Hi.
 
I would like to know if someone who is using the same GPU (EVGA 3090 FTW3 Gaming/Ultra) came across power usage or core load drops that happen infrequently. Problems while mining might not be relevant in my case, so I am asking mostly gamers/testers. I have also check forum topic regarding New World but those doesn't look like the same issue.
 
For a past couple of months I have been testing everything I could think of and I still cannot figure out what might be the problem with my GPU. In different games I have tested drops are varying in frequency and amplitude and they always are accompanied by dips in fps, memory load controller and core clock.
 
To put this into perspective, when I was gaming using GTX1080Ti, I get stable 60-80fps in Death Stranding. In the same location in game, when using RTX3090 I get maximum of 100-120fps that drops to 70-60 or sometimes even as low as 30fps, usually every few/dozen seconds.
 
In DOOM Eternal, drops are not as significant but fps changes in something like sinusoidal wave between 100-200fps all the time. In comparison, when using 1080Ti, fps is stable.
 
Those were extreme examples but I have noticed drops in every title I have "tested" so far: Horizon ZD, RDR2, DOOM 2016 (both DX12 and Vulkan, but Vulkan seemed more stable), Metro Exodus, Cyberpunk, Wicher 3 and GTA V.
 
While playing any game, by using HWiNFO64 as a monitoring and graphing tool I can observe drops in Framerate, GPU: Power, Core Load, Memory Controller Load, (PCIe +12V, 8-pin #1-3) Input Power, Clock, Effective Clock and Total GPU Power (both % of TDP), while GPU Core Voltage has spikes in exact same times. There are also irregularities in Reliability and Max Operating Voltages but those spikes don't coincide with aforementioned entities.
 
What I have tested and, to my abilities, confirmed so far:

• this is not a external (to GPU) power usage issue: in the same hardware set up, I was using 2x1080Ti in SLI that was using regularly at least 150W of power more than 3090; 
• this is not a OCP and transient spikes issue: I use EVGA P2 1000W PSU but to confirm that I also used Corsair AX1600i;
• this is not a external cooling issue: I am not using any case (only bench table) with ambient temp varying between 21-24°C (70-75°F); everything else is water cooled, I have additionally put 2x120mm fans to blow on the backplate and GPU has fans at 100%;
• this is not a driver/PX1/monitoring software issue: it happens on different driver and Precision X1 versions (even latest), w/ or w/o HWiNFO/RTSS tools running;
• this is not OS issue: I was using Windows 10 Pro (21H2) but I have also confirmed it on Windows Server 2019 and clean installation of Windows 10 Pro;
• possibly, this is not high VRAM usage issue: those games use 10-60% VRAM (Horizon ZD in 6880×2880 uses even more);
• possibly, this is not VBIOS or GPU version issue: both are newest (94.02.42.C0.05 and FTW3 ULTRA V2) and it happens in Normal and OC mode;
• finally, it is not an issue with my PC: there were no such problems while using one or both 1080Tis, CPU and RAM is not OCed, everything has more than sufficient power and temperature headroom, but most importantly the same problem occurs when I used 3090 in ASUS RIVBE with 1680v2 on board (although I could only test it in DOOM Eternal and Horizon ZD).

 
Also, GPU is not overclocked (other than FTW3 Ultra factory overclock), core voltage has no problem showing 1.081v and GPU power can climb to 450W.
 
When those drops does not occur? In FurMark and in 3DMark...
 
I bought this card 3 months ago on EVGA EU website and because I do not recollect any situation that might be the starting point for those issues I believe this card behaves that way from the beginning .
 
I have already invested so much time in diagnosing this issue that maybe I have missed something very basic. If someone might have an idea what might be the reason, please do share.
 
Thanks for taking the time to read my (unfinished and without proper editorial) story.

Main PC:

• CPU: 10900k (different frequencies tested: from stock 4.9->5.3 Turbo to 5.4GHz all cores) - every configuration tested with Linpack using MKL from 2022 w/o errors
• RAM: 64GB Crucial Ballistix at stock 4000MHz, XMP I, memtested for few hours, w/o errors
• MB: ASUS Maximus XII Formula, I have tested 2 different PCIe slots
• SSD NVMe: WD Black SN850 and Samsung 980 Pro (different disks for different systems)

In my list of tested games, I have encountered high (Death Stranding, Horizon Zero Dawn) to very low CPU usage (like DOOM 2016 or DOOM Eternal in menu).
 
In Death Stranding is more like 20% lows. This game runs much, much smoother with 1080Ti installed.

Depending on a game but it‘s usually between 4K-6K.
The greater the resolution the greater the problems - drops are more apparent.
Death Stranding is supposed to be getting 120fps avg in 4K and my results are more or less the same. Except I am getting 20fps 0.1% lows.

I assumed you meant to test with slower RAM, so I just changed multiplier w/o messing with timings.
 
In DOOM Eternal (menu) it looks the same, maybe few fps lower. In Death Stranding drops are still huge but I couldn't say if they are bigger or smaller. In 3DMark I am getting 400 marks lower GPU score and for the first time I am observing lower GPU utilization in GPU Test #1 (GPU #2 was falling to 94% GPU core load).
 

x16 3.0. It is 8GT/s all the time

CPU/RAM bottlenecking was my first thought when I was playing Horizon ZD. But after many different tests I just couldn't justify it considering the performance of 10900K. What video game 10900K at 5.2GHz can bottleneck at 4K? And remember, "the greater the resolution the greater" drops I was observing.
Disks are also fine. Both were new just few months ago and both have no more than 10TB r/w with 100% life remaining (as per SMART).
I tested this card with completely different components and drops were still there: "the same problem occurs when I used 3090 in ASUS RIVBE with 1680v2 on board (although I could only test it in DOOM Eternal and Horizon ZD)".

Have you tried the computer in another room to see if the spike might be related to other devices turning on to draw power?

What happened to the idea of a virus?? Melware or something trying to use your pc to mine in the back ground without your knowledge.

Hi Sheeler,
 
through your last RMA´s, the products have been tested "Passed", without any issue - do you have the chance to test your componentes on a different system as well, specially the PSU? Additonal we wrote you an email, please reply back to supporteu@evga.com
 
thx,
 
Markus