PC Graphics

[dexxed]

Describe your system:

• AMD Graphics Card
  • MSI rx 5700xt Mech (3 months old)
• Desktop or Laptop System
  • Desktop (custom)
• Operating System
  • Windows 10 Home 64bit
• Driver version installed
  • Radeon Software 20.10.1
• Display Devices
  • BenQ gl2450HM 60Hz, HDMI, 1920 x 1080 (4 years old)
  • Rift S, 80Hz, DP and USB 3.0, 2560 x 1440 (1 year old)
• Motherboard + Bios Revision
  • Asus ROG Strix B450-f gaming, bios ver. 3103 (1 year old)
• CPU/APU
  • Ryzen 5 2600 (1 year old)
• Power Supply Unit  Make, Model & Wattage
  • EVGA SUPERNOVA 550W (4 years old)
• RAM
  • 32GB corsair lpx 3200 (running at 2933) (4x8G sticks, with two being installed recently) (2x8GB @ 1 year old) (2x8GB @ 2 weeks old)

Describe your issue:

• My computer, during idle or under load, will suddenly and with no forewarning have the screen go black or flash green then go black, along with all power to peripherals being lost. The only way to use my computer after that is to force-power-off with the power button. Seems to happen at random, happened once idling at my desktop when I was in the other room, and has happened twice while playing VR (Rift S) on multiple different apps. No significant errors in Event Viewer.

[pokester]

This behavior is the graphics driver reseting. Most often this is power related either a card or power supply failure. 

The obvious place to start looking in your case is your power supply. 

The 5700xt has a minimum 600 watt requirement and recommended 700 watt. 

In reality if you want to know that your power delivery is not the issue, you will want to go with the recommended spec or better off 700 watts. At 550 your power supply is underpowered to drive this GPU. 

This is likely your issue. I would suggest a quality 750 watt gold plus or better power supply. 

The specs for the 5700xt

https://www.amd.com/system/files/documents/radeon-rx-5700-series-quick-start-guide-en.pdf 

[dexxed]

I looked at the spec sheet a bit after making this post, when I bought the card I remember seeing the wattage requirement somewhere, and I wasn't having issues before, but after upgrading to two new SSD's and more ram as well as a better cooler, I think it might have just put it over the edge. after running ram tests, as well as testing to see when the issue occured, it made me think it was the cooler. This article is the result I most likely drew from when I first bought the card, as well as a lot of the top results on a quick google search showing that 550W would be enough: MSI Radeon RX 5700 XT Gaming X review - Hardware setup | Power consumption. In retrospect I shoulf have just gone with the manufacturer's specifications, however, I think that it is the age of my PSU that perpetuated this issue, being a 4 year old model that was used for a lot of hours daily. I also ran OCCT's GPU, CPU, and PSU tests, with the PSU test being the only one to cause a crash, making me think it is the combined load of the cpu and gpu that causes it to crash, leading towards either a faulty psu or motherboard. I have already ordered a Corsair RM750x, which will arrive between the 30th and the 4th.

[pokester]

Good choice, that should be a nice power supply. Power supplies are like shock absorbers on a car, they begin degrading day one. So you are correct that adding even one more thing drawing power or the PS degrading just a little more can push it over the edge.  Reply back if that new PS takes care of things. 

[dexxed]

Got it in and wired, and ran the OCCT PSU test, which always crashed at 15~20 mins before, ran fine for a whole hour. However, just a few minutes ago as of the time posting this (Oct. 30th, 1:36 am PST) It crashed after not having done anything for hours, all that I was doing was installing a steam game to my other SSD. Long green-screen (5-15 seconds) then the same as before.

[mstfbsrn980]

PSU problems are most likely not causing the green screen. For solving the green screen problem, two same type of connection ways (two HDMI or two DP usage) or two same resolutions etc... a configuration may be selected for the error.

