Repository navigation
Generate ring masks directly as packed bits (#24089) - #24089
Merged
Merged
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24089
Note: Links to docs will display an error until the docs builds have been completed. ⏳ 1 Pending, 1 Unrelated FailureAs of commit 2536527 with merge base 8916832 ( 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. |
meta-cla
Bot
added
the
CLA Signed
Contributor
|
@ThomasJannaud has exported this pull request. If you are a Meta employee, you can view the originating Diff in D126734014. |
This PR needs a
|
meta-codesync
Bot
changed the title
ThomasJannaud
added a commit
to ThomasJannaud/executorch
that referenced
this pull request
Oct 22, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Differential Revision: D126734014
ThomasJannaud
force-pushed
the
export-D126734014
branch
from
October 22, 2026 19:25
f8e5665 to
24544d1
Compare
ThomasJannaud
force-pushed
the
export-D126734014
branch
from
October 22, 2026 21:57
24544d1 to
f347ea7
Compare
ThomasJannaud
added a commit
to ThomasJannaud/executorch
that referenced
this pull request
Oct 22, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Differential Revision: D126734014
ThomasJannaud
force-pushed
the
export-D126734014
branch
from
October 23, 2026 17:16
d0aeceb to
0b63110
Compare
ThomasJannaud
added a commit
to ThomasJannaud/executorch
that referenced
this pull request
Oct 23, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Differential Revision: D126734014
ThomasJannaud
added a commit
to ThomasJannaud/executorch
that referenced
this pull request
Oct 23, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Differential Revision: D126734014
ThomasJannaud
force-pushed
the
export-D126734014
branch
from
October 23, 2026 17:18
0b63110 to
9339d61
Compare
ThomasJannaud
added a commit
to ThomasJannaud/executorch
that referenced
this pull request
Oct 23, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Reviewed By: mvartani-meta Differential Revision: D126734014
ThomasJannaud
force-pushed
the
export-D126734014
branch
from
October 23, 2026 17:41
9339d61 to
4db6155
Compare
ThomasJannaud
added a commit
to ThomasJannaud/executorch
that referenced
this pull request
Oct 23, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Reviewed By: mvartani-meta Differential Revision: D126734014
mvartani-meta
approved these changes
Oct 23, 2026
Summary: Dense ring masks require an avoidable allocation and packing pass. Add `sdpa_bitwise_ring_mask_gen` schemas, meta/reference implementations and a Cadence C++ out kernel that writes packed `uint8` bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes, `int32`/`int64` positions including `INT64_MAX`, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553. AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition. Reviewed By: mvartani-meta Differential Revision: D126734014
mcremon-meta
approved these changes
Oct 23, 2026
This branch was successfully deployed
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
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.
Summary:
Dense ring masks require an avoidable allocation and packing pass. Add
sdpa_bitwise_ring_mask_genschemas, meta/reference implementations and a Cadence C++ out kernel that writes packeduint8bits directly from token positions, sliding window and cache capacity. Keep the existing Python regressions and add seven public direct-call C++ unit tests for packed byte order and polarity, ring wrap, all windows on small caches, realistic cache sizes,int32/int64positions includingINT64_MAX, empty inputs, invalid shapes/dtypes/windows/capacities and rejection of negative positions before writing. Wire the tests into the existing Buck and Cadence CMake test targets. This commit contains only ExecuTorch sources and synchronized mirrors; Jarvis registration remains in D126973553.AI assistance: Codex separated the history and added the public C++ unit tests and build wiring. The production kernel is unchanged by this test addition.
Reviewed By: mvartani-meta
Differential Revision: D126734014