7900 XTX black screen issue

Bigbidoof · 2023-07-12 00:25:25

I just recently built a new machine and have a Sapphire 7900 XTX Nitro+ in it. This GPU is liquid cooled, meaning it has no fans on it. And I suspect that this is possibly why I'm getting blackouts on my machine. Every now and then, but mostly in games, the screen will without warning go black and my monitor will say it lost signal. I then have to kill the power and reboot since the GPU doesn't come back online.

Looking in journalctl after this happens, I can find my GPU complaining how it can't get the fan speed, which makes sense since there aren't any plugged in.

amdgpu 0000:03:00.0: amdgpu: SMU: response:0xFFFFFFFF for index:18 param:0x00000005 message:TransferTableSmu2Dram?
amdgpu 0000:03:00.0: amdgpu: Failed to get fan speed(PWM)!
amdgpu 0000:03:00.0: [drm] *ERROR* [CRTC:72:crtc-0] flip_done timed out
[drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=191198, emitted seq=191200
[drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process DaveTheDiver.ex pid 3386 thread dxvk-submit pid 3468

I've tried setting the fans to be manually adjusted and then to 0, but it looks like it's impossible to do that with the 7000 series cards.
https://wiki.archlinux.org/title/Fan_sp … al_control

Anyone got any suggestions?

ArchDew · 2023-07-14 18:57:02

If that videocard originally had fans, but you made it watercooled, you probably need to edit and flash the vbios on the card and set the fan speed manually to 0. There are tools you can use both for nvidia and amd.
Im not too sure, but i have an msi gtx 960, where the fans somehow wont start and my gpu overheats, so i done the opposite of this, and set the fan speed and some oc on my vbios.

Lone_Wolf · 2023-07-20 08:14:07

This smells a lot like a kernel issue and the fan speed message may be a red herring.

Please upload a full journal output to a pastebin service from the most recent boot with the issue.

Bigbidoof · 2023-07-20 12:59:21

Reposting from the Multimedia and Games thread:

I'm kinda double dipping here since I made a thread on the hardware side of things, but I'm currently losing my mind over this and can't tell if I got software problems or faulty hardware.

I have a RX 7900 XTX in my new built system. Every component is new. But this one seems to be giving me the most grief. It doesn't matter if I'm playing games or just using my desktop, my monitor will suddenly go black saying it lost video signal, and I'll have to reboot my system to get graphics back. Sometimes my system won't even boot. My motherboard will have the indicator light stay on solid saying that there is a VGA problem. According to Asus it means "No VGA detected or faulty hardware". But this is only sometimes and I haven't found a pattern to it, much like everything else. And when I look in journalctl, I don't see the amdgpu throwing any errors. Or really any big red text complaining about crashes from anything. So I don't have much of a scent trail to go on.

In addition, I can't get the IGPU to work with the DGPU. Plugging things into the HDMI out on my motherboard gets no signal, because for some reason the IGPU just sits there unutilized in the background. I can see it looking at amdgpu-top, but it has no activity. But the big issue is why things keep blacking out and dying.

Things that I have tried so far include:
- Re-installing all mesa and amdgpu drivers
- Switching from wayland to X (Launching Deep Rock Galactic in Wayland immediately causes the issue, although it didn't use to, and launching it in X it's back to happening at random)
- Messing with just about every BIOS setting I can find for graphics
- Unplugging the GPU power cables and riser cable and plugging them back in (several times)
- Rerouting the riser cable for the GPU to another PCIe slot (I'm still testing with this to see if it's stable, but I doubt it will make a differce)

I've looked up multiple sources online complaining about blackouts with this card, but I can't find a definitive fix anywhere.

And for those who are curious, hardware specs are:
CPU: R9 7950X
GPU: RX 7900XTX Nitro+
Mobo: Asus Proart X670E Creator Wifi
RAM: 64GB

Bigbidoof · 2023-07-20 13:10:26

Lone_Wolf wrote:

This smells a lot like a kernel issue and the fan speed message may be a red herring.
Please upload a full journal output to a pastebin service from the most recent boot with the issue.

