On HardForums, there was a report (with several photos) of an EVGA GeForce RTX 2080 Ti that abruptly caught fire and, as you might expect, stopped working. It turns out that the damage is reasonably localized, so Allyn and I compared those photos with ones from an xDevs teardown in hopes to pinpoint the most likely component. We did not have that specific card in the office.
Again, this is just our best guess from images over the course of about a half hour.
Image Credit: shansoft at HardForums
Image Credit: xDevs
Image Credit: xDevs
We marked the center of carnage with a red X on both images, which correspond to opposite sides of the PCB. As you can tell… there’s not much there. On the one side, there is an R005 resistor and what looks like two small capacitors. Capacitors, which store energy like batteries, can explode, but they look to be too small to have caused that damage. On the other side, there are a pair of 1R0 1818 inductors, another component that appears to be a capacitor, and four metal solder pads.
Our current best guess, and it’s just a guess, is that something overloaded the card (such as a shorted power phase elsewhere on the card) and that section just happened to be the part that lit up like a fuse. It wasn’t as cut and dry as we were hoping from the start (such as if we saw a giant capacitor with nothing around it) but it doesn’t look like, as some sites are saying, that the VRAM overheated or that the GPU die was defective.
this is the type of stuff i
this is the type of stuff i love reading about. goes to show how much of a rush nvidia is in to do almost nothing. again a separate add on card would have done the job, instead of making an almost “new” card. could have made a card that does ALL of the calculations for PhysX and RTX. sold it for $250. Instead they got greedy and careless.c
You know it is Nvidias fault
You know it is Nvidias fault because you are disliking them for some reason? Okay, makes sense.
I mean, i agree with you, but that’s because i don’t know better. If i were one knowing better, i would of course rather assume that the current (passive) electronic components shortage led the OEM manufacturer for EVGA to choose sub-par components. But then again, i don’t really know better, so bad Nvidia! Bad! Greedy Nvidia bad!
In all sincerity, as long as
In all sincerity, as long as only one individual card is affected, the whole thing is probably just due to a some (manufacturing) defect or a bad PSU frying the card. But that would be thinking based on common sense, and as we all know, common sense is all but common…
The images seem more like a
The images seem more like a personal fuck up than any issue with the card. I have never in almost 20 years of GPU research and testing heard of a GPU catching fire let alone so localized like it shows. Looks more like someone was messing around or even baking the card than an actual fire issue. Though in truth we will never know because if it was personal error the owner will never admit to it. Stop ragging on a brand because you saw a title and don’t care for the price of the new cards.
THIS 100% ^^^^^
THIS 100% ^^^^^
These multi-layer boards can
These multi-layer boards can be quite tricky when it comes to finding the source of the magic smoke. Any layer overheating can trigger a chain reaction melting the next layer and creating a bigger short. Seeing the board layout it could be the inductor but until a lot of similar cards go up in flames this could be a random failure.
Yeah, we realize that this
Yeah, we realize that this post doesn't really say much conclusively.
Costly GPU go POOF!
Costly GPU go POOF!
5 sided shape = pentagon
6
5 sided shape = pentagon
6 sided shspe = hexagon
8 sided shspe = octagon
2080ti onfire = myrtxgon
This post is actually heroic
This post is actually heroic
THIS 100% Owner err9r
THIS 100% Owner err9r screwing aroubd. Also in 25+ years have never heard of this happening and Ive seen some crazy stuff.
*error *around *I’ve
Where
*error *around *I’ve
Where the edit post button???? Lmao “fat thumbs”
everything just works 🙂
everything just works 🙂
Hi,
Computer engineer here,
Hi,
Computer engineer here, recent college graduate so I don’t have a lot of experience, but here’s my theory.
If this is a multi-layer PCB there’s a chance that one of the layers ended up with a lot of current through it, shorted as you said, and heated up components on the card beyond what they were spec’d to take. This may not have caused them to catch on fire directly, but could have changed their values (I.E. a capacitor or inductor changing its capacitance or inductance as it heats up, this even impacts resistors) and as such could have caused a control system in the card (a control system being something that adjusts an electrical signal based on another electrical signal) to be unable to compensate for the change.
This could have caused something similar to diesel engine runaway.
It could have just gotten so hot that it cooked the card, but these don’t look like electrolytic capacitors – they look solid-state – so I’m a bit hesitant to believe that.
Not really much in that part
Not really much in that part of the PCB to cause the fire. Just an inductor and a couple of small ceramic caps.
Without seeing the other side of the burnt PCB, it’s hard to tell for sure where the fire started. Small ceramics do fail short circuit sometimes, but with burnt spot being this big, it could have been an internal short as well.
Let’s hope the owner hasn’t voided the warranty