Not looking for troubleshooting advice at this point, just an idea of whether I might have drawn the short straw this time around...

I bought a premade system with a Core i9-10900K and an RTX 3080 (PC manufacturer's design). The system began to restart randomly after about a week of use when running P3D. It steadily got worse. When the manufacturer tested it, the GPU results came back abnormal and they authorized replacement. Still waiting on that replacement.

Meanwhile, I bought a different premade system with a Ryzen 7 5800x and a PNY 3080. After about the same length of time, it started to do the same thing. Following advice on some forums, I replaced the Nvidia "game ready" drivers with "studio drivers" and dropped the boost clock speeds, and for a few hours, it looked like it was fixed, but then it started doing it again. This is where things stand as of today.

Both systems are dedicated P3D machines that are intended to be upgrades from older systems. They're running the same add-ons as the older RX 580 and GTX 1650 Super systems they were supposed to replace, but they were running 4k monitors instead of the 1080p monitors that the older systems ran. We never had any restarts like this on the older systems.

My question to the community is whether others have had quality issues with the RTX 3080. The warranty process is unsurprisingly slow for a 3080 right now. On the other hand, these systems are for professional use, and if I invest more time in troubleshooting the Nvidia and P3D configuration, I'll be pulling time away from work that could be earning income. Has anyone else running P3D with a 3080 had problems with full-on computer shutdowns, and if so, did you get the problem solved?

I have I9 10900K but with 3090 and I found out that P3D and also MSFS2020 for some reason will overheat initially the GPU followed by CPU in some conditions based on certain textures. The solution for me was to have VSYNC on and limit FPS at 30.

I have to introduce the following; my GPU and CPU were water-cooled and together on the same tubing (system).

Yesterday I replaced my GPU with one that is not water-cooled and reset the Nvidia control panel to default. I found out that CPU temps were normal and the GPU overheated, resulting in a CTD. After more testing, I turned off the EA Mode and my GPU temps went down, but still not happy. I went back to VSYNC on and limit FPS at 30 and my temps are normal now, no CTD or crash anymore.

In my opinion, there is something going on related to textures for both sims that Nvidia doesn't like it.

I had the same issue before with a TITAN RTX 24 GB video card and that setting did fix the problem for me.

Interesting to find out if some other users have the same issues with the ATI or AMD Radeon cards.

Also, there is an Asus Bios Update with Resizable BAR and Intel 10900k that I didn't try yet which is supposed to improve things.

The problem is now people are forgetting about case cooling and air flow I upgraded my MB and CPU,  but first a new case and replaced the installed fans for 4 Black High Performance NOCTUA 140mm fans, my system now runs at 60deg tops at 5.2, most of the latest PC CPU`s and GPU`s run hot and both AMD and Nvidia can pull 300w when in use that ends up getting hot inside your case. 

41 minutes ago, G-RFRY said:

The problem is now people are forgetting about case cooling and air flow I upgraded my MB and CPU,  but first a new case and replaced the installed fans for 4 Black High Performance NOCTUA 140mm fans, my system now runs at 60deg tops at 5.2, most of the latest PC CPU`s and GPU`s run hot and both AMD and Nvidia can pull 300w when in use that ends up getting hot inside your case. 

It would certainly be nice if we could count on PC builders like Alienware and Skytech to properly design their off-the-shelf systems for acceptable cooling. This is either a warranty issue or a software incompatibility issue in this instance. I'd just like to get an idea whether the cards are suffering from a higher than normal failure rate in this community.

Since I have a RTX3090 I have more fan noise even when the Sim is in background and when I´m working with airport design editor in the foreground. Only if I set the sim to pause and minimize the window, the blowing stops. That wasn´t before I got this gpu. I don´t have this all the time, therefore I wasn´t able to find out, if it´s the GPU or the CPU. I also had infrequent BSOD but was not able to find the cause. In recent time the system was stable. Windows and GPU driver were updated meanwhile. So I don´t know, if this has cured it.

