Ethanfel 83e4b5dd98 perf: add torch.compile to PyTorch fallback kernels ...

Wraps _pytorch_softsplat and _pytorch_costvol with torch.compile
for ~6x speedup on ROCm/non-cupy setups. Falls back to eager
execution gracefully if compilation fails.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-04-11 10:13:30 +02:00

..

flowformer

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00

modules

perf: add torch.compile to PyTorch fallback kernels

2026-04-11 10:13:30 +02:00

raft

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00

__init__.py

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00

configs.py

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00

gimm.py

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00

gimmvfi_f.py

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00

gimmvfi_r.py

Add GIMM-VFI support (NeurIPS 2024) with single-pass arbitrary-timestep interpolation

2026-02-13 13:11:45 +01:00