Support needed with 7900 XTX

[Nothingdoing079]

I'm at a loss and trying to get support with an issue which I've been having with a brand new 7900 xtx which was recently brought from Overclockers.

I've changed my GPU from an RTX 3070 to the 7900 XTX, and since doing so have kept having an issue where the PC will randomly restart while I am gaming. All system temps, voltages etc are in line with what is expected, and there have been no random surges which I notice before it restarts.

When this occurs, the PC won't boot past the BIOS Spash Screen, and pressing delete or F2 doesn't do anything. The only way I can get a proper boot, is to turn off the computer, unplug all monitors and then replug them and turn the computer back on.

I've done a number of things to trouble shoot, including the following

- Reinstalling drivers 3-4 times (Using DDU to make sure its a clean install)
- Undervolting the card to see if this had any effect
- Swapping out my HX1000 with a brand new RM1200X (For the PSU I have been using individual PCI-e cables for each power slot, and have not daisy chained at all)
- Changing my CPU to a new 5800x3D
- Installing new 32GB Kingston Fury RAM 3200 MHz
- With the HX1000 trying both a set of sleeved Type 4 cables from Corsair, and also the Type 4 Cables which the PSU came with.
- Doing a full breakdown, clean and reinstall of all parts.

Unfortunately none of these things worked, and I continued to experience crashes. I therefore had to reinstall my RTX 3070, and since doing so, have been able to play games and work with no issues (For example in Starfield I am now playing for 3-4 hours without issue, when before I'd be lucky to manage 10 minutes).

This has led me to believe its the card which is the issue, and have therefore attempted to RMA this with Overclockers.

This was 11 days ago now, and no one in the customer services team has come back to me, and all I have received from them have been emails saying they have combined my queries into the same chain, but have not had any further feedback, or support with this request. Its disapointing that I have to keep chasing considering the issues which I've been having with a £950 card, and not in line with the level of service which I typically have seen from Overclockers.

As I have not had any luck going through the comm channels, I'm now coming to the Overclocker forumns to see if I can get support at all?

Further spec as follows

CPU - 3800x3D
GPU - 7900 XTX Pulse
MB - ROG Strix X570-f gaming (updated to latest BIOS)
RAM - Kingston Fury 32GB DDR4 3200 MHz
Storage - 3xM.2, 2xSSD and 1xHDD
Monitors x 4 - 2x27" 1440p, 1x34" 1440p and 1x10" 1080p

 

[Nothingdoing079]

Yep, its the Sapphire Pulse which I've got.

When monitoring I've been looking at both the Hot Spot and Core Temps. When under load the Core Temp never went above 72C, and the Hot Spot never higher than 85C

 

[Nothingdoing079]

Thanks, the BIOS is the latest version.

I'm contimplating a full reinstall of windows, but really keen to avoid having to do that unless absolutely needed (particuarly as it'll mean hours of reoptimising)

 

[Nothingdoing079]

I can't see it being a windows issue if it doesn't get past the BIOS, unless fast boot is on and it has gotten to the Windows loading stage at which it fails.

Perhaps turn off fast boot (if you haven't already) and see what that yields?

It does sound like a hardware issue with the 3070 having no issues, however.

Interesting, I hadn't thought about if fast boot might be the reason for the splash screen still showing. Will have a look at if fast boot is on or not.

Try a new PSU before concluding the card is bad?

Definately not the PSU, I switched out with a brand new RM1200X and had the same issues arise.

Will be giving Overclockers a call today to see if I can speak with someone on returning the card.

 

[Nothingdoing079]

So I've now spoken to the team and have sent the GPU back this afternoon for replacement

Fingers crossed I'll have a working card soon

 

[Nothingdoing079]

@Nothingdoing079 i had similar issue many years ago and it was a internment short
double check connections are solid (reseat)
double check that your 7900 back panel (where inputs are) is not rubbing against the motherboard i had this once it was touching a screw & would interminently turn off
may check in the fans of the card that nothing is loose

odd things happen its psu/ram or shorts imho (youve checked 2 of 3)

Cheers, the card has now gone back to overclockers so definitely interested in what they have to say when they test it.