I have air cooling on the RTX3090 (KFA) and VAIO cooling for the CPU. Now, without blowing, the Sim running, I have CPU temperatures from 71 to 47 on my cores, and the GPU at 54 deg. C. Unfortunately when the system starts blowing, I never thought about starting HWMonitor to find out more so far.

The 3090 throws the hot air in the case where the 20 series blow it out at the back outside the case. ( Don’t know about the 3080 )

My server has a 25cm (9,5”)  side fan in the door which helps tremendously.

My client has the door open so the hot air can escape.

A proper cooling is crucial, especially with bad designed cooling like Gerard mentioned resp. with that stupid factory OC some brands are doing. And, gentleman: don't forget your M.2 NVMe! That fast little word not alloweds get REALLY hot. 

sometimes I wish I had built a ventilated pod for my computer.  I'm dreading the summer and the hot air the 3090 plus everything else is going to kick out into my room

What about your Power supply? The RTX 30 series cards are very very hungry. I have a very noticeable increase of power consumption. 

I came from a 750 W supply.....had to go up to 1000W due PC shutoffs related to the power spikes my 10900 k and RTX 3090 had...

 

Initially, I’d suspect rebooting or crashing would be a cooling, power, driver, or OC problem in general.

 

I own an EVGA 3080 XC3 (since November) and haven’t had any issues.  I had one or two Device Removed errors after accidentally leaving P3D on the initial setup screen for a few hours, but that’s it.

Edited March 2, 20215 yr by CaptKornDog

4 minutes ago, CaptKornDog said:

I’d suspect rebooting or crashing would be a cooling, power, driver, or OC problem in general.

Indeed, after 2-3 BSOD I reset my BIOS settings to default and then again to the default 5 GHz OC. Since then it´s stable again. I have a 1000W power supply since I built this machine in 2018. It was a good decision.

OEMs tend to skimp on power supplies, both in capacity and in quality.  A system restarting itself is usually a power problem, either excessive voltage droop on a PSU power bus or bad power regulation on the motherboard.  It could also be a flaky power cable.

That said, with two different systems doing this, I'd also have a look at my line power.  In the US, the NEC standard says household circuits are expected to be able to hold 80% of their rated current in continuous use--for a typical 15A rated circuit, that's 12A, or ~1400W.  Stay above that, and the wiring starts to heat and voltage drops.  Now your computer alone won't draw that much, but that rating is for the combined total of everything on every outlet in that circuit.  It's easy to load your house wiring up enough to see a significant voltage drop at the outlet.  I helped out a friend some years back and we discovered he was getting 99v line voltage on the 115v outlet powering his computer.  Turned out he had another computer, a treadmill, large fan, and a plasma TV in the next room on the same circuit, along with lights and the usual miscellany. 

 

You can attend to all above suggestions, its definitely one or a combination of all mentioned. One critical aspect also is video memory overheating on the 3080, but especially on 3090. One NEEDS a very good airflow case for these cards to keep the video memory (GDDR6X) cooled.

7 hours ago, GSalden said:

The 3090 throws the hot air in the case where the 20 series blow it out at the back outside the case. ( Don’t know about the 3080 )

My server has a 25cm (9,5”)  side fan in the door which helps tremendously.

My client has the door open so the hot air can escape.

The Founders Edition also sends air out the back which is why I am trying to pick up an FE model of 3090 no matter what various tests of shown w/ models that pump heat right back into the case--I just think it's a fundamentally stupid design.  A 3090 only because there isn't an nV card between 10 and 24Gb VRAM availaesble in the foreseeable future.  I actually had one in the cart last week, however lost it due to lack of understanding on how to complete getting it to appear in the cart, which I learned later.   Better luck next time 

Noel -- are you sure about exhausting heat out the back of the PC case? See https://www.tomshardware.com/reviews/nvidia-geforce-rtx-3090-review/2

 

Edited March 3, 20215 yr by pgde