I just checked and my 3080 is Rev 0.1, should I be worried? It has been fine so far.

squall-leonhart

I'm no electrical engineer.

 
This much is obvious, because
 

novacane111
Using rabbit ear'd PCIE cables would most certainly go over the 150w design spec. Two 8-pins rabbit-ear'd could easily approach 280-300W on a single PCIE 8-pin cable during game play.

There is no 150w design spec for the cables
8pin cables must provide a minimum of 150w as the 8pin Input on the card has the capability of drawing up to 150w per socket.
 
But most of the cables are specced for up to 280w(or higher as cable grade and insulation increases), hence why the official stance for daisy chaining is that its ok to run #2 and #3 on one cable with a quality, brand name unit (that isn't from the cheaper end, which usually ship with 8pin-8pin only peg cables) - and in normal cases, the 3 8pins balance 130w each and draw only about 65-70 on the slot.
 
Your card is a timebomb waiting to go off, pin #3 is too low and slot draw is exceeding spec.
 

this seems to be my therory as well, with the card not load balancing properly between then #3 8pin annd the Pcie causing the pcie to supply more AMPs / Volts to the circuitry downstream which is only designed to handle the spec 75W 
 

so buildzoid came out with a video stating that it could possibly be the Poor vrm design with the over current protection and over voltage protection feature on the Voltage controllers not being implemented properly https://www.youtube.com/w...lyHardcoreOverclocking

I've now had the following GPU owners express they have had shut downs and failures with New World...
RX590
6800
6800XT
6900XT
3080Ti
3090
So once again, the issue definitely is with SOMETHING in the way the game New World is rendering. This ISNT a 3090'exclusive issue! PERIOD!!

denny88
Looks like FTW3 poor design and cheap components are confirmed:
 
https://www.youtube.com/watch?v=dbCrNBpW-RQ

Thanks for posting the video.
 
At this point is all speculation, but it makes sense.
 

• Buildzoid agrees that cards exploding is not NW's fault but Nvidia's and AIB's fault.
• When users reported that they get a black screen and fans at 100% and they has to reboot the system to get video signal again, that's the a VRM "over current protection" kicking in. This is sorta of the second last protection before a VRM fuse blows. A fuse blowing will produce the popping sound. The 3090 has 3 different power plains, one for memory, other for cache and other for the core, every plain has it's own set of VRM. Which OCP is kicking in is uncertain at this point.
• The interesting theory by buildzoid is you can have an scenario where the core isn't being heavily utilized but let's said the cache is being heavily utilized. This will actually happens on the game menu, where the cache is being frequently required but the core is almost inactive. This would lead to an scenario where the total power consumption will be near close to 450W but the card might be drawing most of it's power towards the cache power stages and if the OCP of the VRM are set too low you can actually blow a fuse before triggering the OCP. That's a design flaw by the AIB but also Nvidia's responsibility given they approved the AIB design in the first place.

EVGA sure makes some crappy cards, this keeps happening again and again.

denny88
Hmm.. this can be ICX failure
 
https://www.igorslab.de/e...n-caused-evga-problem/

I read the whole article, it doesn't detail much other than Igor's insight information which might or might not be true. If it's true he said that the burnt area is the area from the controller of EVGA's fan IC. If that happens to be true... oh boy.. the failure rate of EVGA 30-series is going to be catastrophic over the years.

Outofstock4ever
So to be clear people blaming the game are just being insanely silly, the game has no power to destroy a gpu, any gpu in 2021. If it happens is because someone in the hardware side (nvidia, aib's) didn't do their job. Period.

not sure about this claim. I agree that SW should not be able to brick your HW. In this case it seams the keep unlimited FPS at some stage (Menu/Setting) is killer or at least trouble maker for another cards as well.
I guess the best solution would be that card should not render more that 300 FPS (right now in MENU/Setting you can make few thousands FPS which well could be trouble some).
Probably NVIDIA will put such hard FPS limit to avoid this in future.
 
Summary:
- HW should be able to handle such situation or at least shut down safely
- Driver provider should limit maximum FPS in manner not overloading HW
- right now it seams bricked cards are mostly EVGA FTW3 not sure if this covers all variants (water cooled/hybrid)
 
I will probably install MMC today to see if I kill my card or not :-) worst case is RMA so no BIG deal as EVGA is good in RMA.
 
I wonder if those dead cards were rev 0.1 or even 1.0.

