Skip to content

Conversation

@pull
Copy link

@pull pull bot commented Dec 19, 2025

See Commits and Changes for more details.


Created by pull[bot] (v2.0.0-alpha.4)

Can you help keep this open source service alive? 💖 Please sponsor : )

bjacob and others added 3 commits December 18, 2025 15:24
Some of theses tests were added to the Bazel build where they didn't
belong and bazel-to-cmake couldn't handle the enablement logic for
GPU-specific tests.

As GPU specific tests they are now correctly conditioned on
IREE_HIP_TEST_TARGET_CHIP and also generalized to CDNA4 as there was no
need to restrict them to CDNA3 only.

Signed-off-by: Benoit Jacob <jacob.benoit.1@gmail.com>
#22904)

See llvm/torch-mlir#4355 for more details

The PR disables decomposition `torch.aten.convolution_backward` to the
`torch` dialect - this op now will be directly lowered to `linalg`
dialect. The direct lowering should improve performance of backward
convolutions since we will assign reduction/parallel dims directly,
instead of insertion transposes/expand/collapse ops around.

#### Issues:
- #22502
- #22405

Signed-off-by: Alexandra Sidorova <asidorov@amd.com>
Tighten golden times for benchmarks that exhibit significant %
improvement over current goldens. The new values are those that have
reasonable absolute margins and are derived from the average times of
the 3 most recent passing commits with a 1.1x multiplier.

Signed-off-by: Eric Feng <Eric.Feng@amd.com>
@pull pull bot locked and limited conversation to collaborators Dec 19, 2025
@pull pull bot added the ⤵️ pull label Dec 19, 2025
@pull pull bot merged commit 6c3bd76 into VitalyAnkh:main Dec 19, 2025
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants