7900 XTX Sapphire Nitro+ keeps crashing

[fusseli]

I've tried all combinations at this point. Fresh Adrenalin install, even fresh Windows install, OC bios, non-OC bios, turning all features on or off, etc. and I still get random driver crashes maybe once or twice a day of gaming. It's incredibly infuriating during a competitive game, and still mildly annoying casually gaming. I've sent the AMD bug report more times than I can count.

Any ideas? I may have to flip this thing for a 4080 super

 

[wolf]

It sounds like it needs to be RMA'd given what you tried, last option is perhaps to test in another system, if it's the same it's definitely the card.

Either that or as you say/flip return for a different card.

 

[Super Firm Tofu]

I've tried all combinations at this point. Fresh Adrenalin install, even fresh Windows install, OC bios, non-OC bios, turning all features on or off, etc. and I still get random driver crashes maybe once or twice a day of gaming. It's incredibly infuriating during a competitive game, and still mildly annoying casually gaming. I've sent the AMD bug report more times than I can count.

Any ideas? I may have to flip this thing for a 4080 super

Sounds exactly like what my copy of that same card did. I returned to Amazon and bought the PC Red Devil and it's been flawless since. There's been posts here in the forum about other models of the Nitro+ RDNA3 cards doing it as well.

I'd either exchange or return if that's still an option, or get it in for RMA to have it repaired or replaced.

The 4080 Super is a decent option with the price drop to $999.

 

[fusseli]

It's past the retail RMA or exchange window, I'll see what Sapphire support says.

 

[Ferrum Master]

We had few members who had issues just because their PSU couldn't handle the current spikes.

Just keep that in mind. The whole PSU rating system we have currently is utter bullshit and doesn't tell you if PSU X will work with GPU Y. Despite some amateurish try to introduce new ratings, it creates even more confusion.

 

[Gungar]

We had few members who had issues just because their PSU couldn't handle the current spikes.

Just keep that in mind. The whole PSU rating system we have currently is utter bullshit and doesn't tell you if PSU X will work with GPU Y. Despite some amateurish try to introduce new ratings, it creates even more confusion.

It's a 1200W psu it can handle any spikes of a 7900XTX.

 

[Ferrum Master]

It's a 1200W psu it can handle any spikes of a 7900XTX.

1200W does not mean it handles spikes well.

 

[Vya Domus]

1200W does not mean it handles spikes well.

A PSU with 1200W on the 12V rail will, by definition, be capable of handling spikes of up to at least 1200W. There is no way in hell this card spikes above 1200W. Spikes are an issue when let's say you have a 500W PSU on the 12V rail and the card spikes up to 600W.

PSU issues would result in total system crash, not just driver timeout, I don't know why people still can't differentiate between the two.

 

[Toothless]

Uninstall the last few Windows updates, reinstall drivers, give er' a try.

 

[Ferrum Master]

A PSU with 1200W on the 12V rail will, by definition, be capable of handling spikes of up to at least 1200W. There is no way in hell this card spikes above 1200W. Spikes are an issue when let's say you have a 500W PSU on the 12V rail and the card spikes up to 600W.

PSU issues would result in total system crash, not just driver timeout, I don't know why people still can't differentiate between the two.

Sorry to shatter your ideas... but absolutely not. Current capacity does not mean it handles transient response well. I won't indulge too much into it, but those numbers, larger better and everything comes as a package doesn't apply here. It rarely does anywhere actually. PSU performance consists of many parameters and no badges, ratings say that unit X will work fine with unit Y. For example like RAM compatibility table.
You can easily cause device instability with bad PSU. Our eyes or GPU-Z logs are too slow to see it, but our cards work in few GHz realm at ~500W peaks.

 

[Vya Domus]

Sorry to shatter your ideas... but absolutely not. Current capacity does not mean it handles transient response well. I won't indulge too much into it, but those numbers, larger better and everything comes as a package doesn't apply here. It rarely does anywhere actually. PSU performance consists of many parameters and no badges, ratings say that unit X will work fine with unit Y. For example like RAM compatibility table.
You can easily cause device instability with bad PSU. Our eyes or GPU-Z logs are too slow to see it, but our cards work in few GHz realm at ~500W peaks.

