05-03-2025 08:12 PM - edited 05-03-2025 08:59 PM
I've troubleshooted this for 2 weeks now and am close to pulling my hair out.
I bought an ASUS TUF RTX 5080 OC Edition 2 weeks ago to replace an EVGA RTX 3080 FTW3. This 5080 card is not stable at all in my computer. I'm trying to determine if I have a driver issue or a hardware issue (needing an RMA). I have tried every driver available on NVIDIA's website that they offer for a RTX 5080 (DDU Installed in safe mode each one). I have finally given up and put my 3080 back in and now have no issues. I've even tried with a 4070 I had from another machine, and it has no issues.
The crux of the issue is that in ANY DX12 app, the card crashes and throws me back to desktop within 5 minutes (usually <1m but it sometimes lasts a little longer). If the app gives an error, it's always a DX12 related error message. This also includes all the DX12 benchmarks in 3D Mark which include Speed Way, Steel Nomad, Port Royal & Time Spy. The DX12 games I've tried include CoD BO6, Assassins Creed Shadows, Elder Scrolls IV: Oblivion Remaster, Cyberpunk 2077, Starfield, Star Wars Outlaws, Control & Alan Wake 2. It crashes in every one of those games and benchmarks.
The funny thing is, that as far as I've been able to test, DX11 games don't seem to have this issue. They run pretty stable.
If this is not a driver issue, I suspect the factory OC may be unstable. I have tried downclocking it with GPU Tweak 3 down to both the non TUF 5080 OC clocks and the Nvidia FE clocks. Neither of those settings improved the crash rates.
I've tested this card in two different Rigs. Exact same behavior in both rigs.
Since this RTX 5080 OC Edition card is essentially unusable, I've gone back to my RTX 3080 and it works flawlessly.
Test Rig 1:
GPU: ASUS TUF RTX 5080 OC Edition / Running Factory vBIOS.
Current NVIDIA Driver: 576.28
Tested NVIDIA Drivers: Tried all publicly available drivers for 5080 via Nvidia's website w/ DDU install in Safe Mode.
MoBo: ASUS Crosshair VIII Hero WiFi / Running Bios Version 5002 (Current as of this post)
CPU: AMD Ryzen 7 3700X
RAM: Corsair 16GB DDR4-3200 x4 (64GB Total)
PS: EVGA 1000w
OS: Windows 11 24H2 Build 3775
Test Rig 2:
GPU: ASUS TUF RTX 5080 OC Edition / Running Factory vBIOS
Current NVIDIA Driver: 576.28
MoBo: ASRock Z790M-ITX WiFi
CPU: Intel Core i7 13700
RAM: Corsair 32GB DDR5-5200 x2 (64GB Total)
PS: EVGA 850w
OS: Windows 11 24H2 Build 3775
Solved! Go to Solution.
05-10-2025 02:06 AM - edited 05-10-2025 02:10 AM
TLDR:
I suspected my card was bad. I started the process to return/exchange it with Best Buy a couple days ago. Today, I got my replacement GPU. It's the same SKU as the previous bad card (ASUS TUF RTX 5080 OC Edition). After 6 hours of testing, I can confirm this new card works! I have not had a crash at all. I couldn't go more than 1-5 minutes without a crash before. I'm running the current NVIDIA driver (576.28). I'm now 100% sure that my previous card was bad.
Long:
To confirm stability, I ran both of the 3D Mark Steel Nomad and Speed Way benchmarks for a little over 1 hour each. No crashes or issues at all. On the old card, I would crash out before the first loop even completed on both tests. I also played several DX12 titles and can confirm they did not crash out on me at all. Played each at least 30 minutes, way longer than they ever ran on the previous card.
So I'm not sure exactly what was wrong with the previous card but whatever the problem was only manifested when using the DX12 rendering API. Titles that used the DX11 API functioned just fine.
Also, something else strange I noticed is that this new card runs cooler than the problem card did under load. The temps on the new one hover around 63°-66°C. The previous card with the issues would shoot up to 70°-75°C regularly...
I guess now Best Buy and ASUS need to figure out with each other how their warranties, returns and RMAs work but my particular issue does appear to be solved now.
05-03-2025 10:42 PM
Hi @Talmen
It's not a great start to 5000 series ownership, for sure.
Can you set the GPU power target to 70% and see if the problem persists?
05-04-2025 09:10 AM
I swapped the 5080 OC Edition back in to try your suggestion. Even re-installed the driver with DDU just in case and set power target to 70% in GPU Tweak. 3D mark crashed about 26 seconds in. Tried a couple of the DX12 games I mentioned above, and they all crashed within a minute with the same settings.
Put the RTX 3080 back in my computer and now it's super stable again. Ran 3D mark for almost an hour, no crashes and the games are not crashing in the 5-10 minutes I gave them to run before quitting.
