Multinode NVL/NVLS supports#798

Open

Binyang2014 wants to merge 63 commits into

mainfrom

Commits on Apr 27, 2026

Scale native allreduce/allgather algos for MNNVL/MNNVLS

Binyang2014
and
Copilot
committed

Commits on May 2, 2026

Commits on May 4, 2026

Merge remote-tracking branch 'origin/main' into binyli/mnnvl
Binyang2014
committed

Commits on May 5, 2026

Make NVLS non-zero-copy allreduce algorithms MNNVL-ready

Binyang2014
and
Copilot
committed

Commits on May 6, 2026

Simplify torch-integration tuning example for MPI-only multi-node testing

Binyang2014
and
Copilot
committed
Drop non-MNNVL multi_node regime from torch-integration example

Binyang2014
and
Copilot
committed
Simplify torch-integration tuning example

Binyang2014
and
Copilot
committed
Shorten verbose comments and use THROW in validateIpcDomainSpansWorld

Binyang2014
and
Copilot
committed
Fold validateIpcDomainSpansWorld into getIpcDomainNranks

Binyang2014
and
Copilot
committed
Revert verbose RSAG zero-copy comment; rename NRanksPerNode template param

Binyang2014
and
Copilot
committed
Revert RSAG nBlocks default to 64

Binyang2014
and
Copilot
committed
Tie AllreduceAllpairPacket maxBlockNum_ to MAX_IPC_DOMAIN_NRANKS - 1

Binyang2014
and
Copilot
committed
Strip preflight validation blocks from NVLS pipeline allreduce kernels

Binyang2014
and
Copilot
committed
Default torch example SYMMETRIC_MEMORY env to 1

Binyang2014
and
Copilot
committed

Commits on May 7, 2026

NVLS zero-copy allreduce: support FP16 accumulator for FP8 inputs

Binyang2014
and
Copilot
committed

Commits on May 8, 2026

Commits on May 9, 2026

address hang issue
Binyang2014
committed

Commits on May 13, 2026

Commits on May 15, 2026

Commits on May 16, 2026

Commits on May 18, 2026

Move barrier into setupNvlsChannels and clean up NVLS pipeline state

Binyang2014
and
Copilot
committed

Commits on May 20, 2026

Fix collective topology sizing

Binyang2014
and
Copilot
committed
update
Binyang2014
committed

Commits on May 22, 2026

merge main
Binyang2014
committed

Commits on May 24, 2026

Fix memory leak
Binyang2014
committed

Commits on May 26, 2026

Commits on Jun 4, 2026

Commits on Jun 5, 2026

Commits on Jun 6, 2026

Commits on Jun 8, 2026

add more logs
Binyang2014
committed

Commits on Jun 12, 2026

for buffer pool
Binyang2014
committed

Commits on Jun 13, 2026

update for buffer pool
Binyang2014
committed

Commits on Jun 15, 2026

add test
Binyang2014
committed