ChaseRandall


I just changed my gear out to a new case, and plugged all 3 power cables in, and when I booted up one of them had a red light over them and wouldn't start up. I switched the cables around a bit and it started up just fine. What do you think?

novacane111
I'll just throw this at the wall and hope it helps in anyone's troubleshooting or exploration of the issue.
Just finished playing NW for the night, had GPU-Z recording for the session (about 2.5 hrs). Here's some of what I encountered.
 
3090 FTW3 Ultra (+1000 mem, +55 clock, Target and voltage maxed, fan curve 100% @ 65C)
8700K @ 5.2Ghz
3200Mhz mem
RMx1000 PSU
I have vsync and gsync enabled in Nvidia CP. Capped at monitor 60hz refresh.
4K resolution using the "Very High" preset with little else changed from that. (vsync disabled in game)
 
Board Power Draw
Max: 497.2W (PCIE 80.6W, 8-pin #1 146.5W, #2 139.3 W, #3 118.4W)
Average: 397.3W
 
GPU Voltage
Max: 1.1v (Occurred for multiple readings at various power draws between 225w - 430w)
Average: 1.056v
 
Temperatures
GPU Max: 73.4C
Memory Max: 84C
GPU Average: 65.9C
Memory Average: 77.5C
 
Seeing these numbers, there may be some merit to Jay's thesis about using the rabbit ear'd cables and PSU's rated towards the recommended minimum coupled with an OC'd CPU and GPU. New World seems to tax both the GPU and CPU more than any recent game I know of. (CPU average temp for me was 61.3C, typically I see the 40-50's for most games)
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.
I'm no electrical engineer but noisy power from a maxed out PSU to a GPU running full tilt sounds like a plausible cause.
 

novacane111
I'll just throw this at the wall and hope it helps in anyone's troubleshooting or exploration of the issue.
Just finished playing NW for the night, had GPU-Z recording for the session (about 2.5 hrs). Here's some of what I encountered.
 
3090 FTW3 Ultra (+1000 mem, +55 clock, Target and voltage maxed, fan curve 100% @ 65C)
8700K @ 5.2Ghz
3200Mhz mem
RMx1000 PSU
I have vsync and gsync enabled in Nvidia CP. Capped at monitor 60hz refresh.
4K resolution using the "Very High" preset with little else changed from that. (vsync disabled in game)
 
Board Power Draw
Max: 497.2W (PCIE 80.6W, 8-pin #1 146.5W, #2 139.3 W, #3 118.4W)
Average: 397.3W
 
GPU Voltage
Max: 1.1v (Occurred for multiple readings at various power draws between 225w - 430w)
Average: 1.056v
 
Temperatures
GPU Max: 73.4C
Memory Max: 84C
GPU Average: 65.9C
Memory Average: 77.5C
 
Seeing these numbers, there may be some merit to Jay's thesis about using the rabbit ear'd cables and PSU's rated towards the recommended minimum coupled with an OC'd CPU and GPU. New World seems to tax both the GPU and CPU more than any recent game I know of. (CPU average temp for me was 61.3C, typically I see the 40-50's for most games)
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.
I'm no electrical engineer but noisy power from a maxed out PSU to a GPU running full tilt sounds like a plausible cause.
 

kevinc313

moomeacow
who the hell here is running a 3090 and daisy chaining their pcie cables??

Random people on Reddit.  Purely by coincidence and in no way possibly related or due to user error, there are also random people on Reddit with blown GPU's.
 
As a side note, here is a very good guide to the limits of PSU power cables, geared toward miners, the kings of daisychain PSU cables:
 
https://www.gpuminingresources.com/p/psu-cables.html
 
288w per cable.
 
While I've not been able to confirm it, I've seen indications that some PSU's have per port protect circuitry. YMMV.  I always run individual cables.
 
 

tyranus7
If that were the case (which I don't think it is), shouldn't the GPU manufacturer state on their power requirements that the required PSU should have a ripple smaller than x%.? If the ripple could actually kill a GPU, EVGA should clearly state that as a hard requirement, which is not the case. Now let's take a look at the Asus Strix as an example, that card has tons of capacitors with high capacitance to filter the voltage coming from the PSU, that helps to reduce fluctuations on input voltage. Sadly EVGA doesn't include enough capacitor and that's also on their end. The only way a PSU would kill a video card is if by some reason the PSU had a huge voltage spike and at the same time the ability to deliver huge amount of current, which is also unlikely given that PSU also has fuses to prevent this. Bad quality PSUs will most likely just shut it self down because it's unable to sustain the required voltage at a required current, i.e. it can't deliver the promised power.
 
As I said before there is absolutely nothing that a software could do to burn a good electronically/vBIOS designed GPU. A GPU don't know that it's rendering a video game, a GPU don't know how many fps is delivering a GPU only do mathematics calculations and return bits to the CPU, that's all. If a GPU is burning that's on the hardware level not the software level.

squall-leonhart

Intoxicus

deathlessdd
Class action Lawsuit incoming? Game kills 100s of 3090s for unknown reasons...

Do people not realize that part of the EULA/ToS is a section where you agreed they're not liable for these kinds of things?
It's super duper standard to include clauses like that.

Those clauses are not enforceable.

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

lobstar
Well, I just bought the game.  I've got a Rev1.0 so I'm not expecting any issues.