Swin_v2_t model fails to lower on QNN

[GregoryComer]

🐛 Describe the bug

The swin_v2_t model from torchvision does not successfully lower on the QNN backend. It fails with multiple errors. I see a number of instances of "incorrect Rank 6", though I'm not sure if this is fatal. It fails at the end, fatally, with "Failed to validate op aten_view_copy_default_17".

Output excerpt:

[QNN Partitioner Op Support]: aten.permute_copy.default | False
[ERROR] [Qnn ExecuTorch]:  <E> [4294967295] has incorrect Rank 6.

[ERROR] [Qnn ExecuTorch]:  <E> QnnBackend_validateOpConfig failed 3110

[ERROR] [Qnn ExecuTorch]:  <E> Failed to validate op aten_view_copy_default_17 with error 0xc26

[WARNING] [Qnn ExecuTorch]: Qnn Backend op validation failed with error: 3110

...

INFO:executorch.backends.qualcomm.qnn_preprocess:Visiting: aten_slice_copy_tensor_7, aten.slice_copy.Tensor
[ERROR] [Qnn ExecuTorch]: graph_prepare.cc:3490::ERROR:node 11 has input with id 0

[ERROR] [Qnn ExecuTorch]:  <E> Failed to append node qti.aisw::QNN_StridedSlice                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         [ERROR] [Qnn ExecuTorch]:  <E> Failed to construct common node for graph 257                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            [ERROR] [Qnn ExecuTorch]:  <E> Failed to add node with err 6007                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                         [ERROR] [Qnn ExecuTorch]: Failed to add node to Qnn Graph with error: 6007                                                                                                                                                                                                                                                  [ERROR] [Qnn ExecuTorch]: Fail to compile QNN graph

This can be reproduced with the following test case command or standalone script.

python -m executorch.backends.test.suite.runner models --flow qnn --filter "test_swin_v2_t_qnn_float32$"

Standalone repro:

from typing import Tuple

import executorch
import torch
import torchvision

from executorch.backends.qualcomm.utils.utils import (
    generate_qnn_executorch_compiler_spec,
    generate_htp_compiler_spec,
    QcomChipset,
    to_edge_transform_and_lower_to_qnn,
)

inputs = (torch.randn(1, 3, 224, 224),)
model = torchvision.models.swin_v2_t().eval()

ep = torch.export.export(model, inputs)

backend_options = generate_htp_compiler_spec(
    use_fp16=True,
)

compile_spec = generate_qnn_executorch_compiler_spec(
    soc_model=QcomChipset.SM8650,
    backend_options=backend_options,
)

model = to_edge_transform_and_lower_to_qnn(
    model,
    inputs,
    compile_spec
).to_executorch()

Note that running the backend test case requires executorch's python bindings to be built with the QNN backend. An example build command is below, Note that it will still need the library paths to be set up properly as described in the ET QNN docs.

CMAKE_ARGS="-DEXECUTORCH_BUILD_QNN=ON -DQNN_SDK_ROOT=$QNN_SDK_ROOT" ./install_executorch.sh --editable

Versions

Commit fbda3a9, x86-64 simulator, WSL

cc @cccclai @winskuo-quic @shewu-quic @cbilgin

[cccclai]

@haowhsu-quic can you help with triaging this issue? Note that it's reported from the backend tester, and we can cherry pick changes after branch cut. One thing to note is that, we don't need to have the full model lowered. As long as the repro command work, that's good.

[haowhsu-quic]

Will try to fix it asap.