[dtensor] Remove dead code from _pointwise_ops.py by anshul-si · Pull Request #177208 · pytorch/pytorch

anshul-si · 2026-03-11T22:23:24Z

Stack from ghstack (oldest at bottom):

-> [dtensor] Remove dead code from _pointwise_ops.py #177208

The old strategy registration path (register_op_strategy-based functions
and data structures) was superseded by the register_single_dim_strategy
infrastructure but never cleaned up. This removes ~550 lines of dead
code including: pointwise_strategy, linear_pointwise_strategy,
copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy
functions, list_pointwise_strategy, list_linear_pointwise_strategy,
for_each_ops/for_each_linearity_ops/fused_ops lists, and their
associated helper sets and unused imports.

Authored with Claude.

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

pytorch-bot · 2026-03-11T22:23:27Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/177208

📄 Preview Python docs built from this PR
📄 Preview C++ docs built from this PR
❓ Need help or want to give feedback on the CI? Visit the bot commands wiki

Note: Links to docs will display an error until the docs builds have been completed.

❌ 2 New Failures

As of commit 76825af with merge base dd6252f ():

NEW FAILURES - The following jobs have failed:

pull / linux-jammy-py3.10-clang18-onnx / test (default, 1, 2, linux.c7i.2xlarge) (gh)
ninja: build stopped: subcommand failed
pull / linux-jammy-py3.10-gcc11 / test (distributed, 1, 3, linux.2xlarge) (gh)
'Test'

This comment was automatically generated by Dr. CI and updates every 15 minutes.

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: 86f42be Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: 1bef1ab Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: f8fe5ff Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: 83abc48 Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: 2a9f506 Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: eeb58ff Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: bdb932f Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: ccaf4c2 Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: 83f6815 Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: deb252c Pull Request resolved: #177208

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. ghstack-source-id: f6d6e2e Pull Request resolved: #177208

anshul-si · 2026-03-26T20:39:40Z

@pytorchbot merge

pytorchmergebot · 2026-03-26T20:43:10Z

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

pytorchmergebot · 2026-03-26T21:31:34Z

Merge failed

Reason: 1 mandatory check(s) failed. The first few are:

pull / linux-jammy-py3.10-gcc11 / test (distributed, 3, 3, lf.linux.2xlarge)

Dig deeper by viewing the failures on hud

Details for Dev Infra team

Raised by workflow job

Failing merge rule: Core Maintainers

This reverts commit c311aed. Reverted #177208 on behalf of https://github.com/georgehong due to Breaking internal signals with Undefined import [21]: Could not find a name `list_pointwise_strategy` defined in module `torch.distributed.tensor._ops._pointwise_ops`. ([comment](#177208 (comment))) Co-authored-by: Xia-Weiwen <12522207+Xia-Weiwen@users.noreply.github.com>

