-
Notifications
You must be signed in to change notification settings - Fork 776
Arm backend: Add conv3d support to Tosa/Vgf backends #16093
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/16093
Note: Links to docs will display an error until the docs builds have been completed. ❌ 2 Cancelled Jobs, 4 Unrelated FailuresAs of commit 51335da with merge base 2bc465a ( CANCELLED JOBS - The following jobs were cancelled. Please retry:
FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
BROKEN TRUNK - The following job failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Conv3d is not supported by vela so it can not be supported in u55 or u85. * Adds Conv3D support for FP32, Int8 and int16A8W * Reworks to_tosa_memory_format_pass.py to handle spatial rank 3 tensors (DHW) * Adds support for rank 5 tensors to analyze_output_utils.py * Reworks conv2d passes to handle conv3d and renames them to be more generic Signed-off-by: Ryan O'Shea <ryan.oshea3@arm.com> Change-Id: I0c888b46afe7bdfa0a26c26d5281ab02b945b528
Signed-off-by: Ryan O'Shea <ryan.oshea3@arm.com> Change-Id: I9b096c71da7e1bf943857c04c5606e22ba35da4f
Signed-off-by: Ryan O'Shea <ryan.oshea3@arm.com> Change-Id: Ibce8fdf18a8e3fb2f16a9f5cb0c07d167ba2514c
7a4b5ab to
40a4202
Compare
|
Arm tests failing on test_index_tensor_tosa @oscarandersson8218 is working on a fix |
zingo
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Approved, just make sure CI is green
Conv3d is not supported by vela so it can not be supported in u55 or u85. * Adds Conv3D support for FP32, Int8 and int16A8W * Reworks to_tosa_memory_format_pass.py to handle spatial rank 3 tensors (DHW) * Adds support for rank 5 tensors to analyze_output_utils.py * Reworks conv2d passes to handle conv3d and renames them to be more generic Signed-off-by: Ryan O'Shea <ryan.oshea3@arm.com>
Conv3d is not supported by vela so it can not be supported in u55 or u85. * Adds Conv3D support for FP32, Int8 and int16A8W * Reworks to_tosa_memory_format_pass.py to handle spatial rank 3 tensors (DHW) * Adds support for rank 5 tensors to analyze_output_utils.py * Reworks conv2d passes to handle conv3d and renames them to be more generic Signed-off-by: Ryan O'Shea <ryan.oshea3@arm.com>
Conv3d is not supported by vela so it can not be supported
in u55 or u85.
Change-Id: I0c888b46afe7bdfa0a26c26d5281ab02b945b528
cc @freddan80 @per @zingo @oscarandersson8218 @digantdesai