Commit Graph

7 Commits

Author SHA1 Message Date
eellison
c3d205d598 helper function for replacing nodes in aug graph (#166309)
When we do bucketing, we replace starts and waits with new nodes. This pr adds a helper to transfer the augmented graph additional deps.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/166309
Approved by: https://github.com/IvanKobzarev
2025-10-29 23:08:33 +00:00
eellison
af40828bbb Limit coll bucketing within node idxs (#164944)
Respect max_coll_distance from overlap scheduler in bucketing, also, add an optimization in path searching.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/164944
Approved by: https://github.com/IvanKobzarev
ghstack dependencies: #164738, #164783
2025-10-08 22:18:53 +00:00
PyTorch MergeBot
b5e93ffdcf Revert "Limit path search within range (#164581)"
This reverts commit 415e641572.

Reverted https://github.com/pytorch/pytorch/pull/164581 on behalf of https://github.com/eellison due to merge sets makes this trickier ([comment](https://github.com/pytorch/pytorch/pull/164581#issuecomment-3381955240))
2025-10-08 14:56:21 +00:00
eellison
415e641572 Limit path search within range (#164581)
When we are looking if two nodes are dependent, limit path search within the bounds of their node idxs.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/164581
Approved by: https://github.com/ezyang
ghstack dependencies: #164568, #164569
2025-10-06 18:29:27 +00:00
eellison
92108f4abd Helper to augment graph with additional deps (#163959)
In comm-compute overlap we will have a graph with:

```
def foo(...):
     ag = all_gather(...)
     hiding_compute = mm(...)
     wait(ag)
```

There is no explicit dependency between the hiding compute and the collectives, but we want to add implicit dependencies from wait->hiding_compute, and from hiding_compute->all_gather to preserve overlap.

Additionally, while bucketing, we will merge collective starts and collective waits together. In this case, we will want to treat the two nodes as a single subgraph - each node in the merged set will have the union of all deps in the set.

This pr adds `AugmentedGraphHelper` that adds the apis, and allows querying for dependency with this augmented graph.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/163959
Approved by: https://github.com/v0i0, https://github.com/IvanKobzarev
ghstack dependencies: #163215, #163754
2025-09-30 04:53:58 +00:00
PyTorch MergeBot
84dc54ae5e Revert "Helper to augment graph with additional deps (#163959)"
This reverts commit b5d4d350f5.

Reverted https://github.com/pytorch/pytorch/pull/163959 on behalf of https://github.com/yangw-dev due to seems fails inductor/test_aten_comm_compute_reordering for macos test, see c9b5af9a38 (51526707590-box) ([comment](https://github.com/pytorch/pytorch/pull/163215#issuecomment-3349177940))
2025-09-29 21:53:42 +00:00
eellison
b5d4d350f5 Helper to augment graph with additional deps (#163959)
In comm-compute overlap we will have a graph with:

```
def foo(...):
     ag = all_gather(...)
     hiding_compute = mm(...)
     wait(ag)
```

There is no explicit dependency between the hiding compute and the collectives, but we want to add implicit dependencies from wait->hiding_compute, and from hiding_compute->all_gather to preserve overlap.

Additionally, while bucketing, we will merge collective starts and collective waits together. In this case, we will want to treat the two nodes as a single subgraph - each node in the merged set will have the union of all deps in the set.

This pr adds `AugmentedGraphHelper` that adds the apis, and allows querying for dependency with this augmented graph.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/163959
Approved by: https://github.com/v0i0, https://github.com/IvanKobzarev
ghstack dependencies: #163215, #163754
2025-09-29 20:43:12 +00:00