* Work around upstream PyTorch view op assertion in propagation rules PyTorch PR pytorch/pytorch#177973 fixed a latent bug where a comparison shard.dim == in_dim was always False (comparing int to InputDim), making the double-shard submesh_size validation in propagate_shape_and_sharding dead code. The fix activated that validation, which now raises an AssertionError when a sharded dimension is not divisible by the mesh size (e.g. unflatten nheads=48 on mesh dim size 32). This is overly strict for our use case — we call propagate_shape_and_sharding with strict_view=False to enumerate all possible sharding strategies, and incompatible ones should be silently skipped rather than raising. We now catch the AssertionError and skip the iteration, since the replicated variant is already covered by another strategy. Authored with Claude. * Fix compatibility with latest PyTorch nightly by replacing removed pointwise_strategy PyTorch removed pointwise_strategy from torch.distributed.tensor._ops._pointwise_ops in pytorch/pytorch#177208 as part of a dead code cleanup. Our custom native_layer_norm and native_layer_norm_backward rules imported this function, causing an ImportError on PyTorch nightlies after 2025-03-23. This replaces the import with a local _pointwise_strategy that recomposes the same behavior from PyTorch's new single-dim strategy primitives (_common_pointwise_single_dim_strategy, _fill_single_dim_strategy_placeholders, expand_to_full_mesh_op_strategy). Authored with Claude. * Bugfix The issue was that _get_unique_placements from PyTorch's single-dim strategy path asserts each OpStrategy has exactly 1 strategy — but in autoparallel, input OpStrategy objects have multiple strategies (all possible placements). Replaced it with an inline loop that collects unique placements from all strategies across all inputs. * One more fix The fix: _common_pointwise_single_dim_strategy produces num_outputs output entries per strategy row (e.g., 3 for native_layer_norm's out/mean/rstd). Since all pointwise outputs share the same placement, we collapse to a single output entry before passing to expand_to_full_mesh_op_strategy with input_index=1. This way strategy.output_specs is a single DTensorSpec, matching the old pointwise_strategy behavior — the callers then construct the per-output specs (out, mean, rstd) themselves. * Fix stale tensor_meta when reusing sharding placements across mesh configurations in example_dcp When reconstructing sharding placements from the serialized map for a different mesh, we were only updating .mesh on each DTensorSpec but leaving tensor_meta unchanged. Since the batch size scales with the mesh (bs = 8 * mesh.shape[0]), the tensor shapes differ between the two phases (256 vs 16 on dim 0), so the specs carried stale shapes from the original configuration. This was previously harmless — PyTorch's _maybe_unpad_tensor only used logical_dim_size (derived from tensor_meta.shape) to compute the unpad amount, and with even sharding that was a no-op. PyTorch PR pytorch/pytorch#178210 added a torch._check(orig_size >= logical_dim_size) assertion that now catches the mismatch. Fix: pull correct tensor_meta from the new AutoParallel instance's computed strategies, which reflect the actual shapes for the current mesh and batch size. Authored with Claude.

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. Pull Request resolved: #177208 Approved by: https://github.com/Skylion007 ghstack dependencies: #177186, #177187 ghstack-source-id: f005cbc

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. Pull Request resolved: #177208 Approved by: https://github.com/Skylion007 ghstack dependencies: #177186, #177187 ghstack-source-id: f4e0b82

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. Pull Request resolved: pytorch#177208 Approved by: https://github.com/Skylion007 ghstack dependencies: pytorch#177186, pytorch#177187

…7208)" This reverts commit c311aed. Reverted pytorch#177208 on behalf of https://github.com/georgehong due to Breaking internal signals with Undefined import [21]: Could not find a name `list_pointwise_strategy` defined in module `torch.distributed.tensor._ops._pointwise_ops`. ([comment](pytorch#177208 (comment)))

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. [ghstack-poisoned]

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. Pull Request resolved: #177208 Approved by: https://github.com/Skylion007 ghstack dependencies: #177186, #177187 ghstack-source-id: 5a0cf25

This reverts commit c311aed. Reverted #177208 on behalf of https://github.com/georgehong due to Breaking internal signals with Undefined import [21]: Could not find a name `list_pointwise_strategy` defined in module `torch.distributed.tensor._ops._pointwise_ops`. ([comment](#177208 (comment)))

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 [ghstack-poisoned]

…7208)" This reverts commit c311aed. Reverted pytorch#177208 on behalf of https://github.com/georgehong due to Breaking internal signals with Undefined import [21]: Could not find a name `list_pointwise_strategy` defined in module `torch.distributed.tensor._ops._pointwise_ops`. ([comment](pytorch#177208 (comment)))

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 Pull Request resolved: #178975 Approved by: https://github.com/Skylion007

