[solved] Computer crashing & hard-resetting all over the place (was: Huge issues with RX 5700 XT, now with R9 270X too)

[velis]

Update: the new install is much more stable. I've been running the test game for two hours without an issue. However, I've seen the PC reboot spontaneously from screen turned off twice now. Also one BSOD. Power settings are set to turn screen off after 10 minutes and never sleep.

While stability has improved, I still think there's something seriously wrong with this SSD. Swapping for the old one now to see if I can confirm that.

 

[Shmee]

Hmm sounds like you may have a bad SSD. What did the BSOD say?

 

[Kenmitch]

While stability has improved, I still think there's something seriously wrong with this SSD. Swapping for the old one now to see if I can confirm that.

Did you install Samsung Magician and update the firmware on it yet? I'd probably try that and see how it goes.

 

[velis]

Did you install Samsung Magician and update the firmware on it yet? I'd probably try that and see how it goes.

Oh, I totally forgot that can be done too. Trying it this afternoon. I also found a NVME driver on the download page, so trying that one out too.

 

[velis]

Hmm sounds like you may have a bad SSD. What did the BSOD say?

Sorry, was doing something unrelated at the time, so I didn't remember it, but it was totally storage / GFX unrelated. As were the ones in the previous install.

 

[velis]

Well, I swapped the old SSD back in. Got a couple of hard resets in the first half-hour. Very helpful to disprove the faulty SSD theory. In fact, while typing this message, I had another two. WTH?

To describe: usually, the screen will just turn black, fans will go crazy both on CPU & GFX card, then a few seconds later, there will be a hard reset and a boot. Sometimes no hard reset, sometimes a BSOD, sometimes no fans. It happens under any mode of operation, even idle with monitors off.

However, I'm beginning to think that my previous issues with games crashing - which are no longer happening since reinstall - have been a totally different issue to these hard reboots which now seem to be going on everywhere. TBH, the new SSD "only" had them twice a day, the old one just gave me a shocker now.

I'm running on a spinster now, to check if it might be just the M.2 interface on the MB.

 

[Captante]

I would check PSU as mentioned, though it could also be a RAM/CPU issue, try running memtest.

My guess is PSU as well ... draw from the wall tells you nothing about whats being delivered to your GPU/system.

Your 560 watt unit was below ideal spec for the 5700 out of the box and PSU output only drops over time.

 

[velis]

My guess is PSU as well ... draw from the wall tells you nothing about whats being delivered to your GPU/system.

Your 560 watt unit was below ideal spec for the 5700 out of the box and PSU output only drops over time.

OK, I'll bite. Just ordered a Corsair RM650x. Will be here in a week. Will see tonight how the comp behaves with a SATA SSD.

 

[velis]

M.2 slot theory disproved: having just the spinster in the machine, I just had two reboots.
I'm now copying the system to a SATA SSD just to expedite the boots & testing. Will swap the GFX card next.

 

[EliteRetard]

Well, I swapped the old SSD back in. Got a couple of hard resets in the first half-hour. Very helpful to disprove the faulty SSD theory. In fact, while typing this message, I had another two. WTH?

To describe: usually, the screen will just turn black, fans will go crazy both on CPU & GFX card, then a few seconds later, there will be a hard reset and a boot. Sometimes no hard reset, sometimes a BSOD, sometimes no fans. It happens under any mode of operation, even idle with monitors off.

However, I'm beginning to think that my previous issues with games crashing - which are no longer happening since reinstall - have been a totally different issue to these hard reboots which now seem to be going on everywhere. TBH, the new SSD "only" had them twice a day, the old one just gave me a shocker now.

I'm running on a spinster now, to check if it might be just the M.2 interface on the MB.

I'd put faulty RAM slightly above faulty PSU...just because the issue happens at any time under any load.
Problem also seems to be getting worse, which could be increasing failure and software error buildup.

 

[Kenmitch]

I'd put faulty RAM slightly above faulty PSU...just because the issue happens at any time under any load.
Problem also seems to be getting worse, which could be increasing failure and software error buildup.

That's a possibility. My son's rig was having similar issues with the latest uEFI version for his motherboard when I built his rig. It was somewhat boggling, but reverting back to the uEFI version I was using when I tested out the combo cured all his issues.

I didn't see any mention of ram testing, single stick, relaxed timings, etc.

 

[Smoove910]

Jeezus, to think you are throwing parts at a known issue with AMD drivers as of late. Do a little research rather than spending $$ on replacement parts! There's literally 1900+ responses to the single thread on AMD's site alone.

AMD Community | AMD

community.amd.com

I've read where uninstalling any 3rd party software like MSI Afterburner or Wattman helps. Also you may want to turn OFF 'enhanced sync' if you have it.

 

[velis]

@Smoove910 : My issues are nothing like that. My PC hard reboots out of the blue. Not in a game either.

Anyway, ran a round of memtest86: came out clean. So it doesn't seem to be RAM.
Also moved to Ubuntu, had one reset right when I first got in, but nothing after that. Though it is also true I wasn't doing much with the computer either. I just started copying a few TB of data to a new backup disk. It's almost done now. No reboots during the night.

