In menu screens, loading into multiplayer. Within 3 seconds, the PC shuts down in a weird way. Power and RGB for some components are still on. After turning off PSU and turning it back on, the system refuses to boot up (goes into auto protect mode every time). Only when I disconnect all 3 pci-e power cables will it boot up. Pretty sure it's dead - I have a Rev 0.1 card. Going to submit an RMA, but I heard some rumors that EVGA temporarily stopped accepting 3090's. Is this true, and if so, what would be the turnaround time?
 
Mobo: Asus Prime X570 Pro
CPU: AMD 5900X
RAM: 32 GB DDR4-3200 Trident Z Royal
Drives: 2 NVMe, 3 Sata SSD
PSU: Seasonic Focus GX-850
Case: Cooler Master Mastercase H500 ARGB Airflow
GPU: EVGA 3090 FTW3 Gaming (non-ultra)
 

You aren't the first. Other people were having their 3090s die in that game a year ago. It's not the game's fault. The fault is how the video card mishandles transitions from low load to higher load.

Used my EVGA 3090 FTW3 ULTRA HYBRID. For a year just fine in any game you can think of maxed out no issues. World of warships latest graphics update bricked it when the port loaded........ and then my replacement RMA also died in world of warships port. My 3rd card should get here Monday we shall see how unlucky I really am with build quality QC on a $1800 product

Forza Horizon 5 maxxed out causes the fan controller to request 3 million fan rpms as reported in GeForce Experience. I knew something was starting to give up the ghost.

ty_ger07
You aren't the first. Other people were having their 3090s die in that game a year ago. It's not the game's fault. The fault is how the video card mishandles transitions from low load to higher load.

This isnt exclusive to 3090's.....folks with 3080 n 3080Ti also suffering this same power delivery / spike ...whatever issue....obviously an Engineering problem with these 30xx FTW3 series Cards...

veganfanatic

rjbarker

ty_ger07
You aren't the first. Other people were having their 3090s die in that game a year ago. It's not the game's fault. The fault is how the video card mishandles transitions from low load to higher load.

This isnt exclusive to 3090's.....folks with 3080 n 3080Ti also suffering this same power delivery / spike ...whatever issue....obviously an Engineering problem with these 30xx FTW3 series Cards...

Wonder what is up all these disparate models of cards?  Wonder if the fan control logic is borked?
 

I do want to mention that I have a Rev 0.1 card (received and installed 9/25/2020) so it's part of the same batch as all of the other dead 3090s out there. It unfortunately was a late bloomer that died later than the others. Heard Rev 2.0 is fine, but I could be wrong. In any case, I can finally get rid of the red lips! 

animefreeks
Forza Horizon 5 maxxed out causes the fan controller to request 3 million fan rpms as reported in GeForce Experience. I knew something was starting to give up the ghost.

B O G U S

Anyone who understands fan control knows that is utter bogus.

Completely and totally unrelated. Is a software reading error. Will not cause hardware failure.


What is causing hardware failure is power spiked. NVIDIA power monitoring controls average power. Instantaneous power spikes are allowed and can cause the VRM to fail. Low demanding games cause the card to boost high at low power, and that is the perfect recipe for sudden demand to fry the VRM.

ty_ger07

animefreeks
Forza Horizon 5 maxxed out causes the fan controller to request 3 million fan rpms as reported in GeForce Experience. I knew something was starting to give up the ghost.

B O G U S

Anyone who understands fan control knows that is utter bogus.

Completely and totally unrelated. Is a software reading error. Will not cause hardware failure.


What is causing hardware failure is power spiked. NVIDIA power monitoring controls average power. Instantaneous power spikes are allowed and can cause the VRM to fail. Low demanding games cause the card to boost high at low power, and that is the perfect recipe for sudden demand to fry the VRM.

Yeah I agree...after seeing / hearing mine go "pop" and die in the initial boot sequence points to this for sure. As I said prior, thankfully AX1600i immediately went into safe / shutdown mode, so less risk of frying other components.
With my 1080Ti re-installed (with WB) system seems fine, so no collateral damage ;)
 
Testing my RMA replacement Card today, will provide results in a separate thread.

rjbarker

ty_ger07

animefreeks
Forza Horizon 5 maxxed out causes the fan controller to request 3 million fan rpms as reported in GeForce Experience. I knew something was starting to give up the ghost.

B O G U S

Anyone who understands fan control knows that is utter bogus.

