It has 60% more cores running (in theory) at 40% higher clocks. If you aren’t getting 80% more base performance in conditions where you can actually full all the cores, they’ve majorly f’ed up the architecture.
Given its just a die shrink of Ampere, that seems unlikely.
Now, 1080p and even 4k, I can see just not having enough pixels to fill the pipeline, but at 12Mpixels+, I’ve got enough to fill that pipe.
That’s why I am currently GPU bound, not CPU bound. And why you really need to consider what your objective configuration is before choosing a card. What is over kill for 4K is not the solution of high resolution VR, which is also not even remotely the right solution for 1440p.
How about an example where 22 fps is being interpolated to 44 (ie barely playable to quite smooth)? Come on stress that thing with a useful example. Who cares if you can interpolate a scene that’s already 70 fps up to 140 fps (ie butter and still butter)?
I currently have an AMD Ryzen 7 3800XT and RTX 3070 (it was all I could get when graphics cards were so scarce). I average in the mid-20’s FPS, running at 4K Ultra in MSFS, and I would like to upgrade. But out of the two, it seems (anecdotally) that a CPU upgrade to the Ryzen 7 5800X3D would be a better option to increase my frame rates than updating the graphics card. Would you guys agree with that?
If I do upgrade the GPU, I’m thinking of the regular 3090 which is around £1000 at the moment…
no, just wait. don’t upgrade your system now…unless you want similar performance. Why not sit tight and see what the actual numbers will be with a 7800x3D and 4090. I’m 99% sure that this setup is going to be the holy grail for MSFS for a few years at least.
People seem to ignore the fact that the 4090 is supposed to reduce CPU overhead using DLSS 3.0. It’s not necessarily about raw power anymore, it’s about using this new hardware in a smarter way. With the increase in 7800x3D L3 Cache (which MSFS seems to love), and reducing the CPU overhead using the 4090, I can see 100+ FPS in nearly every conceivable situation including heavy addon aircraft and scenery while at 1440p and possibly 4k. Regardless, 60 FPS in every situation while utilizing this new tech will most definitely be standard.
My advice is just chill. Wait. Plus DX12 is still in beta, you may even get further with your current hardware in just a couple of months.
Who did this benchmark? Paul’s Hardware? It would be helpful to know what is being tested in MSFS for this benchmark. It is common knowledge that MSFS FPS can vary significantly during flight. Also many MSFS configurations are CPU limited. These benchmark numbers imply there aren’t any CPU limitations using ULTRA like “Main Thread”. It would be incorrect to say or assume that installing a 4090 is going to solve MSFS CPU performance problems.
yea, i mean. it’s your money. Personally, I wouldn’t upgrade unless I know it’ll make a marked difference. I’ve been chasing good performance in flight sims for years and rarely upgrade. But when I do, it better be worth it! I’d rather go all out one time, instead of constantly making marginal improvements.
That video is very much an outlier compared to every other review I’ve seen today. Suggests either broken testing methodology or a cynical way of getting hits.
Don’t be silly. Of course it will be an improvement. Never said otherwise. The question is how MUCH of an improvement and whether it’s worth the investment.
From my experience, there is of course also is an improvement in performance between a 3070 and a 3080 TI. However, compared to how OTHER games profit from that upgrade, the difference in MSFS is rather disappointing considering a 3090 will cost double of what a 3070 costs.
So if you now have a 3080TI, 3090 or 3090TI, I personally do not think it’s worth the upgrade considering a 4090 will probably set you back double what you can get for a 3090 used. But since there are no real prices and noone has actually tested the 4090 yet, it’s mostly speculation.
If you’re wavering, there will be a ton of real-world posts here in like 2 or 3 days unless the “lots of supply” rumors are untrue.
I’m coming from a 10GB RTX 3080, so even if I “only” see RTX 3090Ti-level performance in VR and in games that don’t support DLSS 3.0, it’ll still be a nice improvement.
With my current config, I can get 50+ FPS at 5120x1440 flying a non-taxing aircraft around places like the wilds of British Columbia, Florida, the Netherlands, etc. It’s the situations where I see FPS drop down to 20-22 (New York City, London, Seattle, oddly random places in VR, etc) where I’ll be curious as to how much my new rig is going to help. Even in those places, Dev mode claims I’m GPU-limited in DX12 (CPU in DX11).
Quick question:
Since these extra frames are not touched by the cpu, how does the sim know how fast it needs to simulate things? It the cpu can produce 60 updates per second, but the gpu produces 120 frames in the same time… won’t that create all kinds of issues?
Surly they must have thought about this scenario, but I am curious if anyone knows how they solve this.
(i.e. from AM4 to AM5)