MrNeRF / gaussian-splatting-cuda

3D Gaussian Splatting, reimagined: Unleashing unmatched speed with C++ and CUDA from the ground up!
Other
909 stars 74 forks source link

Memory duplication speed up #3

Closed johnryan465 closed 1 year ago

johnryan465 commented 1 year ago

By making the shared memory D times larger, we can reduce the amount of collisions when performing atomic adds reducing the overall time. Current benchmark time: 151 seconds on 3090