[ET-VK] Runtime support for NamedDataMap#13472
Merged
facebook-github-bot merged 3 commits intogh/SS-JIA/290/basefrom Aug 19, 2025
Merged
[ET-VK] Runtime support for NamedDataMap#13472facebook-github-bot merged 3 commits intogh/SS-JIA/290/basefrom
facebook-github-bot merged 3 commits intogh/SS-JIA/290/basefrom
Conversation
Summary: Allow VulkanBackend to load constant tensors from the NamedDataMap instead of the constant data section of the delegate blob. ## Motivation This enables several key results: * Unblocks delegate retargetability with other backends * Allows reducing peak memory usage when loading models by freeing constant weight data as it gets moved to the GPU ## Changes * Allow `TensorRef` to be constructed with a `FreeableBuffer` rvalue * Add ability to load constant data from `NamedDataMap` in `VulkanBackend.cpp` * When prepacking, free the constant data pointer once it's been copied to the staging buffer Test Plan: CI Validate results by collecting memory measurements in the next few diffs. [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/13472
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 1 Unrelated FailureAs of commit 491055c with merge base 8ef9595 ( NEW FAILURE - The following job has failed:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This was referenced Aug 18, 2025
This PR needs a
|
Contributor
Author
|
@SS-JIA has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator. |
Summary: Allow VulkanBackend to load constant tensors from the NamedDataMap instead of the constant data section of the delegate blob. ## Motivation This enables several key results: * Unblocks delegate retargetability with other backends * Allows reducing peak memory usage when loading models by freeing constant weight data as it gets moved to the GPU ## Changes * Allow `TensorRef` to be constructed with a `FreeableBuffer` rvalue * Add ability to load constant data from `NamedDataMap` in `VulkanBackend.cpp` * When prepacking, free the constant data pointer once it's been copied to the staging buffer Test Plan: CI Validate results by collecting memory measurements in the next few diffs. Differential Revision: [D80460035](https://our.internmc.facebook.com/intern/diff/D80460035) [ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80460035 |
lucylq
approved these changes
Aug 18, 2025
Summary: Allow VulkanBackend to load constant tensors from the NamedDataMap instead of the constant data section of the delegate blob. ## Motivation This enables several key results: * Unblocks delegate retargetability with other backends * Allows reducing peak memory usage when loading models by freeing constant weight data as it gets moved to the GPU ## Changes * Allow `TensorRef` to be constructed with a `FreeableBuffer` rvalue * Add ability to load constant data from `NamedDataMap` in `VulkanBackend.cpp` * When prepacking, free the constant data pointer once it's been copied to the staging buffer Test Plan: CI Validate results by collecting memory measurements in the next few diffs. Differential Revision: [D80460035](https://our.internmc.facebook.com/intern/diff/D80460035) [ghstack-poisoned]
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D80460035 |
5bf624b
into
gh/SS-JIA/290/base
102 of 106 checks passed
digantdesai
reviewed
Aug 19, 2025
| ref = compute_graph_->add_tensorref( | ||
| dims_vector, dtype, std::move(buffer.get())); | ||
| } else { | ||
| const uint8_t* tensor_data = constant_data_ + constant_bytes->offset(); |
Contributor
There was a problem hiding this comment.
nit assert offset != UINT64_MAX for debugging
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
4 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):
prepare_pipelines()toprepare()#13478Summary:
Allow VulkanBackend to load constant tensors from the NamedDataMap instead of the constant data section of the delegate blob.
Motivation
This enables several key results:
Changes
TensorRefto be constructed with aFreeableBufferrvalueNamedDataMapinVulkanBackend.cppTest Plan:
CI
Validate results by collecting memory measurements in the next few diffs.
Differential Revision: D80460035