Stable diffusion qaihub script fails

[cccclai]

🐛 Describe the bug

The user follows the tutorial.

The users' device has SM8750 (Galaxy S25) on it, and using QNN SDK 2.37.0.250724 .
It says the image was saved, but it generates totally random noise image.
Due to the log, magic number mismatching and error resizing tensor seems problematic, but on github issue the collaborator says SD2.1 works fine on SM8750, so I wonder what is causing the problem.

The user modified some bits in qaihub_stable_diffusion.py to make the script run ;

qnn_executor_runner_args = " ".join(
        [
            f"cd {adb.workspace} &&",
            f"./qaihub_stable_diffusion_runner {' '.join(qnn_executor_runner_args)}",
        ]
    )
to 
qnn_executor_runner_args = " ".join(
            [
               f"cd {adb.workspace} && "
               f"chmod 755 ./qaihub_stable_diffusion_runner && "
               f"./qaihub_stable_diffusion_runner {' '.join(qnn_executor_runner_args)}"
           ]
    )
and
from executorch.backends.qualcomm.utils.utils import (
    from_context_binary,
    generate_htp_compiler_spec,
    generate_qnn_executorch_compiler_spec,
    get_soc_to_chipset_map,
    QcomChipset,
)
to 
from executorch.exir import ExecutorchBackendConfig
from executorch.backends.qualcomm.utils.utils import (
    from_context_binary,
    generate_htp_compiler_spec,
    generate_qnn_executorch_compiler_spec,
    get_soc_to_chipset_map,
    QcomChipset,
)

Because these resolve permission error of qaihub_stable_diffusion_runner, and import error of ExecutorchBackendConfig.