Regarding shorts I did have a similar thought when I was troubleshooting, and couldn't see any connections touching anywhere.

I didn't check the fans though, so potentially could be something there. There is part of me that now wonders if I missed something there.

 

[Nothingdoing079]

Seen this a few times on here. No fault found and returned back to customer.

That is one of my biggest concerns that they are going come back and tell me there is no issue, when there obviously is.

Given the price paid for the card, I shouldn't see any issues, let alone have had to do everything I've done to try getting it to work

 

[Nothingdoing079]

Keep us updated @Nothingdoing079. Hopefully, you'll get a new card which just works.

Annoyingly Overclockers have rejected the RMA and claim they have found no issues with the card at all, and that it runs perfectly fine.

Their suggestion is to disable something in the bios which I'll be looking into, but if that doesn't work I'm left with an unusable card and don't know what to do.

 

[Nothingdoing079]

Does it do the same thing with just one monitor connected and are you using multiple pcie cables to connect the card to the power supply without daisy chaining.

Ive used separate pci-e cables for each connection.

Tried with every monitor separately too and still had the issue.

Failing that, after checking for potential shorts, as suggested earlier in the thread try undervolting/down clocking the card. A Windows reinstall is painful but if all else fails it is an option (not so bad if you have another drive you can use that's not the main one). The fact the system doesn't progress past the BIOS splash screen would seem to indicate it isn't software related. Did you check fast boot?

I tried undervolting too and that also didn't work. Have disabled fast boot and acceleration in the bios.

I'll be trying with a clean install of Windows off another drive this weekend to see if that does anything.

Additionally I've got a second pc which I'll try it in, but might need to update the bios on that one.

Curious why you havent returned it within first 14 days of purchase ? Would try their advice with BIOS but if that still does not work and you get the option to check it with some other PC as well then id push em further, they cannot sell faulty items and then reject to replace.

At the begining I thought it might have something to do with the PSU, so had to get a new one. I also spent a lot of time trying to troubleshoot I'm between work and a 6 year old, so really was only able to do things on weekends.

When I asked for a RMA I think I was about 5 days over. Overclockers then took 14 days to reply.

 

[Nothingdoing079]

Regarding the second PC, until the suggestion to try it in another PC I had completely forgotten I even had it.

Which I openly admit was stupid.

I'm going to be giving it a try this weekend and see if that works.

 

[Nothingdoing079]

Only times I have ever had this has usually been the power supply - even though it was meant to run the AMD card fine and ran the previous nvidia card fine. Tried a new more beefier psu and the AMD card ran fine.

Already tried that. I replaced the HX1000 with a RM1200X and continued to have the issue happen

 

[Nothingdoing079]

Just as an update on this.

So I tried to get the card working today doing the following

- Disabled Fast Boot and Core Performance in the bios
- Ran DDU to remove the Nvidia Drivers
- Installed the AMD card, and used brand new Corsair Type 4 cables
- Replaced the DP and HDMI cables with brand new ones in case that was causing an issue
- Triple checked the ensure no element of the GPU was touching the Motherboard or other components.
- Installed the latest driver (choosing to install as driver only in case there was an issue with the adrenaline software)
- played cyberpunk 2077 for 30 mins on high setting.
- crashed

- I then tried removing a pci-e m.2 extender in case that might have been causing lane issues.
- crashed

I then moved the card to a second pc, removing all drivers and doing the same thing as before.
- managed 1 hour before the computer crashed

That means I have now tried it on 2 computers and have had a crash on both. The only variable which was the same is the card

All in all I've wasted around 4-5 hours of my day trying to get an expensive component to run like I'd expect it to straight out of the box

How overclockers can claim the card is perfectly fine and refuse to RMA it is beyond me and incredibly frustrating.

I tend to use them for when I need components which I'm now not going to be doing

 

[Nothingdoing079]

That does sound like the card then. I will get flamed for this but due to past experience and personal preference I do prefer nvidia cards as I have never had problems with them. I suggest contacting ocuk on support again and how did you pay debit or credit card? You have more protection with credit card

I do regret deciding to go with AMD this time round, as all the other times I've been Nvidia and have never had an issue.