I agree that I don't think it's actually the fans. Especially since I don't see error messages about it in my journal anymore. Problem is, I can't find any errors in my journal.
I run these two commands each time it happens, and find like nothing.

journalctl | grep amdgpu
journalctl | grep ERROR

I'll have to try to get you a capture of it later today since I don't have the time to boot up a game right now. It happens the most often with games, but sometimes on desktop usage. Or sometimes when my computer is waking up from suspend to RAM, my GPU never wakes up.

I'm hesistant to believe it's even a kernel issue because when I fail to boot, I'm not even reaching the BIOS or bootloader. My motherboard will get stuck on the VGA error light and not make it that far. But, when this happens is random. Sometimes it gets stuck, sometimes it doesn't. That being said, I am growing closer to a total OS re-install to see if it makes any difference.

Last edited by Bigbidoof (2023-07-20 13:18:57)

Bigbidoof · 2023-07-21 21:26:41

Small update. Asus released a BIOS update on the 17th and one of the notes is "Improved system stability". So I'm currently running with that flashed to see if it makes a difference. I also am still using the second PCIe slot and not the top one. I haven't been able to play any games yet to properly test stability, but it's the weekend so hopefully I'll get a chance.

I am very wary that I might have a faulty motherboard though. Just because the past two days have been a nightmare of getting the system to POST. It usually either gets stuck on the VGA light (no VGA detected or VGA is faulty), or on the Boot light. And its very inconsistent. Sometimes I POST, sometimes I don't. I haven't found a pattern to it yet.

Bigbidoof · 2023-07-23 06:51:56

Lone_Wolf wrote:

This smells a lot like a kernel issue and the fan speed message may be a red herring.
Please upload a full journal output to a pastebin service from the most recent boot with the issue.

Back and with a new crash after playing some DRG. And this time, it miraculously left me a gift in the journal. It's got some errors with amdgpu.

http://0x0.st/Heed.txt

Edit: I've had a few more crashes, and similar amdgpu errors showing up in my journal.

Last edited by Bigbidoof (2023-07-23 07:14:59)

Lone_Wolf · 2023-07-23 09:50:08

Jul 23 02:24:17 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: SMU: response:0xFFFFFFFF for index:18 param:0x00000005 message:TransferTableSmu2Dram?
Jul 23 02:24:17 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: Failed to export SMU metrics table!
Jul 23 02:24:18 Lacera plasmashell[243044]: [243044:0723/022418.433646:ERROR:gl_surface_presentation_helper.cc(260)] GetVSyncParametersIfAvailable() failed for 1 times!
Jul 23 02:24:18 Lacera plasmashell[243044]: [243044:0723/022418.442802:ERROR:gl_surface_presentation_helper.cc(260)] GetVSyncParametersIfAvailable() failed for 2 times!
Jul 23 02:24:18 Lacera plasmashell[243044]: [243044:0723/022418.459944:ERROR:gl_surface_presentation_helper.cc(260)] GetVSyncParametersIfAvailable() failed for 3 times!
Jul 23 02:24:19 Lacera DiscoverNotifier[241766]: "Object does not exist at path “/org/freedesktop/NetworkManager/ActiveConnection/8”"
Jul 23 02:24:19 Lacera kded5[241545]: "Object does not exist at path “/org/freedesktop/NetworkManager/ActiveConnection/8”"
Jul 23 02:24:27 Lacera kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx_0.0.0 timeout, signaled seq=118447467, emitted seq=118447469
Jul 23 02:24:27 Lacera kernel: [drm:amdgpu_job_timedout [amdgpu]] *ERROR* Process information: process FSD-Win64-Shipp pid 273175 thread FSD-Win64-:cs0 pid 273309
Jul 23 02:24:27 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: GPU reset begin!
Jul 23 02:24:28 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: IP block:gfx_v11_0 is hung!
Jul 23 02:24:29 Lacera kded5[241545]: "Object does not exist at path “/org/freedesktop/NetworkManager/ActiveConnection/8”"
Jul 23 02:24:29 Lacera DiscoverNotifier[241766]: "Object does not exist at path “/org/freedesktop/NetworkManager/ActiveConnection/8”"
Jul 23 02:24:29 Lacera kernel: [drm:sdma_v6_0_ring_test_ib [amdgpu]] *ERROR* amdgpu: IB test timed out
Jul 23 02:24:29 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: IP block:sdma_v6_0 is hung!
Jul 23 02:24:30 Lacera kernel: Failed to wait all pipes clean
Jul 23 02:24:30 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: soft reset failed, will fallback to full reset!
Jul 23 02:24:30 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: SMU: response:0xFFFFFFFF for index:41 param:0x00000000 message:DisallowGfxOff?
Jul 23 02:24:30 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: Failed to disable gfxoff!
Jul 23 02:24:30 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: SMU: response:0xFFFFFFFF for index:59 param:0x00000000 message:DFCstateControl?
Jul 23 02:24:30 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: [SetDfCstate] failed!
Jul 23 02:24:30 Lacera kernel: amdgpu 0000:04:00.0: amdgpu: Failed to disallow df cstate

