Supposed 2080 Ti vs 3090 Video (8 games tested)

5150Joker

Supreme [H]ardness
Joined
Aug 1, 2005
Messages
4,568
Take a lot of salt w/this video:

https://www.bilibili.com/video/BV1uV411S7Ac?zw

The Chinese dudes at ChipHell calculated the following:
Calculate the percentage of the lead under 1080p: 51, 52, 51, 80, 49, 50, 83, 52 Leading under 2160p: 53, 45, 53, 74, 57, 52, 87, 52 6 games in About 50%, 2 are about 80%. The data is very discrete, which is probably caused by too few samples.
 
I watched and didn't seem impressive for some reason. I hope its fake or they meant to say 3080.
 
So nVidia claims the 3090 can do 8K at 60 FPS, and yet in this we see it barely able to maintain 60 FPS at 4K in current games.

Doesn't really seem legit, but who knows.
 
So nVidia claims the 3090 can do 8K at 60 FPS, and yet in this we see it barely able to maintain 60 FPS at 4K in current games.

Doesn't really seem legit, but who knows.

We know the relative performance of the 3080 from Digital Foundry’s early test of it. The 3090 has 20% more Cuda cores. There is not a chance the 3090 will be playing modern AAA games at non scaled 8k 60 max settings. With heavy implementation of DLSS and the settings turned down maybe and esports games certainly.
 
So nVidia claims the 3090 can do 8K at 60 FPS, and yet in this we see it barely able to maintain 60 FPS at 4K in current games.

Doesn't really seem legit, but who knows.

8K 60fps is only with DLSS, so only a handful of games (of course some older and less demanding games will work fine too). I mean it's 4 times more pixels than 4k...
 
I think it is a pretty good prediction of what we will see in real reviews when comparing apples to apples. It makes sense based on the specs. Expect a much higher jump with any ray tracing though. Even more if you are using DLSS.
 
I think it is a pretty good prediction of what we will see in real reviews when comparing apples to apples. It makes sense based on the specs. Expect a much higher jump with any ray tracing though. Even more if you are using DLSS.

I understand suspicion in the video's authenticity, but not sure why the naysayers on the results. Isn't 50-60% above 2080ti in performance about what's expected from 3090? Especially, if we think 2070 should match 2080ti, and 3080 is ~25% more beefy in specs than 3070, and 3090 is ~25% more beefy than 3080... Doesn't this add up to about 50% above 2080ti.
 
I understand suspicion in the video's authenticity, but not sure why they're are naysayers on the results. Isn't 50-60% above 2080ti in performance about what's expected from 3090? Especially, if we think 2070 should match 2080ti, and 3080 is ~25% more beefy in specs than 3070, and 3090 is ~25% more beefy than 3080... Doesn't this add up to about 50% above 2080ti.

The 3080 has 47% more shading power than the 3070, and much higher memory bandwidth. The 3090 is yet another 20% faster, so if we fly with the reasonable assumption that a 3070 performs the same as a 2080 Ti (mirroring the 970 and 1070 launches) then the 3090 better be a damn lot faster than a 2080Ti.
 
A 390 is about double the specs of the 3070, so in some situations the 3090 might be double the performance of the 2080ti. But for rasterized games I think a 40 to 70% improvement is a reasonable prediction.
 
GPU sellers in China have had release units of 3080 and 3090 for at least a week or more. Heck you can even buy one ahead of schedule if you got the dough. No reason why someone cannot leak a benchmark of it. The numbers look broadly in line with other leaks.
 
Back
Top