Graphics Cards

[AngryMustache]

Hello everyone, first post here, hopefully I can get some insight from other people as I am at my wits end now.

I received a reference 7900 XTX that I ordered from the official store earlier today. Packaging was pristine, nothing wrong with the box. Opening was a delight with the way it was all presented.

Before installing the new card I did use Display Driver Uninstaller and cleaned previous video card drivers off since it was replacing a 2070 super. Got the new card seated and hooked up. Started windows everything was fine, no drivers installed yet. Downloaded the only release listed for 7900XTX (22.12.1). As soon as the driver installation began I noticed weird things happening like both of my monitors being with black backgrounds like explorer had crashed or something similar. Thought nothing of it, let installation finish and restarted the computer.

Herein lies the problem. With drivers installed I am getting freezes, crashes every 15-30 seconds, some being recoverable but a few times it isn't, therefore making the card unusable. Sometimes when it happened would beep at me 3 times. Event viewer is filled with some amd dwm dll has crashed and successfully recovered.

I have tried many things to try and isolate the problem. I have tried separate PCIe cables, disabling expo ram, disabling smart access memory, changing PCIe link speed, fresh install of windows 10/11, only plugging in one monitor, etc. These do not solve the problem, and there is no change. Hard to do much diagnosing because it's also locking up on the windows login screen for 5-10 seconds at a time when it soft-locks.

The only time it does not crash or lock up in any way is in bios and windows when the driver is not installed or when I am in safe mode. I have also tried a driver only installation and the same thing happens.

I have only had this issue once before years ago with a refurbished 1080ti that was also crashing after driver installation and only worked in safe mode. Found out they didn't check the thermal pads, so it was overheating. I'm really hoping it's something simple and I didn't receive a malfunctioning unit.

Would appreciate any insight as to whether I really do have a bad card or I'm missing something simple.

[Rhuz2000]

I am currently in a boat kind of the same with a rx 7900 xt. Downloaded drivers, made sure everythign else was up to date (chipsets, bios, windows, and so on), powered down changed out my 6800xt with new card. Booted up fine. Started install and AMD installer did its reset . Then nothing. Computer will not post, no signals, VGA error light on x570 board and will not post. Corsair 850 watt power supply. Seated good. all power plugged in. Fans spin up on card. Pull and put in 6800xt and it boots just fine. 

[MADZyren]

I had this weird thing happen today. I used 5700XT, then removed it and have used integrated gpu for a week. Now, as a temporary solutions, I installed Nvidia GTX970 which refuses to give any image and AMD motherboard crashes. I never even got to BIOS. Did clear CMOS just in case it was a settings thing. Didn't help. Removed GTX970 and installed it to Intel machine where it works normally. 

[Rhuz2000]

R5 5600x, MSI x570-a pro, 32 gigs ddr4, ROG STrix 1000w psu (did have a Corsair Tx850m). Rx6800xt (Works fine) Rx 7900xt (does not post to bios just VGA error light on board). Have tried cmos reset, bios roll back, all pci-e slots, clean install of AMD drivers and chipset. Aso well as a system roll back. All same results with the RX7900xt. No post, VGA error code on Mobo.  

[ali_d]

Can you provide your system details please?

Motherboard, RAM make and model, Power Supply make and model, how is the graphic card connected to the motherboard, are you using two separate power cables from the Power Supply to the graphic card.

Any additional information would be grate.

[AngryMustache]

Reposting this reply since I think I responded to the topic, not to your post, so maybe you didn't get a notice I had.

Asus B650E-E Gaming Motherboard, 2x16gb G-Skill Flare X5 Series model no. F5-6000J3238F16GX2-FX5, Be Quiet Dark Power Pro 12 1500w psu. 7900 XTX Connected directly to the first PCI express 5.0 x16 slot. I have tried the same and two separate pci express cables from the psu. Have tried changing pcie lane speeds, resetting bios to defaults, etc. Have found nothing that keeps the drivers from crashing constantly (according to event viewer) and getting the report dialog from amd about a driver timeout.

I've tried:

Fresh windows 10 / 11 install

Tried both separate 8 pin pci express cables and two 8pins on same power cable, did a memory test, taking each stick out and testing individually and together, unplugging each monitor leaving only one plugged in, switching to the other monitor, trying different display port cables (mine are 1.4). Clean install of drivers with both amd and ddu and reinstalling the drivers. Most of the time there is some kind of soft crash while even installing the drivers. Only resolutions for me are uninstalling the video card drivers, or going into safe mode. Also the system has no issues when I revert back to my EVGA 2070 Super.