That appears to be when things go wrong.

Your system has a powerful processor , videocard and lots of memory.
What power supply unit are you using (brand + model) ?

Bigbidoof · 2023-07-23 17:15:01

I like to imagine the PSU is giving enough juice. It's a CORSAIR HX Series HX1200, so 1200w.

Lone_Wolf · 2023-07-24 09:36:38

So power issues are unlikely.
Please install linux-lts and add an entry to boot that kernel to your bootloader config so you can test if that has the issues also.

Bigbidoof · 2023-07-24 19:48:23

Tried launching DRG with LTS running and same issue. DRG is an interesting one, because I used to be able to play it for a few hours without the issue. Like 2-3 hours of play, then suddenly black screen. And now it consistently happens every time in the game after launching. I'll get to the main menu of the game, click to into the lobby, and then black screen. Every time.

Here is a crashlog, although it looks to be the same type of issue as before.
http://0x0.st/Het_.txt

Continuing my suspicion that something is wrong with the motherboard; ever since I got it, it has had this weird buzzing noise coming from the VRMs. It doesn't sound like coil whine as I think of it. It's more of static buzzing sound. And it happens constantly, under load or doing nothing. At what point should I open communication channels with Asus?

Bigbidoof · 2023-07-25 03:02:06

And back with more sad news. I bit the bullet and installed windows on a separate partition. It does it in Windows too. Which completely removes it having anything to do with Linux as far as I can tell. This really smells like hardware failure now. I'm open to suggestions on what might be it. Contenders seem to be GPU, motherboard, and maybe the PSU? I don't think it's the RAM or CPU, but who knows.

Lone_Wolf · 2023-07-25 10:12:21

I overlooked your processor has an integrated AND discrete gpu . That suggests an easy way to narrow down which hardware combo is the problem.

Remove the 7900 XTX and test with just the integrated gpu .

Last edited by Lone_Wolf (2023-07-25 10:12:46)

Bigbidoof · 2023-07-25 12:31:44

I haven't gotten to testing it yet, but I've already looked into using the IGPU and it's always dormant. It does nothing while the DGPU does everything. So I don't think it's the one responsible when things go wrong, but I'll give it a try later.

Bigbidoof · 2023-07-25 22:22:28

Lone_Wolf wrote:

I overlooked your processor has an integrated AND discrete gpu . That suggests an easy way to narrow down which hardware combo is the problem.
Remove the 7900 XTX and test with just the integrated gpu .

First up, I tried troubleshooting the RAM just to make sure. Ran both sticks as singles, just to see if it was one bad stick maybe. And the issue still happens with both sticks running single. So I think that probably removes the RAM. I'm still puzzled why one stick will remember its RGB settings and the other won't. And why the RGB sometimes doesn't respond to changes I make. But that could be something else entirely.

Next up was testing with the IGPU. I had to reflash my BIOS to the older version. The latest one removed any options relating to the IGPU and now the BIOS doesn't even think there is one at all. So I downgraded to the old one and set the IGPU as the primary video device. With the 7900 XTX unplugged from the PCIe slot and unpowered, I booted into the windows partition (since the BIOS won't let me enable CSM so that I can get back to accessing grub). In Windows, I once again launched DRG now on the IGPU. It does not blackout, at least not immediately. (it does crash if I try to edit any of the game settings, but that's not the issue I've been having where it crashes the GPU entirely).

So that's where I'm at. It's possible that DRG needs more time for it to blackout, but I don't know. I've got roughly 30 hours in the game and all of them are on this new machine. So I'll get streaks where I can play before it does it. It's possible it hasn't done it to the IGPU yet. This could take hours though, and it's not an ideal way to play.