The old strategy registration path (register_op_strategy-based functions and data structures) was superseded by the register_single_dim_strategy infrastructure but never cleaned up. This removes ~550 lines of dead code including: pointwise_strategy, linear_pointwise_strategy, copy_strategy, common_pointwise_strategy, single_mesh_dim_* strategy functions, list_pointwise_strategy, list_linear_pointwise_strategy, for_each_ops/for_each_linearity_ops/fused_ops lists, and their associated helper sets and unused imports. Authored with Claude. Pull Request resolved: pytorch#177208 Approved by: https://github.com/Skylion007 ghstack dependencies: pytorch#177186, pytorch#177187

…7208)" This reverts commit c311aed. Reverted pytorch#177208 on behalf of https://github.com/georgehong due to Breaking internal signals with Undefined import [21]: Could not find a name `list_pointwise_strategy` defined in module `torch.distributed.tensor._ops._pointwise_ops`. ([comment](pytorch#177208 (comment)))

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of pytorch#177208 * __->__ pytorch#178975 Pull Request resolved: pytorch#178975 Approved by: https://github.com/Skylion007

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 Pull Request resolved: #178975 Approved by: https://github.com/Skylion007

anshul-si mentioned this pull request Mar 11, 2026

[dtensor] Add single_dim_strategy infrastructure for foreach/fused ops #177186

Closed

anshul-si mentioned this pull request Mar 11, 2026

[dtensor] Register foreach and fused ops via single_dim_strategy #177187

Closed

pytorch-bot Bot added ciflow/inductor ciflow/torchtitan Run TorchTitan integration tests release notes: distributed (dtensor) release notes category labels Mar 11, 2026

anshul-si added the ciflow/trunk Trigger trunk jobs on your pull request label Mar 11, 2026

Skylion007 approved these changes Mar 11, 2026

View reviewed changes

pytorchmergebot added the merging label Mar 26, 2026

pytorchmergebot removed the merging label Mar 26, 2026

anshul-si mentioned this pull request Apr 1, 2026

[dtensor][pointwise_ops] removing dead code #178975

Closed

anshul-si closed this Apr 1, 2026

anshul-si added a commit that referenced this pull request Apr 2, 2026

Update base for Update on "[dtensor][pointwise_ops] removing dead code"

2c0482f

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 [ghstack-poisoned]

anshul-si added a commit that referenced this pull request Apr 2, 2026

Update on "[dtensor][pointwise_ops] removing dead code"

5ac05bd

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 [ghstack-poisoned]

anshul-si added a commit that referenced this pull request Apr 3, 2026

Update base for Update on "[dtensor][pointwise_ops] removing dead code"

3816762

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 [ghstack-poisoned]

anshul-si added a commit that referenced this pull request Apr 3, 2026

Update on "[dtensor][pointwise_ops] removing dead code"

f5932d0

Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): resubmission of #177208 * __->__ #178975 [ghstack-poisoned]

github-actions Bot deleted the gh/anshul-si/108/head branch May 2, 2026 02:25

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[dtensor] Remove dead code from _pointwise_ops.py#177208

[dtensor] Remove dead code from _pointwise_ops.py#177208
anshul-si wants to merge 25 commits intogh/anshul-si/108/basefrom
gh/anshul-si/108/head

anshul-si commented Mar 11, 2026 •

edited

Loading

Uh oh!

pytorch-bot Bot commented Mar 11, 2026 •

edited

Loading

Uh oh!

anshul-si commented Mar 26, 2026

Uh oh!

pytorchmergebot commented Mar 26, 2026

Uh oh!

pytorchmergebot commented Mar 26, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

anshul-si commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot Bot commented Mar 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/177208

❌ 2 New Failures

Uh oh!

anshul-si commented Mar 26, 2026

Uh oh!

pytorchmergebot commented Mar 26, 2026

Merge started

Uh oh!

pytorchmergebot commented Mar 26, 2026

Merge failed

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

anshul-si commented Mar 11, 2026 •

edited

Loading

pytorch-bot Bot commented Mar 11, 2026 •

edited

Loading