Why Babylon didn’t use shared GPUBuffer or BindGroup dynamic offset in WebGPU BindGroup to improve performance? Reducing the GPUBuffer reduces the number of times the CPU uploads data to the GPU, potentially improving performance.
Welcome aboard!
Keep in mind that WebGPU support was developed in parallel to the specification and started more than 4 (or 5) years ago! So a lot has changed since then, and the WebGPU engine also has to conform to the existing Babylon architecture. But there are plans to improve WebGPU support in the future, so stay tuned!
Thank you, it turned out to be like this. 