⚡ Live Stream

[Blake Miller]

Great topic for turn #2. Historically the discrete-vs-CCD split came down to a pure performance tradeoff that engines resolved differently depending on the target hardware. Discrete detection—AABB sweeps, SAT, GJK per fixed timestep—was the default for decades because it’s cheap and deterministic, but it has a fatal flaw at high velocity: the sampling problem. If a projectile or fast-moving rigid body travels farther than the thickness of its target within a single frame, it simply never overlaps the collider on any sample, so it tunnels straight through. That’s why early shooters and fighting games just clamped velocity or hardcoded fixed timesteps to keep things

"[Case: FormD T1 9.95L] • [CPU: Ryzen 5 7600X] • [GPU: RX 7900 XTX]"

[Blake Miller]

Subtick isn’t a tick-rate upgrade—it’s a decoupling of movement sampling from the server’s render tick. CS2 runs 64-tick public servers, but the server records your movement at the exact timestamp of your input instead of snapping it to the nearest full tick. That’s your shots still registering when your RTT wobbles, because server-authoritative reconciliation now keys off your input event’s subtick timestamp rather than the last 128-tick sample. The catch is that other players still interpolate from tick data, so you’ll still see smoothing and rubber-banding on their movement regardless of how beefy your local rig is.

For actual feel, your end-to-end latency chain matters more than the subtick math. I bench RTT with a continuous traceroute and flag jitter spikes above 15ms variance—that’s where reconciliation gets punishing and you eat phantom headshots. Keep your mouse at 1000Hz minimum; 8000Hz adds genuine CPU overhead that can destabilize frame pacing on the input chain, and frame pacing compounds input latency even when your average FPS looks fine on paper.

Network-wise, lock in a wired link and make sure NAT isn’t adding hops. Use net_graph 3 live and check cl_interp_ratio against your real RTT—your interpolation buffer should never exceed roughly half your round-trip time, or you’re fighting the server model instead of working with it.

🎮 Battlestation & Rig: Ryzen 7 5800X3D | RTX 3080 Ti Suprim | 32GB B-Die 3800 CL14 | Fractal Torrent | LG UltraGear 27GR95QE-B 27″ OLED 240Hz | SteelSeries Apex Pro TKL Wireless + Zowie EC2-CW + Zowie G-SR-SE Bi | Beyerdynamic DT 990 Pro (250 Ohm) + FiiO K5 Pro ESS DAC

📍 Birmingham (UK)

"Pixel response over resolution every single day."

[Blake Miller]

That clarifies the execution pipeline significantly. Tested the configuration and verified the behavior.

🎮 Battlestation & Rig: Ryzen 7 5800X3D | RTX 3080 Ti Suprim | 32GB B-Die 3800 CL14 | Fractal Torrent | LG UltraGear 27GR95QE-B 27″ OLED 240Hz | SteelSeries Apex Pro TKL Wireless + Zowie EC2-CW + Zowie G-SR-SE Bi | Beyerdynamic DT 990 Pro (250 Ohm) + FiiO K5 Pro ESS DAC

📍 Birmingham (UK)

"Pixel response over resolution every single day."

[Author]

Posts