Aorus Ultra x570 - all gpu's underperforming

Soldato
Joined
11 Jun 2003
Posts
5,010
Location
Sheffield, UK
So... I initially though this was my 2080ti being weird/having issues.

A friend with the same setup (3800X, aorus pro in has case) and a 2080ti, similar memory timings/etc hits about 13800 on timespy.

I'm struggling to break 11000 most runs, which is.... more like 1080ti performance levels. Quite a disparity.

I DO use all 3 M.2 slots generally. Have tried it (in below tests) with just the 1 drive though and seems the same.

Tried all sorts with this now and starting to run out of ideas.

Tried:
New windows install
New windows install, all M.2 removed (just installed on an old sata ssd)
Fresh drivers (several times)
Different bios's flashed to the 2080ti (official and unofficial)
Different tweaks in MSI afterburner

Yet to try:
Different PSU, different motherboard, different chip, if anyone thinks one of these might make a difference.


Now.... until now, I was CONVINCED it was the 2080ti having issues. Since tried a 3070 with fresh drivers/everything/etc and it underperforms by a similar margin (about 20%) so... this SEEMS to be something else.

Have checked through bios for any PCIe options, tried running in PCIe 3 as well with various options, as suggested, removed all the M.2 just to rule out PCIe lanes being used and forcing the 16x to 8x or something similar. I'm... out of ideas at this point though.

Have a PSU coming to test that, mines a seasonic x-series 1250W so... unless it's "gone weird" it's not that either.

Anyone got any ideas?

It FEELS like the PCIe slot is locked to 8x for some reason and is restricting bandwidth. The 2080ti's a seahawk EK-x with a good clock on it so... was outmuscling a very untuned 3070, which was expected. In each case that 20% margin seemed to be there.

My system runs PERFECTLY otherwise, no hang ups/issues reported anywhere.

Can anyone suggest some test tools/directions I can go with checking it all out? Alternative PSU will be tried soon.

For an idea what I'm talking about:

https://www.3dmark.com/compare/spy/...QmWqeXGDihsjOVIO24f-GeekHnfzWcn_mTGP3FOyS3B0#

CPU-z shows the PCIe 16x 3.0 @ 16 1.1. when stressed comes up to proper levels.
 
Last edited:
Soldato
OP
Joined
11 Jun 2003
Posts
5,010
Location
Sheffield, UK
Added the timespy compare with my mate to cover a good bit of that (link in first post). Boosting LOOKS fine, it's a seahawk on a big custom loop too so... about 55c on a bad day, getting hammered. The 3070 I tried was boosting ok too (on air though ofc).

So... doesn't paste nice but, my values first, his second. Better viewed in the link above, open the "graphics card" section.

My cards running about 90mhz faster on core and about 175 on memory. Even the averages have a good bit going my setups way. The GPU score is about getting for 50% behind.

3DMark Score
10 275
13 733 + 33.7 %
Graphics Score
10 361
14 591 + 40.8 %


GRAPHICS CARD
Graphics Card
NVIDIA GeForce RTX 2080 Ti
NVIDIA GeForce RTX 2080 Ti
# of cards
1
1
SLI / CrossFire
Off
Off
Memory
11,264 MB
11,264 MB
Clock frequency
2,025 MHz (1,350 MHz)
1,935 MHz (1,350 MHz)
Average clock frequency
1,992 MHz
1,840 MHz
Memory clock frequency
1,925 MHz (1,750 MHz)
1,750 MHz (1,750 MHz)
Average memory clock frequency
1,925 MHz
1,750 MHz
Average temperature
52 °C
82 °C
Driver version
27.21.14.5655
27.21.14.5655
Driver status
Approved
Approved
 
Soldato
Joined
23 Mar 2011
Posts
13,886
Location
West Side
Added the timespy compare with my mate to cover a good bit of that. Boosting LOOKS fine it's a seahawk on a custom loop too so... about 55c on a bad day, getting hammered. The 3070 I tried was boosting ok too.
What about your cpu is it boosting to higher frequency ?

Mite be worth trying a diffrent bios
 
Soldato
OP
Joined
11 Jun 2003
Posts
5,010
Location
Sheffield, UK
This was.... back when I first noticed it. It's on f31o bios (latest with the tweak in for the "all red" boost), latest nvidia drivers, latest AMD chipset drivers/etc. Still underperforms hugely.

I use the above example as we'd both updated everything that particular day to be comparing apples to apples as best possible.

My mates was largely stock, mine was....vaguely tuned and with v similar hardware, loses badly.

I'm just after ideas for what I can check it with. Some more hardware swaps will be possible in the near future but if there's any software tools I'm not aware of for checking bits out.
 
Last edited:
Top Bottom