Like I said above this all makes me think I got a dud. Since it seems to me hardware acceleration would only be a thing once drivers are installed, and once that happens the card becomes unusable even locking up as early as the windows login screen. I can't really provide anything more like a dxdiag since like I said the system locks up constantly post-install.

[MADZyren]

I have that same motherboard (650E-E Strix Gaming WIFI) and 7950X. No GPU yet, so using integrated. It works except when ever there is any web browser page open with any kind of video. Then it can crash if hardware acceleration is on. So I wonder if you have web browser on or use some software that uses hardware acceleration, even when no video is playing, like perhaps some overlay.

Have you tried disabling MPO? https://nvidia.custhelp.com/app/answers/detail/a_id/5157/~/after-updating-to-nvidia-game-ready-drive... 

Also have you tried running the computer with integrated graphics to exclude other issues?

Just as a warning, when I tried to set IGFX as primary graphics device from BIOS and changed video ram from AUTO to 8 Gb (I have plenty of RAM), computer would not even post and when I had cleared CMOS, my fTPM was empty. Glad I did not use Bitlocker or my files would be in byte heaven... I am using 0421 BIOS. Did not upgrade as some people mentioned there are stability issues with later BIOSes, so I am still waiting for a better BIOS before I flash and this seems to work outside these few hickups. Someone had actually fixed his issues by downgrading BIOS to earliest version he could find.

[AngryMustache]

Yeah I have tried disabling MPO, disabling the integrated gpu. I haven't tested the integrated but switching back to my old 2070 Super and all the issues disappear, so it's definitely the card itself or the drivers. I've done basically all the hardware troubleshooting I can do. Loaded bios defaults, disabling expo profile, and disabling resize bar. With the 2070 super in I have 0 stability problems, no crashes, freezes or lockups.

[MADZyren]

I am still a bit skeptical with BIOS, because when I used my ole 5700XT everything was fine. Then I used IGFX and everything else worked, except HW accelerated video playback and what I mentioned about dedicating memory to it (which basically does nothing, but some games  "see" only the dedicated portion). But I got an old GTX970 and when I install it, nothing happens. I can't even get to BIOS with it. When I put it to an old machine (Intel i7, I think it was 3000- or 4000 something), same GTX970 works just fine. With 7950X I tried pretty much everything from clearing CMOS, but no. Never experienced this before. Despite never booting to anywhere, Windows still, to some extent, must have noticed Nvidia GPU as after I  managed to get computer boot again to Windows, Windows started downloading Nvidia GPU drivers despite GPU was no longer there.

Not sure, but I think there will be some BIOS version one day, hopefully soon, which fixes all this. Also I find it quite strange, video playback can be such an issue with everything released after RX5700XT, from 6000 -series GPUs to APUs to IGFX.

[AngryMustache]

Asus B650E-E Gaming Motherboard, 2x16gb G-Skill Flare X5 Series model no. F5-6000J3238F16GX2-FX5, Be Quiet Dark Power Pro 12 1500w psu. 7900 XTX Connected directly to the first PCI express 5.0 x16 slot. I have tried the same and two separate pci express cables from the psu. Have tried changing pcie lane speeds, resetting bios to defaults, etc. Have found nothing that keeps the drivers from crashing constantly (according to event viewer) and getting the report dialog from amd about a driver timeout.

I've tried:

Fresh windows 10 / 11 install

Tried both separate 8 pin pci express cables and two 8pins on same power cable, did a memory test, taking each stick out and testing individually and together, unplugging each monitor leaving only one plugged in, switching to the other monitor, trying different display port cables (mine are 1.4). Clean install of drivers with both amd and ddu and reinstalling the drivers. Most of the time there is some kind of soft crash while even installing the drivers. Only resolutions for me are uninstalling the video card drivers, or going into safe mode. Also the system has no issues when I revert back to my EVGA 2070 Super.

Like I said above this all makes me think I got a dud. Since it seems to me hardware acceleration would only be a thing once drivers are installed, and once that happens the card becomes unusable even locking up as early as the windows login screen. I can't really provide anything more like a dxdiag since like I said the system locks up constantly post-install.

 

 

[Aboshi]

You are not alone!