I fully appreciate that it can sometimes come down to luck but this has really put me off ever getting another AMD GPU again.

I've sent the above to support so let's see what they come back with

 

[Nothingdoing079]

Definately sounding like the card is faulty. I'd ask OcUK which software they used and how long they tested the card for and if yout able to try the same software on your PC.

Wouldn’t be surprised though if they only the a stress test for 30 mins or so and deemed the card fine.

When they returned it, I asked them what they had tried and got the below

Installed on X670E test rig and connected to 4K display. Ran through all the usual FFXV, 3DMark and Blender benchmarks for a day and then gamed in CyberPunk2077 for around 4 hours and around 1 hour Apex Legends. Tried Ray Tracing on, Ray Tracing off, knocking resolution down to 1440P and messing with settings to change how high the GPU would clock but still remained no issues.

What's surprising in particular is the Cyberpunk thing, as I've not been able to get anywhere close to that amount of time.

I have noted that they tried it on a different platform (AM5 Vs AM4) and a higher resolution (4K) but wouldn't really expect that to have much of an effect.

All I really know at this stage is I'm getting crashes no matter what I try

 

[Nothingdoing079]

Hmm, all I can suggest then is to try a fresh Windows install and see how you get one. It's a bit of a pain but it seems to be the only thing you haven't tried at this point. Could be a software incompatibility somewhere. Could remove current drive and do a fresh install on spare or old drive if you have one.

I could repurpose one of my drives for a fresh install.

I'm 50/50 on if that would do anything considering I also had the same issue happen when I installed in another pc.

That pc is hardly used so windows on it was almost as clean as a fresh install

 

[Nothingdoing079]

You mention you have 4 monitors. Did you try with only one monitor connected?

The reason I ask is that I returned an 7900XTX because of an ever present bug in the driver/software when multiple >60Hz monitors are used at the same time. (Also the terrible coil whine)

With 2+ higher than 60Hz monitors the GPU would idle at ~100W when with one monitor or both monitors at 60Hz it would be at ~30W.

I'm not saying this is the cause of your problem but from reading through all your replies then I would conclude there is something in your setup that differs from OcUK which might be tripping a bug in the Adrenaline driver/software. The key is to finding the difference and I'd start with the monitor.

Try with one monitor connected set to 60Hz. Also definitely try a fresh install OS on a spare SSD. What OS are you using?

I tried it with just 1 monitor connected and had the same issues occur when doing this.

Currently I'm running Windows 11, which is fully up to date.

Going from a nvidia card to amd card you should have DDU'd the old nvidia drivers before installing the amd card, just doing a swap and installing amd drivers on top of nvidia is not a good idea, can lead to all sorts of problems and conflicts, as others have sugested do a fresh install of windows with the amd card installed, download all relevent software and see if the pc and gpu behave normally.

I know, which is why I ran DDU and didn't just plug the GPU in. I've also then ran DDU to remove the AMD Driver, and reinstall it in the hopes that something might have gone wrong then, but no dice.

I will be trying it with a clean OS this evening to see if that does anything, and to give me more evidence if it doesn't.

 

[Nothingdoing079]

Did you try it locked to 60Hz?

I did yes, for the monitors I tried the following

- 1 monitor
- 1 monitor locked at 60 Hz
- 2 monitors
- 2 monitors both locked
- 4 monitors all locked

I also tried with v sync turned off, turned on and with various different settings in adrenaline.

 

[Nothingdoing079]

Further update

Have done a clean install of Windows 10, and it crashed again within 10 mins of playing Starfield

 

[Nothingdoing079]

What kind of crash, a display driver stopped responding, a straight CTD with no error, pc restart or did the pc switch off?

This time round is was a straight CTD with no error showing (which I guess is better than restart as what has been happening)

I'm going give it another go and see if it holds up better in another game

 

[Nothingdoing079]

Crash to desktop is not necessarily a hardware problem

Agreed, based in event viewer it was the AMD display driver failing this time.

I'm going see how things go with it in another game this evening