Frame Generation Latency: NVIDIA DLSS 3 vs AMD FSR 3 vs Reflex Integration: Competitive Benchmark Analysis and Υλικό Interactions

[Chase Gray]

I’ve been bencharking DLSS 3, FSR 3, and Reflex in CS2 and Valorant, and the numbers are getting muddy. Everyone cites the input latency specs, but the real question is what you’re actually feeling at 240Hz. For those running frame gen on top of Reflex in competitive titles — did you notice a measurable input delay penalty versus native rendering, or did Reflex fully close the gap? And does the UI decoupling actually hold up during fast flicks, or do you see the crosshair/environment decoupled from the aim buffer creating a disconnect?

On frametime consistency, how clean are your 1% lows with optical flow interpolation mid-combat? I’ve seen stutter spikes when the flow estimator chases fast-moving particles. Which GPUs are people testing on, and did you see scaling differences between the 40-series and 7000-series when pushing higher refresh rates? Any of you running a high-frame-rate setup with a scope or latency meter pulling hard numbers? I want real-world data before I commit to a build, so share your configs and whatever you’re seeing at the input layer.

🎮 Battlestation & Rig: Ryzen 7 7800X3D | RTX 4080 Super TUF | 32GB DDR5 6000 CL30 | 2TB 990 Pro | BenQ ZOWIE XL2566K 24.5″ 360Hz DyAc+ | Fanatec Gran Turismo DD Pro (8Nm) + ClubSport V3 Load Cell Pedals | HyperX Cloud III Wireless + Dolby Atmos for Headphones

📍 Munich (Germany)

"Hand-lubed linear switches and poron foam are pure therapy."

[Owen Perez]

Great topic for a deep dive. The key thing most people gloss over is that frame generation and latency aren’t the same metric—here’s how they actually interact across these technologies. NVIDIA’s DLSS 3 is deeply integrated at the driver/API level with Reflex, which is a genuine advantage: Reflex actively measures the frame queue depth and injects an extra frame before the GPU-bound frame, effectively pulling the “input latency + frame gen” total back down closer to native levels.

🎮 Battlestation & Rig: i5-13600K | RTX 4070 Super | 32GB DDR5 6000 CL32 | DeepCool LT720 | Samsung Odyssey Neo G8 32″ 4K 240Hz Mini-LED | Custom Tofu60 Lubed Gateron Oil Kings + Lamzu Maya 4K + Artisan Hien Mid | Beyerdynamic DT 770 Pro (80 Ohm) + Focusrite Scarlett Solo Gen 4

📍 Birmingham (UK)

"Hand-lubed linear switches and poron foam are pure therapy."

[Zoe Mitchell]

Short version: for competitive play, neither should be your default, and if you’re benchmarking latency you need to be measuring the right thing. The headline stat everyone misses is that DLSS 3’s real advantage over FSR 3 isn’t image quality—it’s that the frame generation pipeline is hardwired to work with Reflex, so you get latency reduction stacked on top of the extra frames. The RTX 40 series has a dedicated Optical Flow Unit, and when a title supports both Reflex and DLSS 3 frame gen, system latency can actually drop below the native frame rate, which is why you see figures like a locked 165Hz feel at 330fps+ in the latency counters. FSR 3 frame gen is more portable (works on older cards, even some Intel), but historically it shipped with noticeably higher input latency until AMD added driver-level AFMF and Reflex support in

Main Battlestation: Ryzen 5 7600X / RX 7900 XTX / Dell Alienware AW2725DF 360Hz QD-OLED

[Chase Gray]

Update: I finally got it sorted out. I was stuck at constant 15-20fps drops while running DLSS 3 frame gen in competitive shooters, and the lag felt unplayable. Followed the advice from the thread—disabled the frame pacing override in the NVIDIA control panel, turned Reflex on “Boost” mode, and bumped my BIOS p-state setting so the CPU doesn’t bottleneck the injected frames. Framedrops completely stopped and my 1% lows are actually in the 160s now. Thanks guys, this thread was gold.

"gl & hf in your games."

[Sophie Parker]

Great topic. The key thing people undersell is that DLSS 3 and FSR 3 solve two different problems, and the latency story is where they really diverge. NVIDIA’s advantage isn’t just the optical flow accelerator (OFA) hardware baked into the RTX 40 series—that does give them cleaner, more accurate motion vectors and fewer artifacts on fast pans—but their Reflex integration is genuinely more mature. Reflex actively measures and reduces the render pipeline latency, and it’s specifically tuned to offset the extra latency that frame generation adds. AMD’s Anti-Latency (arriving with FSR 3.1) does something similar, but it’s been around

"Sony BVM 20″ Broadcast CRT Monitor + Sega Dreamcast + Super Famicom"

[Author]

Posts