[dynamo] Reserve the tensorrt backend name for torch-tensorrt#94632
[dynamo] Reserve the tensorrt backend name for torch-tensorrt#94632narendasan wants to merge 3 commits intopytorch:masterfrom
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/94632
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ✅ No FailuresAs of commit 0156a81: This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
|
frank-wei
left a comment
There was a problem hiding this comment.
LGTM! We'd keep tensorrt for torch-tensorrt.
|
@frank-wei How can i fix the CI issues? |
|
@narendasan looks like not this PR's fault. Can you rebase? |
|
@pytorchbot rebase |
|
@pytorchbot successfully started a rebase job. Check the current status here |
|
Successfully rebased |
01d32e5 to
fc41c40
Compare
|
The failed tests are flaky which are not related to this PR. |
|
@pytorchbot rebase |
|
@pytorchbot successfully started a rebase job. Check the current status here |
|
Successfully rebased |
2011b3e to
0156a81
Compare
|
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
|
@seemethere @malfet @atalman I know this is coming late in the process but can we review this for Cherry picking into the release branch. cc: @frank-wei |
|
At the very quick glance at the PR, title does not correspond to what the code is actually doing (I.e. the registration is commented out) |
|
We didnt want to introduce a significant functional component this close to release. There doesnt seem to be a central registry for backend names so this was the closest I could think of to reserving the name i.e removing the tensorrt -> ONNX-RT mapping and putting in a placeholder file for tensorrt -> torch-trt |
In PR #93822 the `fx2trt` backend was removed which registered the `tensorrt` backend names to point to `fx2trt` / `torch_tensorrt` and move the name to `onnxrt`. We want to reserve the name `tensorrt` for `torch_tensorrt` to prevent any confusion but due to code-freeze we cannot complete the integration and set up testing for the next release. So we propose leaving out the `tensorrt` name until we can set up the backend and testing for it. Pull Request resolved: pytorch/pytorch#94632 Approved by: https://github.com/frank-wei
…pytorch#94632)" This reverts commit 3064bc4.
In PR #93822 the
fx2trtbackend was removed which registered thetensorrtbackend names to point tofx2trt/torch_tensorrtand move the name toonnxrt. We want to reserve the nametensorrtfortorch_tensorrtto prevent any confusion but due to code-freeze we cannot complete the integration and set up testing for the next release. So we propose leaving out thetensorrtname until we can set up the backend and testing for it.cc @mlazos @soumith @voznesenskym @yanboliang @penguinwu @anijain2305 @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @desertfire