Yeah, I think I'll give it a miss until it's sorted. Thanks.I would avoid it if I were you. The 3080 is pretty much the same GPU as the 3090 just has some parts disabled. Also the ftw3 3080 is identical to the 3090.
Yeah, I think I'll give it a miss until it's sorted. Thanks.I would avoid it if I were you. The 3080 is pretty much the same GPU as the 3090 just has some parts disabled. Also the ftw3 3080 is identical to the 3090.
I've had a few EVGA cards brick on me so I'm absolutely not surprised by this.Reddit - Dive into anything
www.reddit.com
From the Reddit thread, seems like almost everyone affected has an EVGA card.
The EVGA Ampere cards have had power delivery issues so if it's really just them, I wouldn't be surprised. There was a big couple of threads on the EVGA forums a few months back where people were delving into this stuff and it felt like EVGA sort of shrugged and ignored it.Just a wild guess here, but since this only seems to have happened to a couple of SKUs, I'm assuming it's some power delivery/VRM issue. Either they let these cards pull so much power that something fails along the way or something is causing the VRM's temperature/current protections to not shut itself down.
Based on comments I am seeing in this thread... this should be added to the OP.
What? Devs have 0 FAULT at this issueThis is true, for sure, but it doesn't excuse what the devs have done with the menu/queuing system if I understand it correctly. Running the game world in the background of a menu screen with a forced unlocked framerate while people queue in for up to an hour is completely obnoxious. People go to a fair bit of effort tuning fan curves and optimizing airflow within their case to keep things quiet. Nobody wants to hear their system sound like a vacuum cleaner for 30-60 minutes while they wait to play a game.
They certainly wouldnt be the first to run their game world in the background though I agree its a waste of resources. It still isnt their fault a specific hardware is failing while playing their game. The game itself seems demanding on a GPU so the card would eventually fail while playing anyway.This is true, for sure, but it doesn't excuse what the devs have done with the menu/queuing system if I understand it correctly. Running the game world in the background of a menu screen with a forced unlocked framerate while people queue in for up to an hour is completely obnoxious. People go to a fair bit of effort tuning fan curves and optimizing airflow within their case to keep things quiet. Nobody wants to hear their system sound like a vacuum cleaner for 30-60 minutes while they wait to play a game.
What? Devs have 0 FAULT at this issue
Menu with uncapped framerate is normal
Its a beta, and they weren't expecting this much people playing so they didnt count for people waiting for hours
Still, Its NVIDIA FAULT
Wouldn't this be an Nvidia or Nvidia Driver problem? How the hell could a game specifically target and break one GPU?
More or less.Based on comments I am seeing in this thread... this should be added to the OP.
Have there been other games that have caused this issue with these cards.
WTF
How can a game just brick your card? Not even the most hardcore of testing software can do that, the gpu has protections that should kick in.
I wouldn't bother.Based on comments I am seeing in this thread... this should be added to the OP.
RTX 3090 require 3 x 8-pin power. Maybe these guys are using 2 8-pin power connectors from the same cable since most power supplies only come with 2 strands with 2 8-pin connectors each. The TDP is only 350 W which should be fine with 2 cables but maybe something is causing it to go much higher.Just a wild guess here, but since this only seems to have happened to a couple of SKUs, I'm assuming it's some power delivery/VRM issue. Either they let these cards pull so much power that something fails along the way or something is causing the VRM's temperature/current protections to not shut itself down.
It's very common to see people blaming software and not their own hardware. I see this on Steam forums all the time, especially for demanding games. "Developers you suck, your game makes my GPU run at 95C and overheat, this other game I played before didn't do that!!!" Although real blame in such situations is often computer building mistakes, or something as simple as dust build up and lack of maintenance. Hardware is expensive, it takes effort to troubleshoot and replace bad parts. Blaming a game and developers is easy and effortless.I wouldn't bother.
I spent 1 minute on Google and found that people have been complaining about this exact same problem for Halo Masterchief Collection with EVGA 3090 video cards since December.
If people want to believe that New World of Amazon is at fault here then that is what they will believe. No amount of "facts" will change that.
I think you hit the nail on the head... Its Amazon so people are looking at them to fail and are quick to blame them for something that literally can't be their fault.I wouldn't bother.
I spent 1 minute on Google and found that people have been complaining about this exact same problem for Halo Masterchief Collection with EVGA 3090 video cards since December.
If people want to believe that New World of Amazon is at fault here then that is what they will believe. No amount of "facts" will change that.
There's a picture that, if you make it your wallpaper, will brick some phones.Kinda scary how you can create a software that will immediately fry your hardware if you run it.
Kinda scary how you can create a software that will immediately fry your hardware if you run it.
Bad advice. If it can't run or dies at stock settings RMA that shit asap before you are out of warranty period!Given how my EVGA FTW3 card ends up throttling itself on stock settings, this is not surprising news to me.
Undervolt/turn down your power limits, folks.
Oh, it still runs. It just has to push its clocks down significantly in order to keep below the thermal limit.Bad advice. If it can't run or dies at stock settings RMA that shit asap before you are out of warranty period!
Hell no. Stress test that shit and RMA it if it fails.Given how my EVGA FTW3 card ends up throttling itself on stock settings, this is not surprising news to me.
Undervolt/turn down your power limits, folks.
Oh dear, it's EVGA.... I don't know why that brand is so liked by some people. They have some sort of serious bricking or GPU dying problem every generation.
Given how my EVGA FTW3 card ends up throttling itself on stock settings, this is not surprising news to me.
Undervolt/turn down your power limits, folks.
Imagine being a tech youtuber with two or three 3090s at most and your subscribers are demanding you test this issue.
FWIW, this seems to be a problem specifically with the EVGA RTX 3090 FTW3 Ultra and not any other variant of the 3090. That specific card appears to have a defect causing this issue. If you have any other 3090 I wouldn't be concerned.Holy shit what, I just downloaded this game last night. Guess I'm not playing it
FWIW, this seems to be a problem specifically with the EVGA RTX 3090 FTW3 Ultra and not any other variant of the 3090. That specific card appears to have a defect causing this issue. If you have any other 3090 I wouldn't be concerned.
Was there more than one Gigabyte? The only one I saw was the one post on their forum. Could just be a coincidence.There has been reports of Gigabyte 3090s also turning into 2.5k € paperweights.
Whoa what? You got any info or a link about this?There's a picture that, if you make it your wallpaper, will brick some phones.
Shit can be wild!
FWIW, this seems to be a problem specifically with the EVGA RTX 3090 FTW3 Ultra and not any other variant of the 3090. That specific card appears to have a defect causing this issue. If you have any other 3090 I wouldn't be concerned.
Sure thing!
My 3080 goes ham in the menu though and it really fucking sucks with those queue times. My room heats up by a few degrees (it gets really hot in here) because of that :(