But do you understand why transient spikes are a problem in the first place ? It's because usually a card draws the most power in very short time intervals and those spikes can be enough to trigger OPP, OCP but those limits are according to what the PSU is rated for. Ironically this isn't a problem on PSUs with crappy protection circuits because they will fail to trigger if the transients are fast enough.

There is simply no way in hell you're gonna trigger those protections on a PSU rated for 1200W with a 7900XTX because it will never draw that much current, it simply cannot happen.

 

[Ferrum Master]

There is simply no way in hell you're gonna trigger those protections on a PSU rated for 1200W with a 7900XTX because it will never draw that much current, it simply cannot happen.

You don't get the idea sorry... I am too tired to explain how it looks in a scope and how good clean response looks vs sloppy, sagging or ringing one. It has nothing to do with protections here.

If you poke around our QA helpdesk posts, some users have reported that changing PSU have fixed their radeon stability problems.

 

[Vya Domus]

It has nothing to do with protections here.

lol, it has everything to with that, transient spikes cause protections to trigger, that's literally why this is a problem and why a PSU that's rated for a higher wattage fixes the issue.

 

[jesdals]

Windows 10 or 11? I think win 11 is better for AMD 7900 XTX

 

[Ferrum Master]

lol, it has everything to with that, transient spikes cause protections to trigger, that's literally why this is a problem and why a PSU that's rated for a higher wattage fixes the issue.

Lol yourself. Protections are for catastrophic failure prevention. Thus named protection. Sloppy power delivery occurs at all current ranges the PSU work, let it be 100-200W jump.

 

[Assimilator]

The fact of the matter is that numerous users on this forum, who have experienced similar issues with AMD GPUs yet have high-quality and high-output 2.x PSUs, have seen their problems miraculously vanish when switching to an ATX 3.0 unit. Now, while I will be first to say that this is an absolutely shit experience that AMD should be incredibly ashamed of, it's also an option if all else fails. If you do decide to try it, please advise of your experience!

 

[fusseli]

The fact of the matter is that numerous users on this forum, who have experienced similar issues with AMD GPUs yet have high-quality and high-output 2.x PSUs, have seen their problems miraculously vanish when switching to an ATX 3.0 unit. Now, while I will be first to say that this is an absolutely shit experience that AMD should be incredibly ashamed of, it's also an option if all else fails. If you do decide to try it, please advise of your experience!

Yes the 7900XTX with OC power draw averaging ~400watts, and 440 watts extra OC'd (I don't run), and then intel 14 series needing mega power, is why I went with an ATX 3.0 1200W. (I got the 1200w because it was on sale cheaper than a 1000w.) If I switch to default 7900xtx bios the average power draw is 340w or something like that, and it still has driver crashes occassionally.

I've had full blackout system crashes before, I've had every type of crash under the sun over the years and obviously know and prefer 24/7 stability, I agree that type of crash is almost certainly a psu overcurrent or other protection trip and insta-shutdown. I do not think my spurious driver crash is my psu, or windows 11, it's done it since this gpu was new 6 months ago. I started with this gpu on z490/10700k and guess what, it did it then too.

All Sapphire support says to do is file a warranty request with tech@althonmicro.com

 

[Deleted member 237269]

Grey screen with my Saphhire 7900xtx Nitro+ once a day but it might be just an incorrect OC under adrenaline.

Concerning the PSU i’m using the MSI 1300w Platinum. Got this bad boy for 270€~, wanted something who could keep up with the trend of energy hungry degenerence of GPU in 2025.
Here some interesting stats https://hwbusters.com/best_picks/best-atx-v3-0-pcie-5-0-ready-psus-picks-2023-hardware-busters/
The Seasonic TX-1300 is a beast but 510€~

 

[Chomiq]

The fact of the matter is that numerous users on this forum, who have experienced similar issues with AMD GPUs yet have high-quality and high-output 2.x PSUs, have seen their problems miraculously vanish when switching to an ATX 3.0 unit. Now, while I will be first to say that this is an absolutely shit experience that AMD should be incredibly ashamed of, it's also an option if all else fails. If you do decide to try it, please advise of your experience!

But OP already has ATX 3.0 PSU.

 

[Toothless]

Soooo didn't uninstall latest windows updates? Because same card, same drivers, same issues and that was the fix for me.

