB sticks ) and let it run for a while. If it crashes/freezes it’s a hardware issue, if not you need to re-evaluate your installed programs on Win-OS.
*Linux LiveCD’s do NOT touch your Win-OS install. There are several options, but for new users Knoppix ( ← link) is the best entry IMO. Mind that the GUI is different to what you’ve been used to on Win-OS, so take some time to get acquainted.
Probably the wrong direction, but when I’ve had issues like this, I’ve found the non OS SSD/HDD is to blame. May be pull those and try that? … um, unless you’re RAID’ing
Definetely RAM, as jlittle said. I’ve had a similar issue with two sets of G.Skill Trident Z RGB Neo that would crash the Nvidia driver, not to a black screen but in a recoverable status. Call up G.Skill and ask them for a new kit. They have a lifetime warranty on their kits so it’s worth a try. You’ll just need to pay for one way shipping, that’s about it (beside the needed wait, which is three/four weeks).
I’ll definitely try reseating the RAM and running a memtest overnight. I have ran a memory test before, as I tried to run the memory at 3200mhz. Now I have the memory set at the XMP profile which is 3000mhz CL14. I’ll report back when I have ran the memtest.
Since you mention that this is a problem that has recently developed on a system that has been stable for a long time, I’d investigate the cooling on your chassis and CPU. Just make sure all the fans are running and thermal paste is in order. If you have an AIO it might be time for a flush and cleaning if possible.
So, I ran the memtest for about 2 hours and it froze up, didn’t show any error though. The memtest was a bootable one so that would probably rule out software errors etc. I guess I’ll try the memtest again, with just one RAM stick and I’ll unplug any unnecessary USB devices.
I did clean the PC and I have checked the fans and temps, nothing seems to be overheating, the thermal paste on the CPU is not more than 2 years old.
as you have an amd card check eufi/bios and enable hpet if its an option.
enable it if its off.
if there is no option for hpet. boot windows open and elevated CMD>
and type
bcdedit /set useplatformclock true
to enable windows hpet.
bcdedit /deletevalue useplatformclock
to disable hpet.
hpet is a hardware clock that amd uses with its drivers. if its disabled your system will randomly lock up, throw up a banded colour screen or just black, and repeating audio.
one other thing. you can lose performance on some applications with hpet enabled. but i found the stability was more valuable than the extra 10fps when i was already getting more than the monitor i had, could display. so wasnt really an issue for me.
have you run sfc /scannow
or
DISM /Online /Cleanup-Image /CheckHealth
DISM /Online /Cleanup-Image /RestoreHealth
if your memory is unstable your o.s will be full of write errors. which can lead to random crashes and lockups.
with the software checks you test the ram and ssd for write errors
if you find em, then you will need to adjust your ram timings.
with hpet on you will know within 2 days if your stable.
still crashing?.
then the psu or power delivery on the motherboard would be my next stop.
can you test with another psu?.
if so do so.
I’d also suggest that it’s the RAM, I had the very same issue with my G.Skill Samsung B-Die kit, it worked with my first Threadripper 1900X but refused to work stable with my 3960X Threadripper with the problems you’re describing, on any speed or timing settings. I could run memtest as much as I would like and never got an error but it still froze.
I don’t know what it is with Zen 2 that memory compatibility is so much worse than Zen 1, I’m on my third memory kit now and fingers crossed that it works.
Alright, so I pretty much unplugged everything inside the pc and plugged everything back in. I ran a memtest and it passed without any problems. I am now back in Windows, and I’ll see if the pc keeps crashing, but so far so good!
I’ll probably run a longer memtest in Windows, since the bootable memtest only lasts for about 6 hours, and I want to make sure my RAM has absolutely no issues.
If you want to make sure then run HCI Memtest until at least 2000% maybe even 4000% and not just one instance but so many that all your RAM -1GB is utilized. With 32GB RAM that may take about a day to 2000%. So yours with 16GB that may take less.
I ran the HCI memtest and my pc froze again. Looking at the event log, the display driver crashed and recovered multiple times and finally the pc froze. I am running the RAM with XMP enabled so I’ll disable it and try the memtest again.
Just a little update: my pc froze once again running the memtest, this time RAM was at stock settings so no XMP or anything. I contacted GSkill support, hopefully they will let me RMA the RAM. If new RAM doesn’t help, the PSU would be next, luckily my PSU has plenty of warranty left.
Just for anyone wondering how this story ended: I ended up contacting the retailer for an RMA, they accepted and gave me free return shipping. Originally they wanted to send me an identical memory kit back but I asked for a newer one that was in stock, as the old one was not in stock and had no estimated shipping date.
I ended up getting a 32 Gig 3600mhz CL16 kit for free basically (I bought the old one when memory prices were very high and they gave me a refund basically). And no issues so far! I ran a memtest overnight and everything seems to work just fine. Also, the new memory kit runs at 3600mhz just fine, while the previous would struggle at 3200mhz. So I’m a very happy camper!
) and let it run for a while. If it crashes/freezes it’s a hardware issue, if not you need to re-evaluate your installed programs on Win-OS.