423A35C7/pytorch3d
1
0
Fork 0
mirror of https://github.com/facebookresearch/pytorch3d.git synced 2025-12-22 23:30:35 +08:00
Code Issues Packages Projects Releases Wiki Activity
Files
9c586b1351df48a5e36461475bfb2d9ab961d749
pytorch3d/pytorch3d/csrc/pulsar/include
History
Suresh Babu Kolla e17ed5cd50 Hipify Pulsar for PyTorch3D 
Summary:
- Hipified Pytorch Pulsar
   - Created separate target for Pulsar tests and enabled RE testing
   - Pytorch3D full test suite requires additional work like fixing EGL
     dependencies on AMD

Reviewed By: danzimm

Differential Revision: D61339912

fbshipit-source-id: 0d10bc966e4de4a959f3834a386bad24e449dc1f
2024-10-09 14:38:42 -07:00
..
camera.device.h
Hipify Pulsar for PyTorch3D
2024-10-09 14:38:42 -07:00
camera.h
Hipify Pulsar for PyTorch3D
2024-10-09 14:38:42 -07:00
closest_sphere_tracker.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
commands.h
Hipify Pulsar for PyTorch3D
2024-10-09 14:38:42 -07:00
fastermath.h
License lint codebase
2021-06-22 03:45:27 -07:00
math.h
Hipify Pulsar for PyTorch3D
2024-10-09 14:38:42 -07:00
README.md
pulsar integration.
2020-11-03 13:06:35 -08:00
renderer.backward_dbg.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.backward_dbg.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.backward.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.backward.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.calc_gradients.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.calc_gradients.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.calc_signature.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.calc_signature.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.construct.device.h
Apply clang-format 18
2024-04-14 11:28:32 -07:00
renderer.construct.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.create_selector.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.create_selector.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.destruct.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.destruct.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.draw.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.fill_bg.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.fill_bg.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.forward.device.h
missing cudaGetLastError
2023-10-30 09:35:08 -07:00
renderer.forward.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.get_screen_area.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.norm_cam_gradients.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.norm_cam_gradients.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.norm_sphere_gradients.device.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.norm_sphere_gradients.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00
renderer.render.device.h
Hipify Pulsar for PyTorch3D
2024-10-09 14:38:42 -07:00
renderer.render.instantiate.h
Update license for company name
2022-01-04 11:43:38 -08:00

README.md

The include folder

This folder contains header files with implementations of several useful algorithms. These implementations are usually done in files called x.device.h and use macros that route every device specific command to the right implementation (see commands.h).

If you're using a device specific implementation, include x.device.h. This gives you the high-speed, device specific implementation that lets you work with all the details of the datastructure. All function calls are inlined. If you need to work with the high-level interface and be able to dynamically pick a device, only include x.h. The functions there are templated with a boolean DEV flag and are instantiated in device specific compilation units. You will not be able to use any other functions, but can use func<true>(params) to work on a CUDA device, or func<false>(params) to work on the host.