Bigbidoof · 2023-07-25 23:59:45

Back with even more interesting news. I've been screwing around with things and noticed something very suspicious.

As you can see here
The GPU power cables are melted.

So now we got three contenders. The GPU, the cable, and the PSU. Right now it's using Cablemod cables, so I need to see if some OEM cables that came with the PSU work. That's a big pain in the ass though with the PSU just barely fitting in the machine. I have to take it out to plug things in. Will update if/when I find anything else.

dawid1 · 2023-09-15 18:18:18

Hey there! My issues are very similar to yours except for the cable melting. I just built my PC, and everything is brand new. Have you found what caused the blackouts? This issue is driving me crazy...

My system:

PSU: Corsair RM1000x
GPU: Sapphire Nitro+ RX 7900 XTX
CPU: i9 13900K
MB: MSI Pro z790-a wifi
RAM: Corsair Vengence DDR5 6000 MHz, but XMP is turned off so it's 4800 MHz

Bigbidoof · 2023-09-17 16:39:40

dawid1 wrote:

Hey there! My issues are very similar to yours except for the cable melting. I just built my PC, and everything is brand new. Have you found what caused the blackouts? This issue is driving me crazy...
My system:
PSU: Corsair RM1000x
GPU: Sapphire Nitro+ RX 7900 XTX
CPU: i9 13900K
MB: MSI Pro z790-a wifi
RAM: Corsair Vengence DDR5 6000 MHz, but XMP is turned off so it's 4800 MHz

Unfortunately no software fixes. Mine was hardware; it was the power cable. I'm on my third one from Cablemod right now, but I haven't installed it. The OEM power cables work fine, but I've tried two Cablemod ones and they weren't working, and still need to swap the third on in.

kgtuning · 2023-09-18 00:46:45

Wait so it was the Cablemod cable that was causing all of your issues?

Bigbidoof · 2023-09-18 13:17:44

kgtuning wrote:

Wait so it was the Cablemod cable that was causing all of your issues?

Yep. The card works fine as far as I can tell, just bad power cables.

kgtuning · 2023-09-18 13:35:12

Bigbidoof wrote:

kgtuning wrote:
Wait so it was the Cablemod cable that was causing all of your issues?
Yep. The card works fine as far as I can tell, just bad power cables.

Im glad you got it sorted. Maybe pick up some wire, atx pins/connectors and a good atx pin crimper and learn to make your own cables? Its time consuming but I enjoy making cables. Just a suggestion as it can save you some aggravation.

Arch Linux

#1 2023-07-12 00:25:25

7900 XTX black screen issue

#2 2023-07-14 18:57:02

Re: 7900 XTX black screen issue

#3 2023-07-20 08:14:07

Re: 7900 XTX black screen issue

#4 2023-07-20 12:59:21

Re: 7900 XTX black screen issue

#5 2023-07-20 13:10:26

Re: 7900 XTX black screen issue

#6 2023-07-21 21:26:41

Re: 7900 XTX black screen issue

#7 2023-07-23 06:51:56

Re: 7900 XTX black screen issue

#8 2023-07-23 09:50:08

Re: 7900 XTX black screen issue

#9 2023-07-23 17:15:01

Re: 7900 XTX black screen issue

#10 2023-07-24 09:36:38

Re: 7900 XTX black screen issue

#11 2023-07-24 19:48:23

Re: 7900 XTX black screen issue

#12 2023-07-25 03:02:06

Re: 7900 XTX black screen issue

#13 2023-07-25 10:12:21

Re: 7900 XTX black screen issue

#14 2023-07-25 12:31:44

Re: 7900 XTX black screen issue

#15 2023-07-25 22:22:28

Re: 7900 XTX black screen issue

#16 2023-07-25 23:59:45

Re: 7900 XTX black screen issue

#17 2023-09-15 18:18:18

Re: 7900 XTX black screen issue

#18 2023-09-17 16:39:40

Re: 7900 XTX black screen issue

#19 2023-09-18 00:46:45

Re: 7900 XTX black screen issue

#20 2023-09-18 13:17:44

Re: 7900 XTX black screen issue

#21 2023-09-18 13:35:12

Re: 7900 XTX black screen issue

Board footer