These reference cards are 100% trash and I see AMD getting sued in the very near future. I was really rooting for AMD but it’s clear they still don’t have a clue on what they are doing. Think back to the first Ryzen release and every single cpu was muffed with the seg fault issue. They were warrantying them out, but you can imagine the millions and millions lost because zero QC. Now fast forward to these trash ref cards… I have had 3 ref cards so far and never again. 2 Powercolor and 1 sapphire, both direct partners to handle ref cards for AMD. So here’s the issue beyond their supper buggy drivers. I monitored all 3 of this cards pushing 361-378w 

well guess what? Each pcie rail (8 pin) will pull 150w and your pcie slot will pull a max of 75w
I have seen a sudden spike and then instant hard lock in windows! This can happen directly after windows loads in or randomly within 40 mins. It will also happen randomly when playing any games or stressing the gpu. What I suspect is that these trash ref cards are trying to pull even more then the wattage I saw before the hard lock and then windows takes a dump. It’s a piss poor design flaw and the easy remedy would have been to add another 8 pin for an additional 150w if needed and clearly it is!
I have tested these 3 gpus with 2 different psus 850w and 1000w evga supernovas and it’s the same outcome every time. System is rock solid using a rtx 3090, rtx 3080 and the 3090 is sucking down 495w with zero issues. I also tested it with just 2 pcie rails and the 3090 was still rock solid with ZERO lockups or hard freezes. I can say with pretty high confidence that AMDs aib partner cards will be stable because they all have x3 8 pins and will account for the extra power draw. But FFS AMD way to rush out an epic failure of a reference card for a cash grab! I man **bleep** $0.25 could have fixed this issue… now I’m sure it will cost you millions in the long run

Anyway these 3 cards will be returned and if the aib cards fail in the same way, l will be canceling my orders will all distributors and nvidia will be getting the business (and I’d rather not give nvidia my business) so AMD get your ish together!

[Sunny6233]

Currently experiencing this with a hellhound 7900xt and have done everything you have done as well. Send the card back asap. 

Asus Z790p DDR5 motherboard

CPU- i7-12700kf (unlocked)

2x32gb Corsair Vengeance DDR5 RAM 5200mhz

GPU- RX7900XT

[Moist]

It sounds a bit like you and I are having similar issues. I was going to try and install a fresh windows to see if it will fix it. I even got my friend to test the card in his system and he had no issues at all. So there's from what I can tell from my trouble shooting there definitely seems to be some driver/software windows issues. Here is the link to my issues https://community.amd.com/t5/part-recommendations/amd-radeon-rx-7000-series-official-resource-discus... not sure if we're having the same problem but if it is then it's maybe not even your card but the software/driver side is the issue.

[Isabella_Nora]

AMD’s new flagships launched just recently, and based on our own testing, we can say that they’re mostly fit to rival the Nvidia RTX 4080 , although their ray tracing  performance leaves a lot to be desired. However, some users have found the overall performance offered by the RX 7900 XTX and the RX 7900 XT  to be inconsistent, and Kepler_L2 on Twitter may have discovered why.

[AngryMustache]

Update for anyone who might have something similar.

Got approved for an RMA on 12/22 and sent this card back, received new one today 1/7/23

Unfortunately this one seems to have the 110c junction temperature problem, so it's got to be returned as well.

Unusable unless I frame limit cap and turn settings down. Inexcusable for a thousand dollar card.

Wish others more luck than I've had.

[Aboshi]

110c return immediately. I have had 7 ref cards in a row that all hit 110c. They are lying when they claim it’s a small percentage of cards. I was 7 for 7 so unless I won the quantum lottery of bad cards it tells me it’s a major problem. I now have replaced all of those cards with power color red devil cards, and I have ZERO issues! Even when I boost the frq to 3100, 15% more wattage (405w) dropping voltage to 1100 and ram at 2750, fans set to 50% I am 100% stable and max junction temp is 86c under extreme load gpu temp 66c. It goes down even further at 75% fan speed to 78/56c and I’m sure at 100% fan speed it would go down further but I don’t want to listen to that

This shows you just how bad of an issue Amd has on their hand with the ref cards. I think they should recall every one of them and throw a true heat pipe on there as they should have done in the first place. I have some colleagues that have ref cars and they max out 98c hot spot on their ref cards, but that is by pure luck and 98c is still very hot within spec or not.

There are driver issues as well. When using chrome or brave browser. Make sure you turn off hardware acceleration and turn off run in background. If you do not do this your system will hard lock randomly with brave installed, and chrome will hard lock the system for a few second to a minute and might recover. Brave browser will be a hard lock with no warning except your mouse stutter for a second if you happen to be using your system at the time of the hard lock, but you will need to hard power down. Same goes for chrome but might recover if your lucky. When watching YouTube the system will hard lock, 1-30 seconds later you might hear sound again and then it will recover at that point. If not hard power down and reboot.