The logs is following

~/clab/executorch$ python examples/qualcomm/qaihub_scripts/stable_diffusion/qaihub_stable_diffusion.py -b build-android -m SM8750 -s {serial_number of my phone} --text_encoder_bin ../text_encoder.bin --unet_bin ../unet.bin --vae_bin ../vae.bin --vocab_json ../vocab.json --num_time_steps 20 --prompt "a photo of an astronaut riding a horse on mars" /home/dsoo/anaconda3/envs/clab/lib/python3.12/site-packages/diffusers/models/transformers/transformer_2d.py:34: FutureWarning: `Transformer2DModelOutput` is deprecated and will be removed in version 1.0.0. Importing `Transformer2DModelOutput` from `diffusers.models.transformer_2d` is deprecated and this will be removed in a future version. Please use `from diffusers.models.modeling_outputs import Transformer2DModelOutput`, instead. deprecate("Transformer2DModelOutput", "1.0.0", deprecation_message) [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters INFO:executorch.backends.qualcomm.partition.qnn_partitioner:Qnn partitioner will delegate torch mutable buffer with the same I/O address during the runtime, so if your model contains mutable buffer, then you can get the better performance with skip_mutable_buffer=False. If you encounter accuracy issue during the runtime, then please set `skip_mutable_buffer=True` and try again. [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. INFO:executorch.backends.qualcomm.qnn_preprocess:Processing Method(0): (1/1) INFO:executorch.backends.qualcomm.qnn_preprocess:Visiting: ctx_loader_0, ctx_loader_0.default [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters INFO:executorch.backends.qualcomm.partition.qnn_partitioner:Qnn partitioner will delegate torch mutable buffer with the same I/O address during the runtime, so if your model contains mutable buffer, then you can get the better performance with skip_mutable_buffer=False. If you encounter accuracy issue during the runtime, then please set `skip_mutable_buffer=True` and try again. [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. INFO:executorch.backends.qualcomm.qnn_preprocess:Processing Method(0): (1/1) INFO:executorch.backends.qualcomm.qnn_preprocess:Visiting: ctx_loader_0, ctx_loader_0.default [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters INFO:executorch.backends.qualcomm.partition.qnn_partitioner:Qnn partitioner will delegate torch mutable buffer with the same I/O address during the runtime, so if your model contains mutable buffer, then you can get the better performance with skip_mutable_buffer=False. If you encounter accuracy issue during the runtime, then please set `skip_mutable_buffer=True` and try again. [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. INFO:executorch.backends.qualcomm.qnn_preprocess:Processing Method(0): (1/1) INFO:executorch.backends.qualcomm.qnn_preprocess:Visiting: ctx_loader_1, ctx_loader_1.default [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters INFO:executorch.backends.qualcomm.partition.qnn_partitioner:Qnn partitioner will delegate torch mutable buffer with the same I/O address during the runtime, so if your model contains mutable buffer, then you can get the better performance with skip_mutable_buffer=False. If you encounter accuracy issue during the runtime, then please set `skip_mutable_buffer=True` and try again. [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in SAVE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. INFO:executorch.backends.qualcomm.qnn_preprocess:Processing Method(0): (1/1) INFO:executorch.backends.qualcomm.qnn_preprocess:Visiting: ctx_loader_2, ctx_loader_2.default [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: Running level=3 optimization. [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters ./stable_diffusion_qai_hub/qaihub_stable_diffusion_text_encoder.pte: 1 file pushed, 0 skipped. 54.3 MB/s (395891988 bytes in 6.955s) ./stable_diffusion_qai_hub/qaihub_stable_diffusion_unet.pte: 1 file pushed, 0 skipped. 70.6 MB/s (881530388 bytes in 11.915s) ./stable_diffusion_qai_hub/qaihub_stable_diffusion_vae.pte: 1 file pushed, 0 skipped. 82.9 MB/s (61068564 bytes in 0.703s) /home/dsoo/clab/qairt/2.37.0.250724/lib/aarch64-android/libQnnHtp.so: 1 file pushed, 0 skipped. 59.0 MB/s (2465168 bytes in 0.040s) /home/dsoo/clab/qairt/2.37.0.250724/lib/hexagon-v79/unsigned/libQnnHtpV79Skel.so: 1 file pushed, 0 skipped. 86.9 MB/s (9452344 bytes in 0.104s) /home/dsoo/clab/qairt/2.37.0.250724/lib/aarch64-android/libQnnHtpV79Stub.so: 1 file pushed, 0 skipped. 33.3 MB/s (713912 bytes in 0.020s) /home/dsoo/clab/qairt/2.37.0.250724/lib/aarch64-android/libQnnHtpPrepare.so: 1 file pushed, 0 skipped. 105.3 MB/s (69147960 bytes in 0.626s) /home/dsoo/clab/qairt/2.37.0.250724/lib/aarch64-android/libQnnSystem.so: 1 file pushed, 0 skipped. 64.8 MB/s (2549880 bytes in 0.038s) build-android/examples/qualcomm/qaihub_scripts/stable_diffusion/qaihub_stable_diffusion_runner: 1 file pushed, 0 skipped. 78.3 MB/s (5961976 bytes in 0.073s) build-android/backends/qualcomm/libqnn_executorch_backend.so: 1 file pushed, 0 skipped. 31.7 MB/s (646000 bytes in 0.019s) /home/dsoo/clab/qairt/2.37.0.250724/lib/aarch64-android/libQnnModelDlc.so: 1 file pushed, 0 skipped. 64.0 MB/s (2479520 bytes in 0.037s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_list.txt: 1 file pushed, 0 skipped. 0.1 MB/s (290 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_0_0.raw: 1 file pushed, 0 skipped. 0.7 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_1_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_2_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_3_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_4_0.raw: 1 file pushed, 0 skipped. 0.2 MB/s (2560 bytes in 0.014s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_5_0.raw: 1 file pushed, 0 skipped. 0.3 MB/s (2560 bytes in 0.009s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_6_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_7_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_8_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_9_0.raw: 1 file pushed, 0 skipped. 0.3 MB/s (2560 bytes in 0.009s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_10_0.raw: 1 file pushed, 0 skipped. 0.7 MB/s (2560 bytes in 0.004s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_11_0.raw: 1 file pushed, 0 skipped. 0.3 MB/s (2560 bytes in 0.010s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_12_0.raw: 1 file pushed, 0 skipped. 0.7 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_13_0.raw: 1 file pushed, 0 skipped. 0.3 MB/s (2560 bytes in 0.009s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_14_0.raw: 1 file pushed, 0 skipped. 0.3 MB/s (2560 bytes in 0.009s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_15_0.raw: 1 file pushed, 0 skipped. 0.2 MB/s (2560 bytes in 0.012s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_16_0.raw: 1 file pushed, 0 skipped. 0.3 MB/s (2560 bytes in 0.009s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_17_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_18_0.raw: 1 file pushed, 0 skipped. 0.8 MB/s (2560 bytes in 0.003s) /home/dsoo/clab/executorch/stable_diffusion_qai_hub/input_19_0.raw: 1 file pushed, 0 skipped. 0.7 MB/s (2560 bytes in 0.003s) ../vocab.json: 1 file pushed, 0 skipped. 