Completely and totally unrelated. Is a software reading error. Will not cause hardware failure.


What is causing hardware failure is power spiked. NVIDIA power monitoring controls average power. Instantaneous power spikes are allowed and can cause the VRM to fail. Low demanding games cause the card to boost high at low power, and that is the perfect recipe for sudden demand to fry the VRM.

Yeah I agree...after seeing / hearing mine go "pop" and die in the initial boot sequence points to this for sure. As I said prior, thankfully AX1600i immediately went into safe / shutdown mode, so less risk of frying other components.
With my 1080Ti re-installed (with WB) system seems fine, so no collateral damage ;)
 
Testing my RMA replacement Card today, will provide results in a separate thread.

yep, the seasonic did the same. went immediately into shutdown mode. installed an old R9 290 borrowed from a friend and it booted up just fine.

BTW, I was saying bogus about the fan control thing because some reporter started that nonsense theory. I'm not trying to make fun of the OP for hearing that theory and believing it. It's been conclusively debunked. We know why the cards are failing, and its not because the fan controller is trying to make the fan spin 3000000 RPM.

veganfanatic
Boost on video cards can gobble a lot of extra power. This is why I use high-end power supplies which have reserves to handle video card proclivities.

That makes zero difference when the card decides to commit suicide. Power available isn't the problem. The card failing to limit its power consumption -- and frying its VRM in the process -- is the problem.

veganfanatic
The PSU trip suggests a short circuit on the card. That means a catastrophic failure has occurred.

Agreed. As I said. The VRM has fried. This has been researched. The VRM OCP is set way too high to protect itself. (Probably because NVIDIA realized that its power monitoring was too slow to prevent OCP triggering, and OCP triggering too often would annoy people.) The power monitoring circuitry/firmware/driver system is way too slow (and designed to limit average power, not peak power) to protect the VRM from repetitive power spikes that eventually fry it. These low load games are the perfect recipe for killing these cards, and are the games we observe killing these cards. It's a 3d load which will cause the card to boost, the driver knows that you want the card to boost, the card idles along at high boost, and then if there is a sudden load which requires more calculations, the high boost state in conjunction with a high computation requirement causes a high power spike. And zap, it's done. OCP may trigger, shutting the card off, and that would cause the fan to ramp up to 100%, but it might be too late and the damage might have already been done. You turn the computer off, turn it back on, you hear a fuse pop, and that's that. Some times OCP triggers and the card survives, but the next time it happens, you might not be so lucky.

It is a design problem. Cards pushing high power limits with non-beefed up VRM (reference design) are most likely to die. (FTW3s, for example)

veganfanatic
The PSU trip suggests a short circuit on the card. That means a catastrophic failure has occurred.

No kidding !!!!! ;) As said I heard the pop....I was actually looking right at the Mobo and my system when it happened as whenever I have issues I always watch the "post codes" on my Motherboard, was actually watching to see if I got hung up on another "97 post code".....obviously when you hear a "pop" from your hw its not good :)
The Corsair AX1600i has an led on it that is lit green when all good, it immediately went red at the same time the GPU "popped" ....pretty obvious.
I knew it hadnt taken out any hw as was able to reboot and check system using iGPU....
 
The only thing I dont like with this new Maximus Z690 Apex board for my next build is no iGPU input.

RogueMaster

animefreeks
I do want to mention that I have a Rev 0.1 card (received and installed 9/25/2020) so it's part of the same batch as all of the other dead 3090s out there. It unfortunately was a late bloomer that died later than the others. Heard Rev 2.0 is fine, but I could be wrong. In any case, I can finally get rid of the red lips! 

Wait..  There are Rev 2.0 cards now?

Heard unconfirmed reports on Discord, so take with a grain of salt. I'll find out definitely for sure in a few days.

animefreeks
Replacement 3090 arrived, and what the heck? It's a Rev 0.1 card, AND it has the red lips. EVGA sent me another ticking time bomb.

O_O

TacticalBeard
Used my EVGA 3090 FTW3 ULTRA HYBRID. For a year just fine in any game you can think of maxed out no issues. World of warships latest graphics update bricked it when the port loaded........ and then my replacement RMA also died in world of warships port. My 3rd card should get here Monday we shall see how unlucky I really am with build quality QC on a $1800 product

What I want to know is, are you going to try World of Warships again?

Is this your video (WOWs bricking the user's 3rd 3090)?
 
https://www.youtube.com/watch?v=oXmWXyuMKYg