[dexxed]

I don't understand what you mean, could you possibly describe it a little simpler or perhaps link me to an article that discusses this? I don't understand what you mean by two same type of connection ways. I have a HDMI going to my monitor and a DP to my Rift S when I'm using it, monitor runs at 60Hz, while the Rift runs at 80Hz.

[mstfbsrn980]

Do not use DP. Use two HDMI connections.

[dexxed]

I can't, Rift S only works through DP. it's the only connection available, since it needs to run at 80Hz and is not supported through HDMI. I also wasn't using the Rift through some of the crashes.

[mstfbsrn980]

OCCT has been suggested to you to find the cause of the crash. I wrote a suggestion because that DP usage can sometimes create a green screen. When using two monitors I encountered a green screen error before and suggested you find a suitable configuration. You can try to find this out by experimenting.

[dexxed]

I've been using OCCT to stress test it to find the issue. I crashed previously whenever doing the PSU test, as shown in my most-recent reply to pokester. I didn't have this issue using both my monitor and DP before installing the extra ram, new cooler, and SSD's. In OCCT I've ran the OCCT test, Memory, 3D, VRAM, and Power trying to figure out the cause of the issue before, with the only one having it crash being the Power test.

[mstfbsrn980]

EVGA SUPERNOVA 550W (4 years old)

This is the reason for the crashes. You should purchase a new and gold certified PSU of minimum 750w or higher. You can also use a single monitor to fix the green screen problem.

The reason I wrote you the first message was for you to resolve the green screen error only.

[dexxed]

As I listed in my original message, it's not crashing per say, but all peripherals (monitor, keyboard, mouse) stop working, the monitor appears either green or black, but my computer's still running. Computer appears to still be running, and nothing appears in Event Viewer except for the reports that I shut it down by using the power button (since I can't access my computer without forcefully shutting it down first after this). As I mentioned in my replies to pokester as well, I bought a Corsair RM 750x, which is installed currently and what the last "crash" was on. I am not using two monitors at the time of crashing, for the most recent and one before, that is.

[mstfbsrn980]

I think VR is causing this error. Does an error occur when you connect your box to only a good TV?

[dexxed]

I'm inclined to believe it's not VR that causes the error, given that I've used it with my pc before I got this graphics card for 8 months and with this graphics card for two months since getting the new graphics card. This only started happening once I installed the new hardware. On top of that, my hmd wasn't even plugged in during one of the crashes.

[mstfbsrn980]

Have you tried reducing the speed and the voltage a little with WattMan? Could the GPU be doing hardware errors?
-5% GPU frequency and 1162mV maximum voltage can be tried with WattMan settings.

[dexxed]

I will try doing that, but I no longer have a proper way to test whether or not it works anymore- given that my computer now runs occt's power test just fine. And I think I have to use Radeon Settings instead of Wattman.

[dexxed]

I'm going to try and do another OCCT Power test to see if it crashes again with the current configuration.

[dexxed]

Was able to run through a full 45 minutes with the current settings, everything the same as before. It could be two issues possibly with the same outcome. Old psu being the first issue and something else being the one causing it to crash at idle since It 100% of the time would have crashed at the 25 minute mark before. Another thing to note is I just installed the most recent chipset drivers- maybe that will do something. I plan on using DDU and reinstalling the drivers later tomorrow. Before that, I need some sleep though, thank you for attempting to help. Will report back if and when further issues arise.

[mstfbsrn980]

When you used the new PSU and new (my) WattMan settings, did the system crash after 45 minutes?

Edit: Please write the problem and the result clearly. Did it crash after 45 minutes? Has a black screen occurred? Is there a monitor signal? Which WattMan settings did you use? Have you increased the GPU power limit value? Every piece of information is very, very important to comment.

[dexxed]

I can't get it to crash at stock settings- I don't see how it would crash at -5% and not then. I will attempt your settings tomorrow though and report back. I can't stay awake for another 45, already trailing off as-is.

