Serialize NamedData in PTE file#8696
Closed
lucylq wants to merge 2 commits intogh/lucylq/45/basefrom
Closed
Conversation
1. Serialize NamedData in PTE file
2. Add NamedDataStore to EdgeProgramManager
---
Serializing NamedData is slightly different to constant/delegate data as each segment comes with its own alignment.
**An example:**
Given NamedData = {"key": data}. Data is 250 bytes.
- BackendA requires data with alignment=3
- BackendB requires data with alignment=4
Then, data0 should be serialized with alignment of lcm(3, 4) = 12
At serialization, ExecuTorch has a 'segment_alignment' that defaults to 128. Data is now serialized to lcm(12, 128) = 384.
Inside the DataSegment, we want to store the original size of the data (250). The offset of the subsequent DataSegment would be 384 bytes after the start of this one.
**Design**
Introduce a new dataclass 'AlignedData' that stores the buffer and any alignment that's required. This is used when assembling Program.segments to ensure we get lcm(buffer_alignment, segment_alignment).
Note: The default segment_alignment can be overridden inside 'ExecutorchBackendConfig'.
Differential Revision: [D69764150](https://our.internmc.facebook.com/intern/diff/D69764150/)
[ghstack-poisoned]
lucylq
requested review from
JacobSzwejbka,
larryliu0820 and
tarun292
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/8696
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New FailureAs of commit 657228e with merge base bc55c01 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D69764150 |
lucylq
added a commit
that referenced
this pull request
Feb 25, 2025
1. Serialize NamedData in PTE file
2. Add NamedDataStore to EdgeProgramManager
---
Serializing NamedData is slightly different to constant/delegate data as each segment comes with its own alignment.
**An example:**
Given NamedData = {"key": data}. Data is 250 bytes.
- BackendA requires data with alignment=3
- BackendB requires data with alignment=4
Then, data0 should be serialized with alignment of lcm(3, 4) = 12
At serialization, ExecuTorch has a 'segment_alignment' that defaults to 128. Data is now serialized to lcm(12, 128) = 384.
Inside the DataSegment, we want to store the original size of the data (250). The offset of the subsequent DataSegment would be 384 bytes after the start of this one.
**Design**
Introduce a new dataclass 'AlignedData' that stores the buffer and any alignment that's required. This is used when assembling Program.segments to ensure we get lcm(buffer_alignment, segment_alignment).
Note: The default segment_alignment can be overridden inside 'ExecutorchBackendConfig'.
Differential Revision: [D69764150](https://our.internmc.facebook.com/intern/diff/D69764150/)
ghstack-source-id: 268331257
Pull Request resolved: #8696
This PR needs a
|
1. Serialize NamedData in PTE file
2. Add NamedDataStore to EdgeProgramManager
---
Serializing NamedData is slightly different to constant/delegate data as each segment comes with its own alignment.
**An example:**
Given NamedData = {"key": data}. Data is 250 bytes.
- BackendA requires data with alignment=3
- BackendB requires data with alignment=4
Then, data0 should be serialized with alignment of lcm(3, 4) = 12
At serialization, ExecuTorch has a 'segment_alignment' that defaults to 128. Data is now serialized to lcm(12, 128) = 384.
Inside the DataSegment, we want to store the original size of the data (250). The offset of the subsequent DataSegment would be 384 bytes after the start of this one.
**Design**
Introduce a new dataclass 'AlignedData' that stores the buffer and any alignment that's required. This is used when assembling Program.segments to ensure we get lcm(buffer_alignment, segment_alignment).
Note: The default segment_alignment can be overridden inside 'ExecutorchBackendConfig'.
Differential Revision: [D69764150](https://our.internmc.facebook.com/intern/diff/D69764150/)
[ghstack-poisoned]
lucylq
added a commit
that referenced
this pull request
Feb 26, 2025
Pull Request resolved: #8696 1. Serialize NamedData in PTE file 2. Add NamedDataStore to EdgeProgramManager --- Serializing NamedData is slightly different to constant/delegate data as each segment comes with its own alignment. **An example:** Given NamedData = {"key": data}. Data is 250 bytes. - BackendA requires data with alignment=3 - BackendB requires data with alignment=4 Then, data0 should be serialized with alignment of lcm(3, 4) = 12 At serialization, ExecuTorch has a 'segment_alignment' that defaults to 128. Data is now serialized to lcm(12, 128) = 384. Inside the DataSegment, we want to store the original size of the data (250). The offset of the subsequent DataSegment would be 384 bytes after the start of this one. **Design** Introduce a new dataclass 'AlignedData' that stores the buffer and any alignment that's required. This is used when assembling Program.segments to ensure we get lcm(buffer_alignment, segment_alignment). Note: The default segment_alignment can be overridden inside 'ExecutorchBackendConfig'. ghstack-source-id: 268381885 @exported-using-ghexport Differential Revision: [D69764150](https://our.internmc.facebook.com/intern/diff/D69764150/)
tarun292
approved these changes
Feb 27, 2025
lucylq
had a problem deploying
to
cherry-pick-bot
GitHub Actions
Failure
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):
Serializing NamedData is slightly different to constant/delegate data as each segment comes with its own alignment.
An example:
Given NamedData = {"key": data}. Data is 250 bytes.
Then, data0 should be serialized with alignment of lcm(3, 4) = 12
At serialization, ExecuTorch has a 'segment_alignment' that defaults to 128. Data is now serialized to lcm(12, 128) = 384.
Inside the DataSegment, we want to store the original size of the data (250). The offset of the subsequent DataSegment would be 384 bytes after the start of this one.
Design
Introduce a new dataclass 'AlignedData' that stores the buffer and any alignment that's required. This is used when assembling Program.segments to ensure we get lcm(buffer_alignment, segment_alignment).
Note: The default segment_alignment can be overridden inside 'ExecutorchBackendConfig'.
Differential Revision: D69764150