Skip to content

[Qwen3.5_dev] Cherry-pick fix for custom all-reduce #4547

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
yixionghuo merged 2 commits into from
Aug 6, 2026
Merged

[Qwen3.5_dev] Cherry-pick fix for custom all-reduce #4547

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
9b9d94c
Fix: add missing end_sync in cross_device_reduce_1stage (#3514)
zovonoir Jun 4, 2026
37a95e9
fix: synchronize custom collectives before return (#4082)
jpy794 Jul 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions csrc/include/custom_all_reduce.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -467,6 +467,7 @@ __global__ void __launch_bounds__(512, 1) cross_device_reduce_1stage(RankData* _

buf = next_buf;
}
end_sync<ngpus, true>(sg, self_sg, rank);
}

template <typename T, int ngpus, bool is_broadcast_reg_outptr = false>
Expand Down Expand Up @@ -698,6 +699,7 @@ __global__ void __launch_bounds__(512, 1) allgather_naive(
int write_idx = warp_id * size + idx;
result[write_idx] = ptrs[warp_id][idx];
}
end_sync<ngpus, true>(sg, self_sg, rank);
}

template <typename T, int ngpus>
Expand Down Expand Up @@ -725,6 +727,7 @@ __global__ void __launch_bounds__(512, 1) allgather_vec(
int write_idx = warp_id * size + idx;
*(reinterpret_cast<P*>(&result[0]) + write_idx) = ptrs[warp_id][idx];
}
end_sync<ngpus, true>(sg, self_sg, rank);
}

template <typename T, int ngpus>
Expand Down Expand Up @@ -761,6 +764,7 @@ __global__ void __launch_bounds__(512, 1) allgather_lastdim(RankData* _dp,
int write_idx = (ngpus * y + warp_id) * last_dim_size + x;
*(reinterpret_cast<P*>(&result[0]) + write_idx) = ptrs[warp_id][idx];
}
end_sync<ngpus, true>(sg, self_sg, rank);
}

/*
Expand Down Expand Up @@ -796,6 +800,7 @@ __global__ void __launch_bounds__(512, 1) reduce_scatter_first_dim(
*(reinterpret_cast<P*>(result) + store_index) =
packed_reduce<P, ngpus, A>(ptrs, load_index);
}
end_sync<ngpus, true>(sg, self_sg, rank);
}

// fp8 quant all-reduce code start
Expand Down