Arch Linux

jddoxtator

Member

Registered: 2022-01-02

Posts: 43

AsRock RX 7900 XTX Aqua - Drive based instability?

I installed this AsRock RX 7900 XTX not too long ago, and for the most part it works fine in most games.
Some newer titles, especially Unreal Engine 5 titles, the card keeps showing massive stuttering, some times line artifacts across the screen, and crashes of the display driver.

I originally tried re-pasting the GPU block, as manufacturers are known to use very poor quality thermal paste. This did help, but did not entirely eliminate the issue.
Next I tried installing CoreCtrl to manage the power levels of the card for further diagnosis.
This is where strange behavior took on a whole new meaning.
On any other setting other than Performance mode - Fixed - High, the GPU core would only run at about 500 Mhz.
On Performance mode - Fixed - High, it clocks to about 2575Mhz, but what is also interesting is that the crashing and instability is also fixed with performance mode - Fixed - High.

This has me thinking, is there some kind of miscommunication going on between the BIOS of the card and the driver in Linux?
Without this program installed, the system constantly crashes the video driver, Running at the programs theoretical maximum setting the card is perfectly stable.
My only thought is that there is some mismatch between the power setting of the linux driver and the power setting of the VGA BIOS.
Could it be leftover setting from my RX 6800 XT that is just not being removed?
Is this a BIOS issue with the card?
Is it a basic flaw with he current driver?
Something is clearly wrong with how this card is being power managed.

nsfnd

Member

Registered: 2015-12-07

Posts: 5

Re: AsRock RX 7900 XTX Aqua - Drive based instability?

I got a sapphire 7900xtx 6 months ago.
Never had any stability with the card.

Sent back to rma because of high temps (110 degree junction), they sent it back to me saying it works fine.
I repasted PTM7950, temps improved a little, but still around 100 degrees on load, while drawing 370 watts.
Turns out it was a linux issue, in windows while chugging 400 watts, junction stays 85~ degrees, go figure.

About the High profile fixing some things, check out the this issue.
https://gitlab.freedesktop.org/drm/amd/ … te_2477734

Amd gpu linux driver guys are really incompetent, when you report something they ask you to bisect the commits, to figure out which piece of code broke what. They dont test properly apparently, "Works On My Machine(tm), ill commit"

A clear indication that something really is wrong, while running the same game, shadow of tomb raider;
Windows, gpu clock 2525mhz drawing 370watts
Linux, gpu clock 2525mhz drawing 303watts.

Edit: I got crash while writing this

Ara 21 08:00:50 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:50 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:50 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:51 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:51 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:51 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:51 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:52 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:52 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:52 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:54 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:00:54 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to disable gfxoff!
Ara 21 08:00:55 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:55 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:55 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:56 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:56 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:56 enes kernel: amdgpu 0000:0d:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - collecting diagnostic data
Ara 21 08:00:59 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:00:59 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to export SMU metrics table!
Ara 21 08:01:04 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:04 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to disable gfxoff!
Ara 21 08:01:06 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Dumping IP State
Ara 21 08:01:08 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:08 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to export SMU metrics table!
Ara 21 08:01:13 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:13 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to disable gfxoff!
Ara 21 08:01:18 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:18 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to disable gfxoff!
Ara 21 08:01:22 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:22 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to export SMU metrics table!
Ara 21 08:01:27 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:27 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to disable gfxoff!
Ara 21 08:01:32 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:32 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to export SMU metrics table!
Ara 21 08:01:32 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to get fan speed(PWM)!
Ara 21 08:01:36 enes kernel: amdgpu 0000:0d:00.0: amdgpu: SMU: I'm not done with your previous command: SMN_C2PMSG_66:0x00000029 SMN_C2PMSG_82:0x00000000
Ara 21 08:01:36 enes kernel: amdgpu 0000:0d:00.0: amdgpu: Failed to export SMU metrics table!

Last edited by nsfnd (2024-12-21 05:18:07)

cryptearth

Member

Registered: 2024-02-03

Posts: 1,228

Re: AsRock RX 7900 XTX Aqua - Drive based instability?

as this came up a lot lately: any shenanigans like overclocking / undervolting at play here?
as for stuttering and artifacting - I doubt that this was caused by whatever thermal interface material was used - there several videos and even more blogs out there which show that pretty much everything with at least somewhat decent thermal performance can be used
if possible give it a retest on windows to rule out the linux driver - if the card still shows issues it's likely defective -> RMA it