Here's a recording of my Tiny Glade presentation from the Graphics Programming Conference 2024: www.youtube.com/watch?v=jusW... It's an any% speedrun of our GPU-driven rendering, shadows, global illumination, water, and DoF, involving a few weird tricks.

Rendering Tiny Glades With Entirely Too Much Ray Marching

YouTube video by Graphics Programming Conference

youtube.com

In #GPUZen 3, William Bussière and I explain how we render so many instances with Anvil in "GPU-Driven Rendering in Assassin's Creed Mirage" www.amazon.com/dp/B0DNXNM14K

We're very close to the release date of GPU Zen 3! a.co/d/gRPy2Ff

Nice summary of occupancy as a way to hide mem latency, but it is not the only (and sometimes not the best) way. Helping the compiler to add more instructions between the mem read issue and the data use by for eg partially unrolling a loop might help as well (from gpu-primitives-course.github.io)