[mstfbsrn980]

As far as I understand, the OCCT power test goes well for 45 minutes. If so, you don't need to repeat this test any more. Do you have a black screen problem even though your system passed the OCCT test? Is there a monitor signal for a black screen?

Random black screen problem on desktop? How do you get the black screen doing? I did not read the messages you wrote to someone else, so if you write clearly, I can at least comment. Also have you tried the WattMan settings I mentioned? Are you getting improvement with this?

[dexxed]

I have had one crash since installing the new psu. I passed an hour-long OCCT psu test with no issues. At the time of the most recent crash, I was on my desktop, installing a steam game- with chrome open in the background. Green screen, no signal to monitor, no power to keyboard or mouse. Only way I was able to reliably produce this issue before was the OCCT power test, otherwise it's intermittent and random-however that does not crash my system anymore. I have not tried the wattman settings you mentioned, since I can't get it to crash as-is. If I can find a reliable way to get it to crash I will try those settings to see if it helps.

[mstfbsrn980]

I suggest you disable Chrome GPU hardware acceleration. Also, when you experience crashes with A GAME, you can try -5% GPU frequency and 1162mV maximum voltage with WattMan.

I think the problem is caused by Chrome and the driver is corrupted with this program. Have you tried Edge?

Note: Thanks for the clean information...

[dexxed]

It definitely could be chrome that's causing all of these issues. I now see that other people have been having problems with the hardware acceleration turned on when running games. I now recall that I have had both chrome and wallpaper engine(which I think runs as a game) open pretty much always as well, since I use both daily I never thought about them actively. Thank you for your help, I hope these fixes work.

[mstfbsrn980]

So... My goal is not suggestions, we need to find the cause of this error... I wonder if the driver is making an error with VR...

[satanael]

dexxed try this

RX 5000 Series Common Issues & Possible Fixes 

[dexxed]

Thank you for linking me to that, I have done all of them except 2, 6, and 11 I will try 3 right now. Though I now encountered a different issue that might be related? First time getting these error codes. This is also the first time it's rebooted on it's own after the crash/green screen.

Error Source: Machine Check Exception
Error Type: Bus/Interconnect Error

Processor APIC ID: 0

Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 2

Error Source: Machine Check Exception
Error Type: Cache Hierarchy Error
Processor APIC ID: 13

In chronological order. Happened while I was doing some light Blender work. I was also running google chrome, steam, spotify, radeon settings, and discord in the background.

I thought my issue might have been fixed since I was able to play VR for 2.5 hours last night after undervolting and lowering the frequency on my gpu, though now I'm not sure.

Edit: I also played games for 2 hours before the crash and had used my computer for the whole day before that. Temps were within range, usage wasn't super high.

[dexxed]

mstfbsrn980‌ any idea what causes these errors? I can't seem to recreate it with stress testing. Problem was similar to before, but caused a restart and logged this to event viewer. 

[mstfbsrn980]

If you are pasting these errors from where I guessed, there is an error caused by the processor. I can't see any GPU related error.

[dexxed]

I don't see how these errors wouldn't have come up in my near-year of having this cpu / motherboard combo though, and I didn't done anything that could injure the motherboard or cpu when installing my new parts.. I also found on Tom's Hardware Forums a thread stating that the error transferred to another machine with the gpu (it was a different gpu though, being the Sapphire HD 7770). Thread: https://forums.tomshardware.com/threads/is-my-video-card-beyond-repair.1812600/

I finally managed to test the video card on another PC and the same thing happens. As soon as I install the driver, the PC glitches.

Edit: Found another thread where this issue seemed to have stemmed from the gpu: https://www.techsupportforum.com/threads/my-pc-suddenly-restarts.1244268/page-4 

Also found this reddit thread where it was seemingly fixed by installing the driver without AMD software: https://www.reddit.com/r/AMDHelp/comments/hq7jcu/cache_hierarchy_error/

(sorry for being unable to link, otherwise it would need to be reviewed) 

[mstfbsrn980]

I wrote a response to you looking at the error message. So I'm not sure what the problem is.