System shutdown during very specific event

[Trypt]

Hiya! I know this is probably not the forum, but here it is anyway.

System is in the sig. I have no problems at all with games, with temps or performance. Today I started playing Control, in DX12 mode. I made it like 10 mins in, and in a certain place with a big baddie, the computer just shut down, I mean completely powered off, and didn't come back on, had to turn it back on.

I tried again,checked temps (low), checked CPU usage (low), but again, it crashed by just shutting down, instantly, no warning, no game crash, just shut down completely.

So, I noticed that the game comes in both DX11 and DX12 modes and that I was playing the 12 mode, so I tried the other, and of course it's perfect now, been playing for a couple hours (CPU is barely used for the game, GPU is at way lower temp and power consumption than while mining so not that). Also, all voltages from the PSU are normal and it's only used like 50%, so not PSU. I was worried about all that before I figured out the DX12 thing which is what I now think it is since I eliminated everything hardware related and tested the PC with P95, Cinebench, AIDA64.

Just wondering, anything I can do? Maybe re-install DX completely? How do I do that? How about BIOS update, I'm still using the bios that came with the mobo, which was before the release of the actual Ryzen2 cpu.

 

[wingman99]

If it just shut down, like the power going off in the house? The I would say there was a over load switch switching on, powering off the system. There could be a short circuit or short in a switch or wire fan also a PSU failure. Look for burn marks, hot wires, short circuit on the back of the motherboard to the case. Video card failure or over loading the PSU tripping the shut off overload switch.

 

[Trypt]

Um, no, nothing like that,

Anyway, it happened in DX11 as well, after a couple hours, in a boss fight again, I'm starting to think it's the game. It hasn't happened since while mining and watching a movie. I will update bios tomorrow, reinstall DX and see what's up, and play a completely different game for a long while to see if this is a Control only problem. I do like the game tho so I hope I fix it.

 

[wingman99]

Did it power off quickly without shutting down windows first?

 

[Trypt]

No, it just turned off instantly, playing the game, and everything goes quiet and dark, without restarting. I have to power on manually and everything is fine again. This only happens in this game, so it really doesn't make sense that it's a power problem, although I will monitor and try all kinds of things just to make sure.

My system is a 24/7 system, sometimes I don't restart for a month and I never have problems, it's mining always in background except during gaming, and it's my main entertainment center, I would think I'd notice something if it was a hardware problem.

What you think?

Any particular hardware monitoring I could do in the background to see if there are any problems or fluctuations? HWMonitor says all voltages are perfect.

 

[wingman99]

What you describe is not a software problem, sounds like graphics card hardware or power shutting down the PC. Try running graphics card power and stress test FurMark and see if it shuts down the PC, LINK: https://geeks3d.com/furmark/

 

[Mr.Scott]

Instant power cut is only caused by power or temp. It's hardware related for sure. Start testing.
PSU and vid card.

 

[Bill_Bright]

If you can run all day long in DX11 mode with no problems, I don't see this being a PSU problem - especially since P95, Cinebench, and AIDA64 run fine too, and you verified your temps are good.

If me, I would make sure your graphics drivers are updated.

How did you check your voltages?

 

[Trypt]

I'm checking voltages with HWMonitor, I know it's not ideal but they stay perfect. I just ran all night with mining and P95 running full tilt, it's still going now, no problems, no fluctuations in Voltage, in fact, the min/max is so close as to within margin of error.

Also, I played Far Cry 5 for hours and hours which stressed my vid card and cpu more than Control, easily (higher temp for both, higher usage for both vid/cpu), and again, no problem.

I know it sounds like a HW issue but I've never seen it outside this game, and after testing for a day, still good. I'll try running some other graphics intensive tests and see again.

 

[Bill_Bright]

I know it sounds like a HW issue but I've never seen it outside this game

If it only happens with one game and other games and other programs work fine, I sure am not convinced its a hardware issue. It could be a hardware driver issue, however.

Did you check your graphics card's web site to see if new drivers are out there?

Did you run dxdiag?

 

[Trypt]

I just ran FurMark like someone suggested, everything is fine and no fluctuations in PSU voltages at all. I'll run dxdiag and also update graphics but again, not sure what that'll do.

Is it not possible that BIOS may be responsible? I have version 0411, came out before the Ryzen cpu release, since then I see a whole bunch of BIOS releases that say "improves system stability". The reason I haven't updated is cuz the system worked perfectly, although I have d/led the latest ABBA and will probably do that today anyway.

