Jeremy Reizenstein 305cf32f6b Avoid raysampler dict ...

Summary:
A significant speedup (e.g. >2% of a forward pass).

Move NDCMultinomialRaysampler parts of AbstractMaskRaySampler to members instead of living in a dict. The dict was hiding them from the nn.Module system so their _xy_grid members were remaining on the CPU. Therefore they were being copied to the GPU in every forward pass.

(We couldn't easily use a ModuleDict here because the enum keys are not strs.)

Reviewed By: shapovalov

Differential Revision: D39668589

fbshipit-source-id: 719b88e4a08fd7263a284e0ab38189e666bd7e3a

2022-09-21 04:29:44 -07:00

..

common

Remove pytorch3d's wrappers for eigh, solve, lstsq, qr

2022-08-31 13:04:07 -07:00

csrc

Make PyTorch3D C++17 incompatible again :(

2022-08-31 02:38:31 -07:00

datasets

lint fix: raise from None

2022-07-14 04:21:44 -07:00

implicitron

Avoid raysampler dict

2022-09-21 04:29:44 -07:00

io

upgrade pyre version in fbcode/vision - batch 2

2022-08-04 22:18:46 -07:00

loss

validate lengths in chamfer and farthest_points

2022-09-08 15:03:36 -07:00

ops

validate lengths in chamfer and farthest_points

2022-09-08 15:03:36 -07:00

renderer

raysampling stratified_sampling doc

2022-09-06 10:28:45 -07:00

structures

suppress errors in vision/fair/pytorch3d

2022-09-01 11:46:55 -07:00

transforms

Remove pytorch3d's wrappers for eigh, solve, lstsq, qr

2022-08-31 13:04:07 -07:00

utils

Update license for company name

2022-01-04 11:43:38 -08:00

vis

suppress errors in vision/fair/pytorch3d

2022-06-15 06:27:35 -07:00

__init__.py

version 0.7.0

2022-08-09 20:48:51 -07:00