Support eq.Scalar

[cccclai]

It's reported from Discord and this pattern fails to lower to QNN

class EqualFromInplaceCopyDecomp(torch.nn.Module):
    def __init__(self, hidden_size=4):
        super().__init__()
        # a small state tensor
        self.register_buffer("h", torch.zeros((1, hidden_size)))

    def forward(self, x):
        self.h[0] = x
        return self.h[0]

Differential Revision: D86891707

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15792

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 1 New Failure

As of commit 8d9c9af with merge base 8ab589a ():

NEW FAILURE - The following job has failed:

• Lint / lintrunner / linux-job (gh)
  >>> Lint for backends/qualcomm/tests/test_qnn_delegate.py:

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[meta-codesync]

@cccclai has exported this pull request. If you are a Meta employee, you can view the originating Diff in D86891707.

[github-actions]

This PR needs a release notes: label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

[cccclai]

@haowhsu-quic @winskuo-quic @shewu-quic @DannyYuyang-quic Hi team, I might need some help for this pattern. I managed to lower the pattern to fp, but it's failing with the quantized version. I think a copy_ node shows up in the graph indicating it's a mutable buffer. Do you know the best way moving forward with this pattern?

[winskuo-quic]

@haowhsu-quic @winskuo-quic @shewu-quic @DannyYuyang-quic Hi team, I might need some help for this pattern. I managed to lower the pattern to fp, but it's failing with the quantized version. I think a copy_ node shows up in the graph indicating it's a mutable buffer. Do you know the best way moving forward with this pattern?

Hi @cccclai,

I am able to reproduce the issue where I get the following graph finalize error message:
No graph inputs present for graph [0]

I believe this has to do with the behavior of run_decomposition here:

executorch/exir/program/_program.py

Line 1176 in b433278

program = program.run_decompositions({})

The input node is actually not connected to the graph after run_decomposition for quantization flow, as shown in the image below. The copy_ node also disappeared.

On the other hand, floating point flow is working properly because the input node is still connected to the graph after run_decomposition.