36.9 MB/s (862328 bytes in 0.022s) I 00:00:00.000196 executorch:qaihub_stable_diffusion_runner.cpp:100] Stable Diffusion runner started I 00:00:00.000249 executorch:runner.cpp:72] creating module: model_path=/data/local/tmp/executorch/qaihub_stable_diffusion/qaihub_stable_diffusion_text_encoder.pte I 00:00:00.000254 executorch:runner.cpp:72] creating module: model_path=/data/local/tmp/executorch/qaihub_stable_diffusion/qaihub_stable_diffusion_unet.pte I 00:00:00.000259 executorch:runner.cpp:72] creating module: model_path=/data/local/tmp/executorch/qaihub_stable_diffusion/qaihub_stable_diffusion_vae.pte I 00:00:00.000263 executorch:runner.cpp:127] Loading Tokenizer from json [INFO] [Qnn ExecuTorch]: Deserializing processed data using QnnContextCustomProtocol [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: QnnContextCustomProtocol expected magic number: 0x5678abcd but get: 0x2000000 [INFO] [Qnn ExecuTorch]: Running level=1 optimization. [INFO] [Qnn ExecuTorch]: Deserializing processed data using QnnContextCustomProtocol [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: QnnContextCustomProtocol expected magic number: 0x5678abcd but get: 0x2000000 [INFO] [Qnn ExecuTorch]: Running level=1 optimization. [INFO] [Qnn ExecuTorch]: Deserializing processed data using QnnContextCustomProtocol [INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1 [INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2 [INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE. [INFO] [Qnn ExecuTorch]: QnnContextCustomProtocol expected magic number: 0x5678abcd but get: 0x2000000 [INFO] [Qnn ExecuTorch]: Running level=1 optimization. I 00:00:01.123758 executorch:runner.cpp:346] Start generating E 00:00:01.134330 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134366 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134373 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134375 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134403 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134405 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134407 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134408 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134430 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134441 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134443 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134445 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134467 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134476 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134478 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134480 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134501 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134510 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134512 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134514 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134535 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134537 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134538 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134540 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134561 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134563 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134565 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134566 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134588 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134590 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134592 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134593 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134614 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134622 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134624 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134626 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134647 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134654 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134656 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134657 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134678 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134687 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134688 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134690 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134711 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134716 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134718 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134719 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134741 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134743 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134745 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134746 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134767 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134769 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134770 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134772 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134793 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134794 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134796 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134797 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134818 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134827 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134829 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134830 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134851 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134853 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134854 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134856 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134877 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134878 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134880 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134881 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134902 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134904 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134905 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134907 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134928 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134935 executorch:method.cpp:1110] Error resizing tensor at input 0 E 00:00:01.134936 executorch:tensor_impl.cpp:83] Attempted to change the tensor rank which is immutable: old=2, new=4 E 00:00:01.134938 executorch:method.cpp:1110] Error resizing tensor at input 0 I 00:00:01.295804 executorch:runner.cpp:551] Total Number of steps: 20 I 00:00:01.295912 executorch:runner.cpp:557] Tokenizer Load Time: 0.548000 (seconds) I 00:00:01.295945 executorch:runner.cpp:563] Model Load Time: 0.576000 (seconds) I 00:00:01.295955 executorch:runner.cpp:569] Generate Time(Tokenize + Encoder + UNet + VAE): 0.168000 (seconds) I 00:00:01.295963 executorch:runner.cpp:576] Tokenize Time: 0.000000 (seconds) I 00:00:01.295971 executorch:runner.cpp:582] Text Encoder Execution Time: 0.010000 (seconds) I 00:00:01.295978 executorch:runner.cpp:588] Unet Aggregate (Cond + Uncond) Execution Time: 0.001000 (seconds) I 00:00:01.295985 executorch:runner.cpp:594] Unet Average Execution Time: 0.000000 (seconds) I 00:00:01.295992 executorch:runner.cpp:600] Unet Aggregate Post-Processing Time: 0.000000 (seconds) I 00:00:01.295999 executorch:runner.cpp:607] Unet Average Post-Processing Time: 0.000000 (seconds) I 00:00:01.296006 executorch:runner.cpp:613] VAE Execution Time: 0.156000 (seconds) [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend [INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters [INFO] [Qnn ExecuTorch]: Destroy Qnn context [INFO] [Qnn ExecuTorch]: Destroy Qnn device [INFO] [Qnn ExecuTorch]: Destroy Qnn backend /data/local/tmp/executorch/qaihub_stable_diffusion/outputs/: 1 file pulled, 0 skipped. 30.5 MB/s (3145728 bytes in 0.098s)

