Samuel NordmannandMisha Goin 4a99ca40d6 [Perf][MoE] Eliminate staging from NCCL symmetric reduce-scatter (#49194) ...

Signed-off-by: samnordmann <snordmann@nvidia.com>
Co-authored-by: Misha Goin <mgoin64@gmail.com>

2026-10-06 07:37:01 -07:00

..

__init__.py

[Core][Refactor] move parallel_utils into vllm/distributed (#3950)

2024-04-10 15:33:30 -07:00

aiter_custom_all_reduce.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

all2all.py

[Perf][MoE] Eliminate staging from NCCL symmetric reduce-scatter (#49194)

2026-10-06 07:37:01 -07:00

all_reduce_utils.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

base_device_communicator.py

[Perf][MoE] Eliminate staging from NCCL symmetric reduce-scatter (#49194)

2026-10-06 07:37:01 -07:00

cpu_communicator.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

cuda_communicator.py

[Perf][MoE] Eliminate staging from NCCL symmetric reduce-scatter (#49194)

2026-10-06 07:37:01 -07:00

cuda_wrapper.py

[Bugfix][KV Offload] Register the offload region in chunks (#51081)

2026-09-17 13:19:56 +03:00

custom_all_reduce.py

[Feature] Release the CUDA graph pool on sleep (#59160)

2026-10-04 01:13:10 +00:00

flashinfer_all_reduce.py

[Bugfix] Fix FlashInfer all_reduce backend selection (#56891)

2026-10-03 22:18:58 -04:00

flashinfer_pcie_ipc_all_reduce.py

[Distributed] Add opt-in FlashInfer PCIe IPC all-reduce backend (#53576)

2026-09-01 03:34:56 +00:00

mnnvl_compat.py

[Kernel] Add FlashInfer MoE A2A Kernel (#36022)

2026-03-15 23:45:32 -07:00

pynccl_allocator.py

[Core] Model console logging as CLI configuration (#57205)

2026-09-25 11:36:59 +01:00

pynccl_wrapper.py

[Distributed] Support pre-shared ncclUniqueId rendezvous for weight transfer (#53784)

2026-09-01 14:55:57 +08:00

pynccl.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

quick_all_reduce.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

ray_communicator.py

Add pydocstyle to the ruff rules (#52136)

2026-09-16 16:44:29 +00:00

shm_broadcast.py

[CI] Ignore ruff D209, rejoin the docstrings it split, and silence incompatible-rule warnings (#57212)

2026-09-16 10:57:24 -07:00

shm_object_storage.py

[Bugfix][Multimodal] Fix reference counting of the SHM processor cache (#59695)

2026-10-02 22:14:13 +08:00

symm_mem.py

[Rubin] Enable NVLink all-reduce paths on SM107 (#49647)

2026-07-29 20:53:25 +00:00

xpu_communicator.py

[Feat][XPU] VLLM_BATCH_INVARIANT support for Dense/MoE models (#55881)

2026-09-22 11:06:54 -04:00