@DavidGraham supplied a benchmark on one of the prior two pages.
I don't understand the distinction. 'Better' is a relative metric - choose your baseline and compare.
That was before AMD pulled the driver from their download site without explanation. No point in getting worked until official driver are released -- performance deltas could either go up or down.
It's a proportionate response and it's precisely that he's going to engage in confirmation bias with a specific adversarial pattern as a rationale for it. The discussion was already headed in the direction of beyond reasonable discourse ...
@Bold David already presented that data for everyone. In the worst case scenario both referenced hardware are dead even at the in-game benchmark according to gameGPU or DSO.
It really isn't easy to conclude especially if an opponent like David is going to insist on including dubious data from a tester where both the 7900 XT/XTX are literally equal in performance at 1080p or 1440p which is virtually out of line with everyone else's results. If anyone is going to resort to using plausible deniability as a basis to continue their line of reasoning then they shouldn't get an easy pass for it since it becomes a tedious exercise for those who did honourably exchanged their thoughts with them ...
Is Nvidia underperforming at lower resolutions or is AMD underperforming at higher resolutions?
No one is "underperforming" anywhere. RDNA2 GPUs tend to show more performance loss in comparison to Ampere GPUs when going into higher resolutions because they have narrower buses and slower DRAM, and their cache doesn't help as well as it does in lower resolutions. This is pretty much remedied with RDNA3 vs Lovelace which tend to have similar buses now and both are using big caches to cover the bandwidth deficit.
It could also be looked at as recent Nvidia architectures gaining less from lowering resolution. Infinity cache is less useful at higher resolutions as AMD has stated, but do these huge compute arrays on big Nvidia GPUs have utilization issues at lower resolutions?
Lower resolution relative performance issues are linked to Ampere mostly and it is hard to say what the reason for them is but my guess would be the CPU limitations which Nv h/w used to hit more often than AMD's. Lovelace doesn't show the same behavior to the same degree at least.
Well, according to ComputerBase, RDNA3 lose their edge fast in AC Mirgae, the 7900XTX starts out 15% faster than 4080 at 1080p, then collapses to a 7% lead at 4K. Same thing happens with RDNA2, the 6800XT vs the 3080, 18% lead at 1080p collapses to a 5% lead at 4K. This is using the latest driver from both vendors.
www.computerbase.de
Yeah, I wouldn't look at Mirage (or Valhalla previously) as it's very obviously an outlier among other titles.
I am interested in knowing why would you say that, other UE5 games don't behave like the Remnants 2 or Immortals of Aveum.
Doubtful that this has anything to do with h/w RT being supported or not. A game can still be better optimized for some specific h/w even if it's using a third party engine.
