Binyang Li 9aab9cacc0 support rocm7.2 (#819) ...

This pull request introduces support for ROCm 7.2 across the build
system, CI pipelines, Docker images, and documentation, while also
improving ROCm FP8 type selection and CUDA IPC memory handle management.
It updates dependencies and configurations to ensure compatibility with
ROCm 7.2, adds new options for native FP8 variants, and refines some
benchmarking and internal memory handling logic.

Pls notice: there is an issue in rocm7.2 (rocm7.2 user lib + rocm6.2
driver) when execution code in this order: allocating memory -> ipc
communication -> allocate new memory -> free old memory.

---------

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

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

..

__init__.py

Add collective benchmark and correctness check (#814)

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

allreduce_bench.py

Torch integration (#692)

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

allreduce.cu

Fix #458 (#568)

2025-07-13 18:02:56 -07:00

bench_collective.py

support rocm7.2 (#819)

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

comm.py

Add collective benchmark and correctness check (#814)

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

correctness.py

support rocm7.2 (#819)

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

gpu.py

support rocm7.2 (#819)

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

mscclpp_op.py

Torch integration (#692)

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

nccl_op.py

Torch integration (#692)

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

tuner.py

support rocm7.2 (#819)

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

tuning_config.py

Add collective benchmark and correctness check (#814)

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