 

[Bill_Bright]

Is it not possible that BIOS may be responsible?

It seems like you are looking to blame your computer. If all those other programs work fine, and you did every check of your hardware you could and they revealed no issues, but the problem occurs when you play one specific game, I would looking at that game.

 

[Trypt]

Yeah, I agree, however I have gone deep into event viewer and found this:

A corrected hardware error has occurred.

Component: PCI Express Endpoint
Error Source: Advanced Error Reporting (PCI Express)

Busevice:Function: 0x1:0x0:0x0
Vendor IDevice ID: 0x1CC1:0x8201
Class Code: 0x8802

The details view of this entry contains further information.


In fact, I find these all over event viewer, it's a WARNING, not an ERROR, and not much online about this particular error. I never noticed it before, and can't be completely sure it happened during shutdown, but I found the unexpected shut down in event viewer as well, telling me the time, and while it doesn't actually give an error, around that time there is also the above warning.



On one of the three shutdowns I had, this occured exactly at the time system says it shut down:

ERROR

The CldFlt service failed to start due to the following error:
The request is not supported.

but only on one of the three shutdowns, the other two say nothing around the time the error says it was an unexpected shutdown. HWMonitor has been opened for a long time now, for min/max reasons.

- - - Auto-Merged Double Post - - -

Here is the HWmonitor info, minus the temps as I couldn't get it all in the screenshot, but they are 48C min and 70C max for the Ryzen, totally normal. I left everything else showing.
View attachment 207601

Another weird thing is the Radeon driver, the AMD driver. I have a driver that is NEWER than the upgrade recommends, and even NEWER than the optional driver, what is going on here? Here is the screenshot:

 

[trents]

Yes, very weird. Try rolling back the GPU driver. Do a complete wipe of the existing driver first, however.

 

[Trypt]

I just updated BIOS to the ABBA 1201 version, the latest from about a month ago. At least it fixed the TRC value for ram being 75 instead of 51, and it works. We'll see how it goes now, I'll also uninstall the amd driver, but probably I'll just avoid playing Control for now, test everything else, play a different game today, then if I feel brave I'll try Control again, or another game in DX12 (although as I said, Control did end up causing system shutdown even in DX11 mode, but after 2 hours of straight playing, again against a boss fight, coincidence probably but who knows, the 1st boss also crashed the game twice, but in DX12 only).

Nevermind, the Row Cycle Time is still 75, really annoying, it should be 18+36 for 54 and yet after months of bios updates the XMP profile still doesn't do that properly.

 

[trents]

I was going to suggest running memtest86 overnight to test the RAM. If the timings aren't right that could well explain the shutdowns.

 

[Trypt]

I highly doubt it, every Ryzen x570 system has Row Cycle Time set at 75 on DOCH profiles, it's something Ryzen recommends, although after testing by YTers, you can gain 1% performance by setting it to whatever the ram is actually rated at, RAS + tRAS I believe, so 54 in my case. My ram is 16-18-18-36-54 but xmp profile, like in all Ryzen cases in x570 chipsets, sets it to 16-18-18-36-75

 

[trents]

So is this a RAM incompatibility issue? Was your RAM on the QVL list?

 

[EarthDog]

The symptoms don't match what is going on...bleh....almost as bad as intermittent problems (on anything, lol).

If it only happens in one game, I'm betting it's something with that title or the drivers (psu may be it too). It is a new game. Have you tried to look up this issue on the web to see if anyone else is seeing the same thing by chance?


Edit: I'd also refrain from running furmark. It's a power virus (says amd and nvidia). Nvidia throttles the hell out of card and lowers clocks close to the BASE to compensate. What's funny is because of that throttling, it manages to use only a couple percent more power. Loop unigine heaven or similar for a better gauge of real world clocks and loads... instead of choked throttling.

 

[Trypt]

Yeah, I wondered what was up with furmark, it ran my gpu at at least 10% underclock.

- - - Auto-Merged Double Post - - -

Anyway, I did a BIOS update and rolled back the AMD driver to the last official one. The new one I had I got from the AMD chipset driver, it is newer but somehow it doesn't seem polished, and you can't get it at all from the Radeon update or even on the AMD Radeon driver site, only as a AMD chipset.

Well, did more testing, Novabench, 3DMARK, no problems.

Also, been playing Assassins Creed Odyssey for over 4 hours straight now, temps normal, 30% CPU 100% GPU usage, this game takes way more resources than Control does, and it's perfectly fine, no problems.