Radeon Rx 5700 Xt Green Screen

This is not a support thread, I merely figured I would share what I have learned afterward searching around a lot. On my current computer I have an AMD gpu, specifically the Radeon RX 5700 XT, and there are some problems with information technology. It is rare, and extremely random, but sometimes during gaming my screen will only plow green and the figurer locks up. I tin’t practise annihilation and simply have to force reboot it. I can become for a week or more than without experiencing the event, or I might see it with just a couple of days in between. It has only happened while gaming so far.

At first I was worried I had a lacking card and I might need to go a new 1. But when I started searching for this effect, I realize that a LOT of people have the same problem, even on Windows. I have seen multiple suggestions on how to fix information technology, and I figured information technology would be best to try out one thing at a time.

For now, I take disabled Dynamic Power Management. I tin can’t really know if it worked because of the random nature of this result, merely if I don’t see this problem again for a couple of weeks and then I’ll presume that was the prepare. I will be sure to study dorsum here if I see the issue, or if I become for a couple of weeks without problem.

To disable DPM, you demand to add:
radeon.dpm=0
equally a kernel parameter to your bootloader.

I don’t think that is the trouble seeing that on Windows people went from crashing every time they launched a game, to now existence able to play without this happening much. In whatever case, internal temperatures in my chassis in the aforementioned as my room temperature since I got ii huge front end fans. I accept also been monitoring my ain temperatures.

This GPU has iii temperature sensors, “border”, “junction”, and “retentivity”. Junction is apparently some kind of hotspot so it is expected to be higher than the others. When I am playing the almost demanding games, all three temperatures stay well within spec. Edge temp will commonly be between 65 and 70 when playing demanding games, Junction volition go every bit high equally 95-98 (AMD has said that 110 is inside spec and expected for this sensor), and memory is usually somewhere betwixt lxxx and ninety.

The issues have also not surfaced while I have played some of the well-nigh demanding games. And so far, the game that seems to trigger this consequence the virtually in my case is Starcraft 2. I take also seen this result with War for the Overworld, but but once or twice. Doom, which I’ve played quite a lot lately never triggers this.

For now I have disabled Dynamic Power Management, which has been known to cause issues on Linux with newer AMD cards. If I don’t experience issues again for a couple of weeks then I tin probably assume that disabling DPM fixed information technology for me. Hopefully they will eventually fix the issue properly with newer drivers. As I said, they have the issue on the Mesa bugtracker, and it is a well known problem both on Windows and Linux.

But it really does not bother me that much, as I said in the original post, I can become for more than a week without seeing the problem.

Disabling DPM was non a solution, I just had another i of those green screens while playing War for the Overworld. Temperatures are normally effectually 70 degrees C in this game, which is far from overheating. I take a 1200W Corsair power supply, so I likewise doubt it would be a power issue. Perhaps I will just need to wait for the effect to hopefully exist solved in a future driver at some point.

I’ve been playing quite a lot for over a week without the trouble, and then it really is very random. Merely so far information technology has never triggered in the more enervating games. Both Doom (2016) and Doom Eternal have been running with all settings on maximum and no crashes, and I’ve played those games for hours.

Yeah at the moment I’m but hoping that this will be fixed with a newer commuter eventually. And there are no logs at all, the arrangement really difficult crashes. Sometimes I left it on for 10 minutes in this country, hoping that the log would still get written to the difficult drive, but no. The system log contains only normal stuff until it just cuts off when my organization freezes, and continues once again with the forced reboot.

It has to be a driver upshot, since Windows users have the exact same green screen on crash.

AMD’south firmware goofed on memory allocation – if you’ve got the option to gear up a specific memory window, east.g. 2GB or 4GB, set information technology to that, rather than to Automobile, and the problem should go away. The green screens seem to happen whenever information technology tries to allocate itself more than memory.

Thank you for the help, although I don’t seem to have the selection to set that in my bios. I am as well non familiar with any software for changing gpu settings for AMD cards. It does not happen often these days, actually in the last 3-4 weeks or so it has just been Starcraft 2 that triggered it, and only in one case or twice.

So, I’yard a petty late to the green screen party.

I simply got a rx 5600 xt and I was having these green screen issues (when gaming).

I downloaded the unigine sky benchmark and ran it a few times on ultra settings @ 1080p (vsync on, vsync off, tess on, tess off, aa, yous proper noun it). Every single time, greenish screen.

And then, I decided to monitor the gpu stats while benchmarking and I could see some things:
i) Green screen started when VRAM temps reached ~85 degrees (C).
2) The GPU was very serenity. I could not hear the fans kick in.
3) Green screen starts as dark-green flashes, then it hangs.
iv) Green screen happen even with DPM disabled.

A lot of people were having these issues even on windows, merely somehow using the latest drivers and upgrading the vbios took care of the outcome. My GPU is already running the latest vbios and I’m running amdgpu drivers (the free ones, not the proprietary amdgpu-pro). This vbios update boosted the vram clock, so I thought I could downgrade the vbios. Turns out you can’t.

So, what I did was:
1) enabled dpm.
2) set the gpu to run at slower pstate (and then the vram clock was not running at max):

# echo “manual” > /sys/class/drm/card0/device/power_dpm_force_performance_level
# echo “2” > /sys/course/drm/card0/device/pp_dpm_mclk
3) installed amdgpu-fancontrol (this puppy https://github.com/grmat/amdgpu-fancontrol)
four) edited the conf and then fans would be boot in early (just and so I could run benchmarks and bank check how things were doing)
five) ran the benchmarks, vram didn’t fifty-fifty achieve seventy degrees. NO Light-green SCREEN.

Afterward that, I simply tweaked a niggling the fan curves so I don’t finish upwards deaf (the gpu is loud as f. when fans are at 100%) and set dpm to automobile once more. With fans really running, I’m not getting green screens anymore.

Turns out you can tweak a lot a stuff with these cards even with the free driver. The archlinux wiki has a lot of stuff on that, even the power draw [https://wiki.archlinux.org/index.php/AMDGPU#Overclocking]

Stay cool

Groovy to hear that you found something that works for you lot! I’ll definitely keep this in mind if I experience the trouble again.

For some reason, I have non had this happen for several weeks now. I’yard also still playing the games that used to cause the green screen crash, but now everything seems to run smoothly. In my example, with a 5700 card, I never found whatsoever correlation between temperature and the crash. Afterward all, the most demanding games I play don’t trigger the issue, and benchmarks too runs just fine. Withal, I have updated my bios in early september, and the driver every bit well every bit gpu firmware have had several updates recently on Arch, and then maybe the newer driver/firmware fixed the upshot.

If I do encounter whatsoever more than crashes I will definitely report back here, and test your solutions as well. But for now, it seems to be fixed on my finish.

And so.. another thing to check:
Organisation RAM
X.M.P with auto timings & frequency, seems to work fine about of the time, simply under very heavy load (CPU + GPU) I will get green flashes.
So, I set up the RAM frequency to 2666mhz (both my mobo and memories allow me go fashion past that, but ryzen and rx5600xt are being a b. about it) and left XMP enabled. Have been benchmarking on the background and working (a bunch of docker containers, intellij and some heavy scala stuff) on the foreground for near 4hs directly, non a single dark-green wink.

So, fan curves + dram settings.