I have been playing this a bit about 10 hours each on a 3090 Strix blocked, and a 6900XT Ultimate blocked.
At 4K hitting between 90-120 FPS and yeah, crazy temps.
 
Just limiting to 60 FPS now, Temps on the 6900XTU went from 53 core 83 hotspot! to 38 core 42 hotspot.
it IS about 77F in room though.
 
gonna see how it plays on kingpin system only because it has the 11900K.. 
 

denny88
Looks like FTW3 poor design and cheap components are confirmed:
 
https://www.youtube.com/watch?v=dbCrNBpW-RQ

Thanks for posting.  Buildzoid has a better understanding of GPU design than anyone posting here.

Meguas2

denny88
Looks like FTW3 poor design and cheap components are confirmed:
 
 

 
 
I think you need to rewatch the video because nowhere does he say it's specific to EVGA.  If anything it's nvidia as a whole not properly implementing OCP... and even that per buildzoid is speculation.

They should not have released a video card that does 350w in reference trim and 420w in AIB custom.  Especially with peaky power draw. 

funny that the most expensive card is also the one with the highest failure rates now

The game has stated an announce that they would cap (in a patch) the FPS on menus, so maybe people trying to reproduce the fault will not be able to do it anymore.
Anyway, I agree with people that states, that there's NO way that software in any kind could kill hardware, if it does, then it is a hardware fault.

Hoggle

denny88
Looks like FTW3 poor design and cheap components are confirmed:
 
https://www.youtube.com/watch?v=dbCrNBpW-RQ

We would have seen a pattern beyond just one game if this was true. Cheap components and poor design would be something that would show up in benchmarking and overlocking tests and really if you look at reviewers they are not issuing a warning that the FTW3 is a poor choice for overlocking compared to another company. Also if it was NVIDIA using poor parts and design you would think that AMD would be able to catch up.

spider1701

Outofstock4ever

spider1701
 
not sure about this claim. I agree that SW should not be able to brick your HW. In this case it seams the keep unlimited FPS at some stage (Menu/Setting) is killer or at least trouble maker for another cards as well.
I guess the best solution would be that card should not render more that 300 FPS (right now in MENU/Setting you can make few thousands FPS which well could be trouble some).
Probably NVIDIA will put such hard FPS limit to avoid this in future.
 
Summary:
- HW should be able to handle such situation or at least shut down safely
- Driver provider should limit maximum FPS in manner not overloading HW
- right now it seams bricked cards are mostly EVGA FTW3 not sure if this covers all variants (water cooled/hybrid)
 
I will probably install MMC today to see if I kill my card or not :-) worst case is RMA so no BIG deal as EVGA is good in RMA.
 
I wonder if those dead cards were rev 0.1 or even 1.0.

you can put that the way you want, but this is in no way the game's fault. The card should hardware based WORKING measures to avoid a catastrophic failure. They all do, but apparently they just aren't working in this cards. This is not even the 1st game that was uncapped fps.
Going away from games, every benchmark is made to max the cards, what would be the point to limit them? Should syntetic benchmarks start caping fps too?

none of the synthetic benchmarks generates 1000+ FPS in given resolution. Please note that benchmarks are build measure card by given set of capabilities and there is no way to overload card during bench. For reference in the past FURMARK was killing cards too until driver makers made specifically filters for it.
Also please note that I have not told that it it is not HW fault.
 
To be honest it is both HW and SW. It could be even improper NVIDIA design of platform as whole.
One thing is certain it is SW causing the card to die in conjunction with HW/Driver design (so if you want NW than keep FPS limiter on)

I agree it is clearly a highly abnormal load situation that any normal stress testing regimen would not replicate, exposing a weakness in the card, possibly vram temps and/or vram power delivery.  
 
BTW, the ancient ATI tool 3d render will let you run 6000fps at about half power.  People use it to test max stable core clock.

kevinc313
 

I agree it is clearly a highly abnormal load situation that any normal stress testing regimen would not replicate, exposing a weakness in the card, possibly vram temps and/or vram power delivery.  
 
BTW, the ancient ATI tool 3d render will let you run 6000fps at about half power.  People use it to test max stable core clock.

cool... at least is is on 50% power :-) (I am tempted to try it.. any download link for it?)
 
What makes me wonder Amazon claims it is not their fault yet they are quickly deploying patch limit FPS in menu :-)
this will be interesting to follow up in the future what outcome there will be for them.