05-04-2025 11:38 AM - edited 05-04-2025 11:39 AM
I will also add that I've seen reports of 5080 owners having many problems on the 576.xx drivers and they were able to resolve them by rolling back to later 572.xx drivers. I did state above that the behavior I'm experiencing was present in all the 572.xx drivers as I did test every 572.xx driver Nvidia has on their website and my card exhibited the same issues on every driver.
I have also tried with G-Sync On/Off, DLSS On/Off and Frame Gen On/Off. Those settings do not affect the outcome at all with my card and will still crash just as reliably.
05-04-2025 12:43 PM
Thanks for testing. Before I move this post to the ROG Care for attention of service, can you confirm you've disabled any system overclocking? This includes XMP/EXPO memory profiles. Reset the CMOS and make sure to test at Optimised Defaults.
05-04-2025 01:23 PM
Yes, I've been running Optimized defaults on my MoBo for a while (and would reload optimized defaults every couple of DDU runs). The behavior is present weather or not I have the memory overclocked with the XMP/DCOP profiles... Though I have been testing with them turned off for most of the time to reduce the variables in play.
05-06-2025 06:58 PM
So I guess I'm trying to figure out what my next steps are. I'm past my return window for Best Buy so my only option now is to deal with ASUS... unless this is a driver issue which I'm not convinced it is since no drivers released that support an RTX 5080 work.
Is there a way for me to initiate a formal support or RMA ticket referencing this post (so I don't have to explain everything again)?
05-06-2025 09:45 PM - edited 05-06-2025 09:45 PM
Have you monitored the max temperatures in furmark and whilst gaming, do they seem nominal?
Are you using a riser cable? If so, be sure to test the card directly connected to the board.
However, the fact that the behaviour is identical in two completely different systems indicates an issue with the GPU.
I've reached out to CS here on the forum to assist you further.
05-07-2025 01:04 AM
I think the hottest I remember seeing the GPU get was 69°C or 70°C. It would never run long enough in a graphics load for the system to stabilize at a temp.
Furmark did strangely work. Didn't mention it before because... well I've done sooo many things trying to troubleshoot this and I just forgot to mention it. Furmark is OpenGL though. DX11 and OpenGL seemed to run stable. I can't remember the temps when running that, but it never got high enough to cause me alarm.
Other random things I've tried that I have seen suggested was:
-- Set PCI Lanes to 4.0 instead of auto
-- Set PCI Lanes to 3.0
-- Disable Resizable Bar
-- Flip between the Performance and Quite vBIOS
-- Down clock the CPU
-- Down clock the GPU
-- Set Power Limit to Max (~111%)
-- Set Power Limit to Minimum (70%)
-- Turn off D.O.C.P memory profiles
-- Load Optimized Defaults in UEFI BIOS
-- Disable all Overlays (Nvidia, Steam, Ubisoft Connect, etc...)
-- Close all running apps on PC except for the 3D App/Game
-- DDU'd computer in Safe Mode & tried every available Nvidia Driver
-- Run games on minimum settings
-- Disable DLSS & Frame Gen
-- [ Probably more things that I can't remember right now ]
I did all of this on Rig 1 and a subset of these on Rig 2 (from my OP). As Rig 2 was experiences the exact same issues, I didn't dedicate all the time needed to re-test everything from Rig 1 on Rig 2. Nothing in that list of stuff I tried seemed to affect the stability of the card. it would always crash out within a minute or two. It was a hard crash where Windows actually gave the "device removed" sound and then the "device connected" sound. The screen would then pop back on, and the Event Log showed an error about the driver crashing and the device being removed.
05-07-2025 03:25 PM - edited 05-07-2025 03:34 PM
So, update:
I tried to submit a support / repair request with ASUS on this card, and it will not let me. As soon as I put the S/N in, it tells me to take it to Best Buy (where I bought it from) for repair and warranty work and then does not let me proceed any further.
I took it to Best Buy for repair/exchange, and they were a little surprised that ASUS was sending me back to them. They had no idea what to do with it since it was outside the return window by like 6 days. I had to show them as I logged into my ASUS account that it was registered and had a warranty though 2028 through ASUS. Then I showed them the message I get when I tried to open a repair order with them. As soon as I put the S/N in, the system saw that it was purchased at Best Buy and directed me to Best Buy for warranty and repair work. The Best Buy Geek Squad employee was perplexed as he had nothing that he could do to take it in... at first anyway.
We thankfully figured out a work around with me buying a + membership which extends my return window out to 60 days and then just doing an exchange (with me waiting for stock from another store to be shipped in).
Long story short, ASUS is referring GPU's purchased from a Best Buy back to Best Buy for repair and warranty work, but Best Buy does not seem to be aware of that and would have declined my return had I not been a + member... thus putting me a weird spot where ASUS would not let me send it for repairs and Best Buy would not take it... even though ASUS shows an active warranty on it.
Once the new card comes in, I'll update this post on if it works correctly.
Image from ASUS's website showing that I have to have Best Buy Geek Squad handle repairs when entering the cards S/N.