Tyehn22

TWO...died.. what PSU are you using?  also.. are you using a UPS?

 
Two different PSU's. First was time was a Thermaltake Toughpower 1200W, this time was a Be Quiet Dark Power Pro 12 1500W. Both are on conditioned power. 

ty_ger07
I'm sorry to hear.
Try to get EVGA to cover your fees.

neteng101

dalten22
I'm deathly curious what is causing most of the failures. Would be interesting if you sent it to a real engineer or youtuber who understood PCBs really well.

The information is out there although not in a single place.  Check out Buildzoid's PCB analysis for the 3080 FTW3 he did for Gamers Nexus, his own PCB analysis of the 3080 Zotac card, and the super long discussion threads on the 3080 and 3090 on OCN.  These cards are crippled by the uP9511 PWM controller and the microcode that was unscrambled from the FTW3 indicated the cards were originally designed for a 2x8-pin plus 1x6-pin power input instead of the final 3x8-pin configuration that went to production.
 
Two problems with this - the analog uP9511 relies on a back signal from the GPU to tell it what voltage to output to the GPU, and the power input was never changed so the 3rd 8-pin consistently underdraws power (its running like its a 6-pin, which indicates the components in the input path was not modified, only the connector was changed).  The higher power limits of the FTW3 cards means when they boost, they can run at higher voltages > 1.050v.  That's not a problem normally, but there's voltage overshoot that few people know - in short as the GPU starts demanding more voltage to boost, the initial surge goes beyond the target voltage.  If you want to see voltage overshoot (can only be seen with an oscilloscope and not in monitoring software) and understand why both overshoot/undershoot is problematic, there's a bunch of videos from Buildzoid on the topic but you'll have to spend a lot of time watching his videos (he tends to ramble a lot so it takes some patience to really get an understanding).
 
Until EVGA redesigns the 3080/3090 FTW3 cards, to me its a matter of when, not if these cards go...  older games which tend not to run the cards at steady state maxed out causes more power state changes and kaboom there goes yet another 3080/3090 FTW3.  The 3090 FTW3 is especially vulnerable.  The only thing anyone can do with these cards is to undervolt them conservatively, this should help limit any overshoot to a level that hopefully doesn't damage the card.  They also need to redesign the input to balance out the 3 PCIE power inputs, and use a better PWM controller.
 
Its interesting the new EVGA 3060 XC3 actually uses the uP9512 instead of the 9511 EVGA has been using a ton in Ampere cards...  not sure if Nvidia actually changed the reference design to require the 9512 for the 3060, but that's a possibility too.

flg2010
TWO...died.. what PSU are you using?  also.. are you using a UPS?

ivangogh

ty_ger07
I'm sorry to hear.
Try to get EVGA to cover your fees.

 
How does one approach them about that?  
I tried asking EVGA to consider the high shipping cost for my first RMA.  Not a word in response to the ask.  It cost over $200 to send from Canada.

Tyehn22
Went to go play an older game (Alien Isolation) and my system shut down and video never came back. At least this one last for a few months. Looks like I have to wait until Monday to get the cross ship process started. Again. 

ivangogh

atfrico
Your replacement might be failing for having cheap capacitors installed in the board, please see below video, very interesting.
 

I'm pretty sure this specific model has these parts alright.

The clip talks about an early-discovered issue that got dealt with through drivers. Still unfortunate for owners, but as far as I know there was a fix.