File-backed mmap for XNNPACK packed weights (#19862)#19862
Draft
doggeral wants to merge 1 commit into
Draft
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/19862
Note: Links to docs will display an error until the docs builds have been completed. ❗ 2 Active SEVsThere are 2 currently active SEVs. If your PR is affected, please view them below:
❌ 2 New Failures, 2 Unrelated FailuresAs of commit 24af41f with merge base 4de16d0 ( NEW FAILURES - The following jobs have failed:
BROKEN TRUNK - The following jobs 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
|
@doggeral has exported this pull request. If you are a Meta employee, you can view the originating Diff in D106673663. |
This PR needs a
|
Summary: Add file-backed mmap support to `XNNWeightsCache` so that packed weight allocations go to a `MAP_SHARED` file instead of dirty heap. After `msync(MS_ASYNC)`, pages become clean file-backed and drop out of iOS `phys_footprint`. ## How it works 1. `set_packed_cache_path()` configures the cache file path via `BackendOptions` 2. `initialize_for_runtime()` opens the cache file 3. Each `reserve_space()` call extends the file via `ftruncate` and creates a `MAP_SHARED` mmap region — XNNPACK packs weights directly into file-backed pages 4. `finalize_for_runtime()` calls `msync(MS_ASYNC)` on newly added regions only (incremental sync), making pages clean 5. On Windows, mmap is unavailable — all code paths fall back to heap allocation automatically (`packed_file_fd_` stays -1) ## Expected savings ~400 MB packed weights move from dirty heap to clean file-backed pages (0 `phys_footprint` on iOS). Differential Revision: D106673663
meta-codesync
Bot
changed the title
doggeral
force-pushed
the
export-D106673663
branch
from
eeb479e to
24af41f
Compare
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
1 participant
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:
Add file-backed mmap support to
XNNWeightsCacheso that packed weight allocations go to aMAP_SHAREDfile instead of dirty heap. Aftermsync(MS_ASYNC), pages become clean file-backed and drop out of iOSphys_footprint.How it works
set_packed_cache_path()configures the cache file path viaBackendOptionsinitialize_for_runtime()opens the cache filereserve_space()call extends the file viaftruncateand creates aMAP_SHAREDmmap region — XNNPACK packs weights directly into file-backed pagesfinalize_for_runtime()callsmsync(MS_ASYNC)on newly added regions only (incremental sync), making pages cleanpacked_file_fd_stays -1)Expected savings
~400 MB packed weights move from dirty heap to clean file-backed pages (0
phys_footprinton iOS).Differential Revision: D106673663