Binyang Li 325f79f9dc Add configurable FP8 low-latency dispatch ...

Quantize BF16 dispatch payloads to FP8 E4M3 with format-defined block scales while preserving BF16 expert outputs for combine. Clean up the sender structure, payload metadata, vector conversions, Python API, and multi-rank coverage.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Copilot-Session: efbacae6-f679-430b-bc16-b45ae162fc76

2026-07-13 18:47:21 +00:00

..

csrc

Binyli/ep revise (#828)

2026-07-06 21:14:29 -07:00

mscclpp

Add configurable FP8 low-latency dispatch

2026-07-13 18:47:21 +00:00

mscclpp_benchmark

Torch integration (#692)

2026-01-21 20:32:24 -08:00

test

MoE Commnucator design doc (#818)

2026-06-29 13:32:51 -07:00

CMakeLists.txt

Disabling Nanobind Memory Leak Warnings in Release Builds (#745)

2026-02-23 11:55:17 -08:00

requirements_cuda11.txt

Support python wheel build (#787)

2026-04-16 21:24:45 -07:00

requirements_cuda12.txt

Support python wheel build (#787)

2026-04-16 21:24:45 -07:00

requirements_cuda13.txt

Support python wheel build (#787)

2026-04-16 21:24:45 -07:00

requirements_rocm6.txt

Support python wheel build (#787)

2026-04-16 21:24:45 -07:00