Hey,
 
I've bought a 2080Ti Black (2281-KR) back in April without knowing about all that A vs Non-A desaster.
Today I sat down and did some benchmarks and it seems that I got the absolute low-end of these low-end chips...
 
Benchmarks
Firestrike Ultra: 3092 ()
Firestrike: 11736 ()
Those results put it in GTX970 or 1060 territory...
 
Weird Behaviour
The GPU only boosts in very short bursts (2-3 sec) to around 1600-1800 (sometimes even 2055) and then falls back down to 1350MHz.
Power consumption caps out at around 50% or 135 Watts. (reported with GPU-Z, PrecisionX1, HWMonitor)
Here is a GPU-Z log during a run of Firestrike Ultra (score: 3092)
 
System
CPU: 9900K
RAM: GSkill Trident Z 3600MHz, 32GB
Board: Gigabyte Aorus Master Z390
PSU: AX1200i (using seperate cables/rails; delivers solid 12.03 - 12.06V)
Cooling: custom watercooling (GPU 45°C idle / 60°C under load)
 
Step up
Back in may I wasn't aware of this awful performance numbers but when I found out about the Non-A chips I wanted to use my step up for an 2282-KR anyway.
But unfortunately I was on vacation while the step up was approved and it expired without me having a chance to accept it.
So sadly that is not an option anymore.
 
RMA the only option?
Do you guys have any idea what I could try before opening an RMA?

Gave it another try with all 7 fans and the pump on max which resulted in around 50°C under load and 4408 points in Firestrike Ultra.
Thats better but still not even close to my 1080 Non-Ti I've had before. (which delivered this performance at 60-70°C with no problem)
 
And the 2080Ti in my understanding should also behave like that. I mean what's the point of setting a temp target of 80°C when the card throttles itself anyway?
 
Power consumption still caps at about 57% or 140 Watts at a merely 50-55°C
>> Thats still odd... Why is that?
 
In tests on various sites (guru3d, ...) the card ran at 70°C+ and gave them well over 8000 points in Firestrike Ultra.
So what am I doing wrong? I mean its not my first build but never had problems like this... (2x 1080, 2x Titan, 690, ...)
 
Or is it because they've tested the Founder Edition version (which is a A chip) and it performs THAT much better?
This would put the whole A vs Non-A on the edge of false advertising... I mean at such a huge difference (>50%) they shouldn't have labeled them both as "2080Ti".

When running the gpu-z render test does your bus interface run at pci-e 3.0 x16?