Nvidia DLSS 4.5 boosts RTX 50 frame rates

The test results around CONTROL Resonant, a path-traced title due Sept. 24, show why Nvidia keeps leaning on DLSS. With Super Resolution, Ray Reconstruction and Multi Frame Generation enabled, the 6x setting lifted average frame rates about 4.1 to 4.9 times versus frame generation off across three resolutions. The GeForce RTX 5090 reached 313.9 fps at 4K and 616.0 fps at 1080p, while the RTX 5070 Ti posted 172.2 fps at 4K. Even the RTX 5060 Ti 16GB and RTX 5050 could show triple-digit or near-triple-digit averages in some cases, numbers that on paper make 4K play look accessible far below flagship pricing.

But the article’s more important message is that frame rate alone is no longer enough. Internal PC latency told a much harsher story. The RTX 5090 held 4K latency at 42.8 ms, comfortably usable for many players. The RTX 5070 Ti rose to 76.1 ms at 4K, suggesting WQHD is the better balance point. The RTX 5060 Ti climbed to 129.8 ms, and the RTX 5050 hit 476.0 ms at 4K, with the detailed log showing a range of 402.3 ms to 565.3 ms. In other words, AI-generated frames can make motion look smooth, but they cannot fully erase the underlying render burden or the responsiveness penalty that comes with too little native GPU headroom.

For investors, that is the key commercial takeaway. Nvidia’s gaming franchise benefits most when consumers believe the top-tier cards unlock a visibly better experience, not just higher benchmark numbers. The data here supports that message. RTX 5090 buyers can plausibly target 4K path-traced gaming, RTX 5070 Ti owners are pushed toward WQHD, and lower-end RTX 50 models are steered back to 1080p or lower settings. That segmentation is exactly what supports pricing power across the stack.

It also points to the broader economics of the GPU market: AI features are becoming as important as transistor counts in defining product tiers. If DLSS 4.5 can make demanding titles look and run better on Nvidia hardware than on competing platforms, the company strengthens its moat in a segment that remains one of the few consumer businesses where software differentiation can lift hardware margins. Still, the latency data is a reminder that AI upscaling is an enhancer, not a substitute for brute-force silicon, which leaves the flagship and upper-midrange cards the most defensible products.

The near-term watch point is whether more AAA releases adopt path tracing and Nvidia’s latest DLSS stack as a baseline feature set. If they do, the RTX 50 family could keep extracting premium demand from enthusiast buyers. If latency complaints become more visible at mid-tier price points, though, the upgrade narrative narrows quickly to the very top of the stack.