Versions

Post it for users, don't have this info

cc @winskuo-quic @shewu-quic @haowhsu-quic @DannyYuyang-quic @cbilgin

[zero-DS]

I am the user who questioned on discord channel. I tried other scripts for additional information.
First, executorch/examples/qualcomm/scripts/mobilenet_v2.py

output

QNN_SDK_ROOT=/home/dsoo/clab/qairt/2.37.0.250724 [WARNING] the 1 th arg of node hardtanh_ is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh_ is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__1 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__1 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__2 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__2 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__3 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__3 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__4 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__4 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__5 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__5 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__6 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__6 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__7 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__7 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__8 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__8 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__9 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__9 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__10 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__10 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__11 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__11 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__12 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__12 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__13 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__13 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__14 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__14 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__15 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__15 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__16 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__16 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__17 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__17 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__18 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__18 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__19 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__19 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__20 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__20 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__21 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__21 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__22 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__22 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__23 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__23 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__24 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__24 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__25 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__25 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__26 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__26 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__27 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__27 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__28 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__28 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__29 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__29 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__30 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__30 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__31 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__31 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__32 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__32 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__33 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__33 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__34 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__34 is NumberType, might need to lift No quant config is implemented for op, aten.dropout.default Quantizing(PTQ) the model... [WARNING] the 1 th arg of node hardtanh_ is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh_ is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__1 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__1 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__2 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__2 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__3 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__3 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__4 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__4 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__5 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__5 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__6 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__6 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__7 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__7 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__8 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__8 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__9 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__9 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__10 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__10 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__11 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__11 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__12 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__12 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__13 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__13 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__14 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__14 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__15 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__15 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__16 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__16 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__17 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__17 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__18 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__18 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__19 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__19 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__20 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__20 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__21 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__21 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__22 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__22 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__23 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__23 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__24 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__24 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__25 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__25 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__26 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__26 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__27 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__27 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__28 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__28 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__29 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__29 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__30 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__30 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__31 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__31 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__32 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__32 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__33 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__33 is NumberType, might need to lift [WARNING] the 1 th arg of node hardtanh__34 is NumberType, might need to lift [WARNING] the 2 th arg of node hardtanh__34 is NumberType, might need to lift [QNN Partitioner Op Support]: aten.linear.default | True [QNN Partitioner Op Support]: aten.view_copy.default | True [QNN Partitioner Op Support]: aten.adaptive_avg_pool2d.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.add.Tensor | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True [QNN Partitioner Op Support]: aten.hardtanh.default | True [QNN Partitioner Op Support]: aten.convolution.default | True ====== DDR bandwidth summary ====== spill_bytes=0 fill_bytes=0 write_total_bytes=4096 read_total_bytes=4499456

[QNN Partitioner Op Support]: aten.convolution.default | True
[QNN Partitioner Op Support]: aten.hardtanh.default | True
[QNN Partitioner Op Support]: aten.convolution.default | True
[QNN Partitioner Op Support]: aten.hardtanh.default | True
[QNN Partitioner Op Support]: aten.convolution.default | True
[QNN Partitioner Op Support]: aten.convolution.default | True
[QNN Partitioner Op Support]: aten.hardtanh.default | True
[QNN Partitioner Op Support]: aten.convolution.default | True
[QNN Partitioner Op Support]: aten.hardtanh.default | True
[QNN Partitioner Op Support]: aten.convolution.default | True
top_1->69.0%
top_5->86.0%

So, mobilenet seems working.

But when i compile add.pte from tutorial and then try to run it with qnn_executor_runner,

~/clab/executorch$ adb shell "cd /data/local/tmp && export LD_LIBRARY_PATH=/data/local/tmp && export ADSP_LIBRARY_PATH=/data/local/tmp && ./qnn_executor_runner --model_path ./add.pte" | tee log2.txt
I 00:00:00.000535 executorch:qnn_executor_runner.cpp:232] Model file ./add.pte is loaded.
I 00:00:00.000574 executorch:qnn_executor_runner.cpp:242] Using method forward
I 00:00:00.000579 executorch:qnn_executor_runner.cpp:289] Setting up planned buffer 0, size 48.
[INFO] [Qnn ExecuTorch]: Deserializing processed data using QnnContextCustomProtocol
[INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1
[INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2
[INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE.
[INFO] [Qnn ExecuTorch]: QnnContextCustomProtocol expected magic number: 0x5678abcd but get: 0x2000000
[ERROR] [Qnn ExecuTorch]: QnnDsp <E> Failed to create transport for device, error: 4000

