[LocalTensor] Cache DeviceMesh.get_coordinate results in LocalTensorMode#173836
Closed
wconstab wants to merge 7 commits intogh/wconstab/511/basefrom
Closed
[LocalTensor] Cache DeviceMesh.get_coordinate results in LocalTensorMode#173836wconstab wants to merge 7 commits intogh/wconstab/511/basefrom
wconstab wants to merge 7 commits intogh/wconstab/511/basefrom
Conversation
The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
This PR needs a
|
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/173836
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 1ad1fb9 with merge base 4b0f7fb ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
wconstab
added a commit
that referenced
this pull request
Jan 29, 2026
The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. ghstack-source-id: 02756a0 Pull Request resolved: #173836
…ocalTensorMode" The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
This was referenced Jan 31, 2026
…ocalTensorMode" The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
This was referenced Jan 31, 2026
…ocalTensorMode" The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
…ocalTensorMode" The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
…ocalTensorMode" The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
wconstab
added a commit
that referenced
this pull request
Feb 3, 2026
The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. ghstack-source-id: df2f1e5 Pull Request resolved: #173836
…ocalTensorMode" The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. [ghstack-poisoned]
wconstab
added a commit
that referenced
this pull request
Feb 4, 2026
The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. ghstack-source-id: 97e54a8 Pull Request resolved: #173836
dzmitry-huba
approved these changes
Feb 6, 2026
Contributor
Author
|
@pytorchbot merge |
Collaborator
Merge failedReason: This PR needs a If not, please add the To add a label, you can comment to pytorchbot, for example For more information, see Details for Dev Infra teamRaised by workflow job |
This PR needs a
|
Contributor
Author
|
@pytorchbot merge |
Collaborator
Merge startedYour 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 |
radeksm
pushed a commit
to radeksm/pytorch
that referenced
this pull request
Feb 20, 2026
…ode (pytorch#173836) The get_coordinate method was being called repeatedly with the same DeviceMesh during operations like DTensor.from_local, recomputing the same coordinate mapping each time. This adds a per-mode cache keyed by mesh id to avoid redundant computation. In profiling of sharding rule validation, get_coordinate accounted for ~86% of from_local call time. With caching, from_local latency dropped from 4.55ms to 0.76ms (83% reduction). Authored with Claude. Pull Request resolved: pytorch#173836 Approved by: https://github.com/dzmitry-huba
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Stack from ghstack (oldest at bottom):
The get_coordinate method was being called repeatedly with the same
DeviceMesh during operations like DTensor.from_local, recomputing the
same coordinate mapping each time. This adds a per-mode cache keyed
by mesh id to avoid redundant computation.
In profiling of sharding rule validation, get_coordinate accounted for
~86% of from_local call time. With caching, from_local latency dropped
from 4.55ms to 0.76ms (83% reduction).
Authored with Claude.