Pin ExecuTorch to the release/1.4 branch head - #4434
Open
shoumikhin wants to merge 1 commit into
Open
Conversation
main pins the ExecuTorch dependency to the release/1.3 branch head (6118688a), which predates the shared extension_cuda caller-stream work. Bump the pin to the release/1.4 branch head (cd380e7aefd18c171271cc228d3a155455095219), the first ExecuTorch release line that contains both pytorch/executorch#20158 (shared extension_cuda caller-stream library) and pytorch/executorch#20498 (caller CUDA stream for H2D/D2H copies). This follows the existing convention of pinning a release-branch head commit; the new_git_repository rule takes a commit, not a branch name. The four Bazel files that pin ExecuTorch are updated identically: MODULE.bazel, docker/MODULE.bazel.docker, docker/MODULE.bazel.ngc, and toolchains/ci_workspaces/MODULE.bazel.tmpl. Swap the branch-head commit for the immutable v1.4.0 tag once ExecuTorch publishes it.
shoumikhin
force-pushed
the
pin-executorch-release-1.4
branch
from
July 28, 2026 22:05
7b76b42 to
8002a42
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why this PR is needed
Torch-TensorRT currently builds native ExecuTorch code from the ExecuTorch 1.3 release branch.
New native integrations, including the shared CUDA caller-stream support in #4421, require APIs that are available in ExecuTorch 1.4. This PR updates the repository-wide source pin once, in a small standalone change, so feature PRs do not each carry their own dependency bump.
What changes
The same pin is updated in all four Bazel workspace files:
Files changed:
MODULE.bazeldocker/MODULE.bazel.dockerdocker/MODULE.bazel.ngctoolchains/ci_workspaces/MODULE.bazel.tmplNo Torch-TensorRT runtime or Python behavior changes directly in this PR. It only selects the newer ExecuTorch source revision for native builds.
Example
A native build continues to use the normal TensorRT targets:
The difference is that Bazel now fetches ExecuTorch from the 1.4 release line.
Dependency order
Merge this PR before #4421. After this lands, #4421 can rebase onto
mainand its temporary copy of these four pin changes will disappear.Validation
Current GitHub CI completed the full matrix. Most jobs passed. The remaining failed jobs are in unrelated model/runtime tests and package-environment setup, for example Windows platform-name mismatches, missing TensorRT shared libraries, model test failures, and JetPack package-index dependency resolution. None of those failures point to the four pin files changed here.
Follow-up: replace the release-branch head with the immutable
v1.4.0tag commit when ExecuTorch publishes that tag.