[ERROR] [Qnn ExecuTorch]: QnnDsp <E> Failed to load skel, error: 4000

[ERROR] [Qnn ExecuTorch]: QnnDsp <E> Transport layer setup failed: 14001

[ERROR] [Qnn ExecuTorch]: QnnDsp <E> Failed to parse default platform info: 14001

[ERROR] [Qnn ExecuTorch]: QnnDsp <E> Failed to load default platform info: 14001

[ERROR] [Qnn ExecuTorch]: QnnDsp <E> Failed to parse platform config: 14001

[ERROR] [Qnn ExecuTorch]: Failed to create device_handle for Backend ID 6, error=14001
E 00:00:00.005913 executorch:QnnManager.cpp:340] Fail to configure Qnn device
E 00:00:00.005916 executorch:QnnExecuTorchBackend.cpp:95] Fail to initialize Qnn Manager
E 00:00:00.005920 executorch:method.cpp:114] Init failed for backend QnnBackend: 0x1
F 00:00:00.005929 executorch:qnn_executor_runner.cpp:312] In function main(), assert failed (method.ok()): Loading of method forward failed with status 0x1
Aborted

[cccclai]

But when i compile add.pte from tutorial and then try to run it with qnn_executor_runner,

I think add.pte example is the simplest model and ideally should run. I notice that we didn't push v79 so file in the old instructions. Can you try pushing it and see if it works?

[zero-DS]

Right, pushing v79 so file makes add.pte work.

add.pte log

I 00:00:00.000967 executorch:qnn_executor_runner.cpp:232] Model file add.pte is loaded.
I 00:00:00.000986 executorch:qnn_executor_runner.cpp:242] Using method forward
I 00:00:00.000990 executorch:qnn_executor_runner.cpp:289] Setting up planned buffer 0, size 48.
[INFO] [Qnn ExecuTorch]: Deserializing processed data using QnnContextCustomProtocol
[INFO] [Qnn ExecuTorch]: create QNN Logger with log_level 1
[INFO] [Qnn ExecuTorch]: Initialize Qnn backend parameters for Qnn executorch backend type 2
[INFO] [Qnn ExecuTorch]: Caching: Caching is in RESTORE MODE.
[INFO] [Qnn ExecuTorch]: QnnContextCustomProtocol expected magic number: 0x5678abcd but get: 0x2000000
[INFO] [Qnn ExecuTorch]: Running level=1 optimization.
I 00:00:00.118086 executorch:qnn_executor_runner.cpp:313] Method loaded.
E 00:00:00.118131 executorch:method.cpp:1245] Output 0 is memory planned, or is a constant. Cannot override the existing data pointer.
I 00:00:00.118143 executorch:qnn_executor_runner.cpp:373] ignoring error from set_output_data_ptr(): 0x2
I 00:00:00.118153 executorch:qnn_executor_runner.cpp:376] Inputs prepared.
I 00:00:00.118235 executorch:qnn_executor_runner.cpp:570] Input list not provided. Inputs prepared with default values set.
I 00:00:00.118861 executorch:qnn_executor_runner.cpp:579] Model executed successfully.
I 00:00:00.118875 executorch:qnn_executor_runner.cpp:582] Perform 0 inferences for warming up
I 00:00:00.118984 executorch:qnn_executor_runner.cpp:604] 1 inferences took 0.098000 ms, avg 0.098000 ms
I 00:00:00.119083 executorch:qnn_executor_runner.cpp:615] Write etdump to etdump.etdp, Size = 576
[INFO] [Qnn ExecuTorch]: Destroy Qnn backend parameters
[INFO] [Qnn ExecuTorch]: Destroy Qnn context
[INFO] [Qnn ExecuTorch]: Destroy Qnn device
[INFO] [Qnn ExecuTorch]: Destroy Qnn backend

But it seems qaihub_stable_diffusion.py already pushes required so files.

Is magic number error negligible?

However, I re-installed executorch on a new conda environment , and remembered I changed this code, too.

so i changed range to builtins.range, after importing builtins in qnn_preprocess.py

Is other people running SD2.1 modifying these small bits , too?

[cccclai]

I feel the aihub flow might not be guarded. @haowhsu-quic @shewu-quic @winskuo-quic, any thought on this?

[haowhsu-quic]

Yeah, we're thinking to deprecate this flow and let AIHUB to host ET artifacts directly in the future. Since the IO order might change once they update the pre-built context binary, which is not very friendly to predefined runner.

[cccclai]

Yeah, we're thinking to deprecate this flow and let AIHUB to host ET artifacts directly in the future. Since the IO order might change once they update the pre-built context binary, which is not very friendly to predefined runner.

That will be awesome! Let's discuss about it more