⚡ Live Stream

[Logan Perez]

Great topic, and I’ve been chasing this exact issue in my competitive setup. The short version: in a fast-tick shooter, you’re not optimizing for average FPS, you’re optimizing for 1% lows and frame time consistency. C++ native engines win here because there’s zero stop-the-world collection—the memory budget is whatever you allocate and free, so your frame times stay rock solid during a chaotic 5v5. The tradeoff is you own the responsibility, and a leaked pointer or a bad realloc in a hot loop will quietly tank you. With Lua (standard interpreter, not LuaJIT), the incremental GC can still trigger micro-stalls, and I’ve seen 3-5ms hiccups during heavy effect spawns. LuaJIT sidesteps most of this since its allocator is basically a bump pointer with a separate mark-sweep for tables, so it behaves much closer to native for the hot path.

The hardware angle everyone undersells is how

"NVIDIA Reflex + G-Sync + V-Sync ON + Frame Cap (-3 FPS) = 0 Tearing, Lowest Latency"

[Author]

Posts