Isotr0py be8a72f0a6 Stack torch.compile with CUDA graph decode instead of choosing one ...

Regionally compile the DiT blocks even when enable_cuda_graph_decode is set
and let graph capture record the compiled (fused) kernels; inductor's own
cudagraphs stay off so the two graph layers never nest. Measured on GB200
(1024x1024, 50 steps, seed 42, BF16): combo 3.04s vs compile-only 3.2s,
graph-only 4.0s, eager 4.3s per image; 44.5-45.6 dB PSNR vs eager.

2026-09-22 08:58:32 +00:00

..

__init__.py

[Feat] Support step-boundary abort in diffusion (#1769)

2026-04-01 17:25:02 +08:00

diffusion_model_runner.py

Stack torch.compile with CUDA graph decode instead of choosing one

2026-09-22 08:58:32 +00:00

diffusion_worker.py

[Core][Diffusion] Add pause_generation(mode="keep") for diffusion stages (#7685)

2026-09-22 01:31:22 +08:00

input_batch.py

[LingBot World] Eleven bit-identical removals of repeated work in the realtime path (#7648)

2026-09-21 06:12:45 +00:00

request_batch.py

[Core][Diffusion] Add a typed pre-D2H video media contract (#6615)

2026-09-10 09:50:54 +00:00

utils.py

[Feat][OmniVoice]Support Varlen Attn, Request-Batch and Step-Execution (#6408)

2026-09-08 18:58:38 -07:00