Binyang Li 5ebbd340bb Enable efficient 16-rank HT EP ...

Extend direct-fabric HT launch and runtime support to 16 ranks. Decouple TMA contributor staging from global rank discovery and restore adaptive warp geometry.

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

2026-07-15 06:35:34 +00:00

..

csrc

merge main

2026-07-13 22:01:36 +00:00

mscclpp

Enable efficient 16-rank HT EP

2026-07-15 06:35:34 +00:00

mscclpp_benchmark

support rocm7.2 (#819)

2026-06-24 16:09:34 -07:00

test

merge main

2026-07-13 22:01:36 +00:00

CMakeLists.txt

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

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

requirements_cuda12.txt

Add collective benchmark and correctness check (#814)

2026-06-04 09:22:10 -07:00

requirements_cuda13.txt

Add collective benchmark and correctness check (#814)

2026-06-04 09:22:10 -07:00

requirements_rocm6.txt

Add collective benchmark and correctness check (#814)

2026-06-04 09:22:10 -07:00

requirements_rocm7.txt

support rocm7.2 (#819)

2026-06-24 16:09:34 -07:00