Nothing seems a huge surprise. The top cards (less powerful) seem to run 4K at way lower than 1/2 of 1080p. The bottom cards (better cards) there is less of penalty for moving to higher resolutions.

 

......................FPS............................
GPU..............1060p...1440p...4K
RX 590...........30...........21........12
1660ti.............41...........30........16
RX 5700XT...57............39.......21
2070 SUPER..57..........46......28
2800 SUPER..58........52........33
3060ti.............58.........52.......32
RX 6800.........59........59.... ..34
3070................53........56........37
RX 6800 XT...62.......56........41
3080...............66........66........51
RX 6900 XT...60......62........43
3090...............59.......56........52

 

No way the RX-6800 and RX-6900XT are that close together, I had both here and they are miles apart in FPS in MSFS.
The 6900xt is superior over the 6800, we're talking at times easily a 20% difference.

The test results are invalid, completely invalid as something strange going on with those numbers.

When I got the 6800 I was disappointed in the MSFS performance, so I returned it and got a 6900XT.
The 6900XT also smokes it in RDR 2 by even more, by about 25% to 30% at times. 

I underclocked my 6900XT and I was still beating the FPS in MSFS that an overclocked 6800 was giving me.

 

Edited July 25, 20223 yr by Alpine Scenery

You might get 60 fps sometimes, but that should not have been listed as the average.
The average FPS is lower, unless he was using SU 10 or something (I don't know  - didn't watch the video).

Edited July 25, 20223 yr by Alpine Scenery

RTX 3080 Ti vs RX 6950 XT - Test in 10 Games - YouTube

11 hours ago, Alpine Scenery said:

The test results are invalid, completely invalid as something strange going on with those numbers.

Like the 3080 being better than the 3090 at 1080/1440p and basically equal at 4K.

The  6950XT is faster than the 3800 TI in a properly optimized game, whichever one is faster in MSFS is version specific and varies on version to version of the game and the updates. Even though true, the 3080TI is the safer purchase for other reasons, but the 6900XT and 6950XT generally beat it in normalized testing.

That said, in some games the DLSS would give the Nvidia back the advantage, but I think in a few more months most of these differences will level out as drivers and settings mature, and there is more AMD specific implementations and support.



There are too many variables in FPS tests and people doing the tests that don't have proper experience normalizing results, and normalizing the results are somewhat involved to get all the variables leveled out. Even the slightest version difference or update can fry the results.

 

Edited July 25, 20223 yr by Alpine Scenery

In the end, the 3090, 3080 TI, and the AMD 6900xt / 6950xt are so close that it doesn't matter much for FPS differences, just get whatever deal you can find within reason of taking into account that AMD has less market share which causes more performance issues in games sometimes (but getting better slowly) due to slower fixes and optimizations.

However, it was a no brainer for me getting a refurb Red Devil 6900xt for $600, as when I bought the equivalent 3080 TI would have cost $300+ more, and the 3090's even more than that. It's a bit risky buying a refurb card, but it came from Amazon directly and so far the card runs just like new. Of course a refurb is always riskier without a warranty, but it was such a huge difference in price I figure if the card fails, it's most likely the prices will be lower and I will likely still break even worst case (or close enough).

I probably would have gone NVIDIA if the prices were within $100 to $150, but after the difference reaches $200+, I'd rather just spend the difference on some other upgrade to level the field.

 

Edited July 26, 20223 yr by Alpine Scenery

Side Note:
The 6900XT is superior to the 6800 for MSFS because it gives you just enough headroom where you don't have to overclock.

Heck, I rarely hit 75-80c on the junction (hot spot).
I am averaging 55c card temp and 65c junction hot spot, no water cooling (just multiple "quiet" Noctua fans running 1000 - 1400 RPM).

I see many people running their cards in the 90 c's, which could be borderline depending how hot your other components are getting.
80's should be ok but you have to watch other temps, averaging in the 70's is good, but 60's is bliss and nirvana.

 

Edited July 26, 20223 yr by Alpine Scenery

1 hour ago, Alpine Scenery said:

The  6950XT is faster than the 3800 TI in a properly optimized game, whichever one is faster in MSFS is version specific and varies on version to version of the game and the updates. Even though true, the 3080TI is the safer purchase for other reasons, but the 6900XT and 6950XT generally beat it in normalized testing.

That said, in some games the DLSS would give the Nvidia back the advantage, but I think in a few more months most of these differences will level out as drivers and settings mature, and there is more AMD specific implementations and support.



There are too many variables in FPS tests and people doing the tests that don't have proper experience normalizing results, and normalizing the results are somewhat involved to get all the variables leveled out. Even the slightest version difference or update can fry the results.

 

I remember a shocking increase in fps 5 minutes after I installed SU5 (compared to update 4). It was enormous. And soon it went away because everyone complained of delayed scenery popping when switching to a side window view, therefore Asobo rolled back their tweaks with a quick 'patch'. People suspected that SU5 fps increase was for the new Xbox players.

38 minutes ago, Fielder said:

I remember a shocking increase in fps 5 minutes after I installed SU5 (compared to update 4). It was enormous. And soon it went away because everyone complained of delayed scenery popping when switching to a side window view, therefore Asobo rolled back their tweaks with a quick 'patch'. People suspected that SU5 fps increase was for the new Xbox players.

Cool, I am fine with people complaining about performance (especially when video cards were so overpriced)...

My best advice to people is to buy something to test it on your own system, and if not happy with the increase, return it (like I did), and go up another level until you are happy.

Problem solved and you don't have to rely on FPS tests, you can see with your own eyes how it looks and feels...

 

This will run MSFS on six monitors 😀.

Nvidia Reportedly Readies 800W RTX 4090 Ti With 18,176 Cores, 48GB GDDR6X | Tom's Hardware (tomshardware.com)