what does rma stand for.
They are cheap so it is best way to test if this solves your problem as @Echoa said.
RMA
Return Merchandise Authorization.
that is what i do at this moment as well. it would take 2 or 3 weeks without a gpu. That is sad.
Well it could be that it’s is defective. 
The exact power draw is going to depend on the version of the 5700 you have, 160w isnt the peak but the average you will peak higher. Also the issue is the stability of the power the cable can deliver, daisy chain cables deliver less stable power vs individual cables.
I.E the 5700xt (yours being 5700 i know) is rate for around 180w but because AMD measures power draw at the VRMs and its less accurate you may be +/- a significant amount but its a good enough ballpark. My Sapphire Pulse 5700xt draws over 200w during load.
If you have a board partner card with a better cooler its likely been set to run higher/faster and peaks will likely exceed 200w. Due to power losses with the cable and depending on the PSU design you may be experiencing issues related to the PSU/Daisy Chain cable especially if the cables are made of aluminum which lower priced PSU tend to use (not sure of the tier yours is)
1 Like
how can i control it if it goes higher up? there some tool to keep an eye out for this?
you can test power issues by lowering the power limit in corectrl, try lowering it 50 or so Watts and see if you have issues. You cant really control/accurately measure your peaks/transient power draw with software but you can try to lower its peaks that way and test.
you tell me to downpower the gpu to 50 watts that right? Default in corectrl is 160. The 50 seems very low compared to that.
What if the supply is not enough will it power down or just have issues doing what it is told to do ?
If the PSU power output wasn’t enough you usually trigger over current protection (some PSU dont have it) but I don’t believe the total output is the issue but the output stability which can cause the GPU to act weird also yes lower it 50w from 160. Its a test to see if its a power related issue.
okay so i disabled all profiles, set the globla to 50w and am going to start games now or what am i doing with this low watcount?
no lower it by 50w to 110w lol
set it to 110w global and test your games for green screen
the 50w were not kept. i took a peek at the corectrl stuff and it just got to 116 in one jump while loaded and spawned on a map.
yeah 116w is fine itll not be 100% at the 110w, youre just testing telling it to use as little as possible. Just minimizing the power while keeping some performance so you can test is all no need to constant monitor the draw/usage
okay so with corectrl set to 110w maximum i see the voltage value spike over it regularly. 135w in one application was reached easily.
Others did not use more than what wsa set as limit but were far away from running fine. (even before the limit so no news there)
Did you green screen like you have been?
no none of the games has recently. since i can’t reproduce it i hve no idea why not. i also noticed. Some games do not use the potential the gpu would offer. Example will be payday 2. it stays around 40 w and lags horribly.
verdun on the other side does use over 110w and runs with 20frames. Is there a way to force the voltage ?
How have you made out with the green screening?
@Tads
I have added some other pkgs for my Radeon to see if it made any difference based off of this post.
https://forum.manjaro.org/t/amd-radeon-5700-xt-fully-working/118189
Edit: This is what i have installed.
[ricklinux@eos-xfce ~]$ pacman -Q | grep amd
amd-ucode 20200519.8ba6fa6-1
amdvlk 2020.Q2.4-1
lib32-amdvlk 2020.Q2.4-1
opencl-amd 20.20.1089974-2
xf86-video-amdgpu 19.1.0-2
[ricklinux@eos-xfce ~]$ pacman -Q |grep radeon
lib32-vulkan-radeon 20.1.1-2
vulkan-radeon 20.1.1-1
[ricklinux@eos-xfce ~]$ pacman -Q | grep vulkan
lib32-vulkan-icd-loader 1.2.141-1
lib32-vulkan-radeon 20.1.1-2
vulkan-headers 1:1.2.143-1
vulkan-icd-loader 1.2.141-1
vulkan-radeon 20.1.1-1
vulkan-tools 1.2.140-1
[ricklinux@eos-xfce ~]$ pacman -Q | grep clang
clang 10.0.0-3
lib32-clang 10.0.0-1
[ricklinux@eos-xfce ~]$ pacman -Q | grep mesa
lib32-libva-mesa-driver 20.1.1-2
lib32-mesa 20.1.1-2
lib32-mesa-vdpau 20.1.1-2
lib32-opencl-mesa 20.1.1-2
libva-mesa-driver 20.1.1-1
mesa 20.1.1-1
mesa-demos 8.4.0-4
mesa-vdpau 20.1.1-1
opencl-mesa 20.1.1-1
[ricklinux@eos-xfce ~]$
Excuse the long absence.
The GPU has been returned to the manufacturer and i had to wait 5 weeks for the new one. they apparently changed the cards and so far there has not been any trouble.
Only thing is: I can’t really say if it is the software having fixed the bug or the hardware.
Thanks for following on this frustrating ride and all the good ideas on how to fix it. Sadly no fast Solution has been found.
2 Likes