While I am not interested in FS2020, here is an incredibly detailed review of performance benchmarks of 37, yes 37 graphics cards in FS2020.

https://www.guru3d.com/articles-pages/ms-flight-simulator-(2020)-the-2021-pc-graphics-performance-benchmark-review,1.html

 The 37 graphics cards used in this test are: 

• GeForce GTX 1060
• GeForce GTX 1070
• GeForce GTX 1070 Ti
• GeForce GTX 1080
• GeForce GTX 1080 Ti
• GeForce GTX 1660
• GeForce GTX 1660 Super
• GeForce GTX 1660 Ti
• GeForce RTX 2060
• GeForce RTX 2060 Super
• Geforce RTX 2070
• GeForce RTX 2070 Super
• GeForce RTX 2080
• Geforce RTX 2080 Super
• GeForce RTX 2080 Ti
• GeForce RTX 3060
• GeForce RTX 3060 Ti
• GeForce RTX 3070
• GeForce RTX 3080
• GeForce RTX 3090
• GeForce Titan Xp
• Radeon RX 470
• Radeon RX 480
• Radeon RX 5500 XT
• Radeon RX 5600 XT
• Radeon RX 570
• Radeon RX 5700
• Radeon RX 5700 XT
• Radeon RX 580
• Radeon RX 590
• Radeon RX 6800
• Radeon RX 6800 XT
• Radeon RX 6900 XT
• Radeon RX Vega 56
• Radeon RX Vega 64
• Radeon VII
• Radeon R9 Fury

Edited March 6, 20215 yr by pgde

^

Basically, the GPU is the controlling factor in MSFS performance🥱. For 4K, a 2080SC is the bare minimum.

The results of the 6900 XT are simply false. I extensively tested the 6900 XT against the 3080 and it performs on par or even a bit better than the 3080. You can see my test results here on the forum. 

19 minutes ago, carlito777 said:

The results of the 6900 XT are simply false. I extensively tested the 6900 XT against the 3080 and it performs on par or even a bit better than the 3080. You can see my test results here on the forum. 

Did they test with the same GPU and ram amount as you?

I can say that's pretty accurate for my card. With my 2070S I get exactly the fps mentioned both on my 4K monitor as well as on my 1440 one. For a smooth experience I must consequently scale down on 4K, but that's still looking good. 

Even though I am not in the market for new hardware this is (in my opinion) the most meaningful benchmark for GPUs since it is not simply a "raw" comparison like a 3D Mark but a more realistic measure of relative performance. Now that I have an accurate benchmark for my 2060 Super it is not unreasonable to state that the latest and greatest is now about twice as fast.

So when the higher end mainstream cards are about double the performance of my 2060 Super, for me that is usually the optimal time to upgrade. I could never pay out that sort of top end money to get maybe 18 months of top end performance when if I just waited that long I'd probably get the same or very similar performance for half the price.

Ever since I got my 3Dfx Voodoo 5500 at the princely sum of around $600 AUD back in 2000, I have limited myself to that sort of price range ever since.

Edited March 9, 20215 yr by JonP01

40 minutes ago, JonP01 said:

Even though I am not in the market for new hardware this is (in my opinion) the most meaningful benchmark for GPUs since it is not simply a "raw" comparison like a 3D Mark but a more realistic measure of relative performance. Now that I have an accurate benchmark for my 2060 Super it is not unreasonable to state that the latest and greatest is now about twice as fast....So when the higher end mainstream cards are about double the performance of my 2060 Super...

If your 2060S does 35fps, and 3090 does 50fps, that is a 42% increase over your 2060S which is certainly substantial but nowhere near twice as fast.  I have a 2070S, and a 3090 is 25% faster than it.  I bought the 2070S in March 2020 to tie me over til something better arrived, so much for that idea.  When I bought it using Passmark GPU benchmarks I think it showed the 2070S was only about 18% slower than 3090 in that benchmark, which was astounding to me, since the 2070S sold for $499.  I've been scheming on a 3090 but for two reasons I think it's ultimately a crazy purchase whose results will largely be underwhelming:

1.  Vents heat into the case. That just seems lame to me.  

2. There is no way you will use anywhere near all of that VRAM that helped make the card as expensive as it is, even when DX-12 arrives.  The top 10Gb will remain virgin in large part for most of its life cycle it seems. You can set render scaling up to 140 and get up to 20Gb I've  read, but I've looked at RS at 100 (my default) and 140, and quite frankly the difference if any is underwhelming at best.  Here, without looking at GPU load, which one do you think was rendered at 100 v 140, #1 or #2?  No peekie at the OSD first!

#1

#2

It proves the point if you are not using a 4K monitor the 3000 Series has no effect 3800 to me would be 1 frame more, and I am now thinking wait to see what the 4000 Series will bring at the moment the best around 9 frames for £2000 I don`t think so.

Edited March 9, 20215 yr by G-RFRY

On 3/9/2021 at 2:54 PM, Noel said:

If your 2060S does 35fps, and 3090 does 50fps, that is a 42% increase over your 2060S which is certainly substantial but nowhere near twice as fast. 

I based that on the toughest test because even if I might not wish to run an actual resolution that high, I crank the image quality way up to max including the very slow supersampling options. So I think in the real world with the sort of monitor I will upgrade to next and the image quality I will expect from it, better to work off the "worst case" scenario which is relative performance when pushed to the image quality limits.

My current FSX setup is such that I actually pretty much have both the CPU and GPU level pegged meaning one will never compromise the other and both are pushed to their limits. But obviously with FSX being old DX9 and acknowledged to be CPU bound, I am "pushing" the card with the image quality and AA options completely maxed out with nothing further to give. But based on my experiences with other modern DX11 / 12 games with the same card, that doubling of performance when pushed to the limit does seem pretty reasonable. Heck, at the asking price I would want double!!