Anyway, moving to Ubuntu and having a crash in there too seems to drastically reduce the possibility that it would be the drivers.
It looks like it supports the PSU theory, though it seems ridiculous that PC would reboot when the GFX card is not loaded. That would indicate a general PSU failure. Should that be the case, great timing! Will see on Wednesday...

 

[Captante]

Anyway, ran a round of memtest86: came out clean. So it doesn't seem to be RAM

Although my money is still on the PSU, it can't hurt to rule out a memory error. If using memtest86 however you need to let it run for quite awhile to be sure... I suggest setting it up looping overnight.

 

[velis]

I have observed another symptom: it seems that OS install USBs that I create come out corrupt. Initially I thought it was a one-off when Ubuntu desktop failed to install. It was easy to solve by placing a network install ISO onto the stick and then it just downloaded itself while installing, but the full image failed stating that some files were not found on the stick.
Now I try to reinstall Windows - and it too fails with identical message. I'm going to re-download the image afternoon, reflash and see if it stops at the same percentage again. Highly doubtful.

 

[Shmee]

That sounds like flaky RAM then. Run memtest86 for 4 passes. Use UEFI mode if possible.

 

[Captante]

That sounds like flaky RAM then. Run memtest86 for 4 passes. Use UEFI mode if possible.

Agreed ... bad RAM can manifest itself in strange/unexpected ways.

Have to say though I've had RAM turn up bad after passing more then 4 memtest runs a few times. I wouldn't trust the results of running it looping for less then minimum a few hours.

The other utility I've found useful for this sort of thing is
OCCT.

 

[velis]

I've had a really bad bout of crashes today. So I pulled out gfx card, inserted the NV back in. Same deal. 5700XT back in. Then I pulled ONE RAM stick out. Not a failure since ?!?!?!
I think I'm now SWAPPING the RAM sticks

Edit: This RAM (if really the culprit) must be really shoddy - I already had to replace one stick because it came to me faulty. Not fully dysfunctional, but causing all types of havoc. Just a lot more apparent as this round is...

Edit2: the bout of crashes I had were BSODs saying that storage failed - something about a HW interrupt not being processed - I never take a pic, stupid!

 

[Stuka87]

I've had a really bad bout of crashes today. So I pulled out gfx card, inserted the NV back in. Same deal. 5700XT back in. Then I pulled ONE RAM stick out. Not a failure since ?!?!?!
I think I'm now SWAPPING the RAM sticks

Edit: This RAM (if really the culprit) must be really shoddy - I already had to replace one stick because it came to me faulty. Not fully dysfunctional, but causing all types of havoc. Just a lot more apparent as this round is...

Edit2: the bout of crashes I had were BSODs saying that storage failed - something about a HW interrupt not being processed - I never take a pic, stupid!

You can open up the crash dumps with a BlueScreen reader. You should have at least several of those in the minidump folder.

 

[Chicken76]

@OP: I would not completely rule out the motherboard either. When I built my Ryzen system last year, I went with a cheap B450 board, because I don't overclock and bought a 65W CPU. It gave me similar crashes to yours and the problem was degenerative (took less and less time between crashes). The mobo was the last thing I suspected, Murphy's law I guess. I memtested the ram, no errors. Memtested the ram on a different computer, no errors. Changed the GPU, similar crashes. Changed the PSU, crashed all the same. Tried Windows, Linux booting from USB, both crashed. Sent the mobo back as defective. Seller confirmed it was broken and sent me a replacement. No problems since.

 

[Kenmitch]

I would not completely rule out the motherboard either.

As well as the uEFI version the MB is running. I've tested some uEFI versions that wouldn't even play nicely with b-die.

 

[velis]

As well as the uEFI version the MB is running. I've tested some uEFI versions that wouldn't even play nicely with b-die.

UEFI version? I can influence that when I flash latest BIOS? Afraid I'm getting too old for this

Update: 9 passes of MemTest86 in UEFI mode without a single issue (both sticks). PSU just delivered by the postman. Thinking of asking my friend to loan me his MB (he's building a Ryzen system this week).

 

[Shmee]

I would install the new PSU first and see if that fixes it. You said you already changed motherboard and CPU to the 2700X and B450 from your old intel system, so I doubt that it is related to those now, if you had the same issue on the Intel system. What was the Intel CPU and motherboard btw, and was it using the same RAM?

 

[velis]

Thanks, yes, I will definitely do that. However, I think I may have established that the issues changed, I just didn't know it originally because everything was crashing all over the place. Please see #31 on this page.

Intel CPU was the i7-6700K, MB was ASUS Z170M-E D3 with DDR3 RAM.

But, as I said, the original issues seem to be gone. Those were more games crashing with reports that seemed to indicate that the GFX card simply vanished mid-game. Those issues were gone simply by inserting the old SSD (which still had the year-old Windows install) vs the new SSD which had a clean, but corrupt install (done on the Intel MB though).

My current issues are different: screen goes blank, disk activity (crashed driver?), then maybe a hard reset (or I have to do one because the computer is non-responsive anyway).

 

[velis]

Second day with the new PSU and without a single crash. I'm getting my hopes up!