Skip to content

Fix group name lookup crash for flattened mesh collectives in debug_helpers #364

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
fmassa merged 1 commit into from
Mar 16, 2026
Merged

Fix group name lookup crash for flattened mesh collectives in debug_helpers #364

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
19 changes: 14 additions & 5 deletions autoparallel/graph_passes/debug_helpers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -125,11 +125,20 @@ def custom_runtime_estimation(node: torch.fx.Node, override_size=None):
target = node.target
if target == torch.ops._c10d_functional.wait_tensor.default:
return 0
# TODO: figure out mesh without reading from global scope
mesh_topo = MeshTopoInfo.build_from_mesh(mesh)
groups_name = tuple(g.group_name for g in mesh.get_all_groups())
# Resolve group_name to a mesh and dimension. Collectives may
# use per-dimension groups or the flattened mesh group created
# by mesh._flatten() in _apply_placement_common.
group_name = get_group_name(node)
mesh_dim = groups_name.index(group_name)
groups_name = tuple(g.group_name for g in mesh.get_all_groups())
if group_name in groups_name:
cost_mesh = mesh
mesh_dim = groups_name.index(group_name)
elif mesh.ndim > 1:
cost_mesh = mesh._flatten()
mesh_dim = 0
else:
return 0
mesh_topo = MeshTopoInfo.build_from_mesh(cost_mesh)
t = node.args[0].meta["val"] # type: ignore[union-attr]
comm_bytes_gb = t.numel() * t.itemsize / 2**30
if override_size is not None:
Expand All @@ -138,7 +147,7 @@ def custom_runtime_estimation(node: torch.fx.Node, override_size=None):
torch.ops._c10d_functional.all_gather_into_tensor.default,
torch.ops._c10d_functional.all_gather_into_tensor_out.default,
}:
comm_bytes_gb *= mesh.shape[mesh_dim]
comm_bytes_gb *= cost_mesh.shape[mesh_dim]
return allgather_cost(comm_bytes_gb, mesh_topo, mesh_dim)
elif target == torch.ops._c10d_functional.reduce_scatter_tensor.default:
return reduce_scatter_cost(comm_bytes_gb, mesh_topo, mesh_dim)
Expand Down
Loading