 

[Dr. Dro]

Any ideas? I may have to flip this thing for a 4080 super

And it'd be about the best decision you ever made. I believe the thread's contents more than justify my statement.

A brand new, ATX 3.0 native, high-end 1200 W PSU from a reputable manufacturer and your computer's crashing? Just replace and upgrade it to a higher capacity unit to handle "spikes" bro. Driver issues? Install and uninstall different driver versions and select Windows updates to workaround the problem, otherwise it's user error. Expensive flagship AIB model from flagship AIB partner has problems? Just exchange it for another flagship design from a very high end AIB, that's gotta solve all the problems!

Honestly, just buy yourself an RTX 4080 and actually spend your time playing video games instead of troubleshooting problems that will never be fixed because the Radeon community would rather stonewall, deflect and deny rather than demand AMD do anything about it. When I did demand, I was just one guy. Feedback entering from one ear and leaving through the other. So, yeah. I understand that there's genuine good intentions behind each and every post but... guys. Really. You need to blow up AMD's customer feedback channels to get them to do something about these recurrent problems. We just can't reasonably expect that end users work around their problems and shortcomings.

I'm just so frustrated.

 

[Toothless]

And it'd be about the best decision you ever made. I believe the thread's contents more than justify my statement.

A brand new, ATX 3.0 native, high-end 1200 W PSU from a reputable manufacturer and your computer's crashing? Just replace and upgrade it to a higher capacity unit to handle "spikes" bro. Driver issues? Install and uninstall different driver versions and select Windows updates to workaround the problem, otherwise it's user error. Expensive flagship AIB model from flagship AIB partner has problems? Just exchange it for another flagship design from a very high end AIB, that's gotta solve all the problems!

Honestly, just buy yourself an RTX 4080 and actually spend your time playing video games instead of troubleshooting problems that will never be fixed because the Radeon community would rather stonewall, deflect and deny rather than demand AMD do anything about it. When I did demand, I was just one guy. Feedback entering from one ear and leaving through the other. So, yeah.

I'm just so frustrated.

We're in the same community and I actually see fixes come out. My issue was Microsoft being extra dumb for no reason. Not sure being in the program is good for you with that attitude.

 

[Dr. Dro]

We're in the same community and I actually see fixes come out. My issue was Microsoft being extra dumb for no reason. Not sure being in the program is good for you with that attitude.

If you mean Vanguard, I haven't been active. I don't even interact other than checking the Discord's off-topic channel every now and then anymore. I won't say they haven't done anything, but the recurrent fundamental problems with stability just seem to never go away. These threads never stop springing, and it's like they don't even care. So why should paying customers care? End of the day that's how I see things.

 

[Toothless]

If you mean Vanguard, I haven't been active. I don't even interact other than checking the Discord's off-topic channel every now and then anymore. I won't say they haven't done anything, but the recurrent fundamental problems with stability just seem to never go away. These threads never stop springing, and it's like they don't even care. So why should paying customers care? End of the day that's how I see things.

Sure NVIDIA drivers are more stable, that's why Vanguard exists no? To help the community by being the bug testers? What about your contribution to the forums? What about everyone running WCG?

Point is while yes AMD has driver issues, work has to be put in for them to be better. Hope being one day they're just as good or better than NVIDIA's and I've seen quite a bit of good come from Vanguard. It's not the end of the world and every person I've seen running AMD in my friend's list, IRL, etc doesn't have issues with their cards. Sadly when issues pop up people point fingers fairly quick.

 

[Dr. Dro]

Sure NVIDIA drivers are more stable, that's why Vanguard exists no? To help the community by being the bug testers? What about your contribution to the forums? What about everyone running WCG?

Point is while yes AMD has driver issues, work has to be put in for them to be better. Hope being one day they're just as good or better than NVIDIA's and I've seen quite a bit of good come from Vanguard. It's not the end of the world and every person I've seen running AMD in my friend's list, IRL, etc doesn't have issues with their cards. Sadly when issues pop up people point fingers fairly quick.

I agree... had OP been running pre-release nightlies, but this isn't the case. Recurrent problems such as these shouldn't affect stable, certified drivers for a product that's been in-market for over a year. Between you and I, Vanguard has been the best thing that has ever happened to AMD. I just wish the suits who actually have the final say saw it the same way - because I actually fear where they'd be without it.