James Heslop

Being part of the game and engine dev community, there’s an easy trap we all fall into early on: Putting together a decent renderer, calling it a "game engine", and claiming it runs faster than this or that industry standard tech. Sure, those mainstream engines are so bloated that hardly anyone enjoys using them anymore. We’ve all been there, and I believe that's why we do what we do. But the reality is that we can’t evaluate an engine's true performance without an actual game running on top of it (or at least part of it). Early isolated test scenes (that damn Sponza...) do a decent job simulating GPU load, but they don't tell the whole story when it comes to real world game production where a bunch of systems come together. Assuming that your renderer is indeed a capable one that meets certain standards, a complete game has to handle far more than that... Physics, audio, UI, localization, asset streaming, editor tools, complex gameplay logic, you name it... And all that compete for CPU and cache bandwidth. Getting 500 FPS in an isolated test scene is great as an academic exercise, but it's the classic case of lab vs real-world data failure. So forget about test scenes and before you declare war on the giants, at the very least prove your tech can drive a small but real, modern game concept without falling apart. To avoid any misunderstanding (given that I'm also building my own engine) I say this out of a genuine hope to see more decentralized, transformative gamedev tech again. Especially now, when the playing field appears to be somewhat equalizing for independent game devs.