Cannot run Stable Diffusion model on NPU

[Shravanthi]

Hi,

We are trying to run stable diffusion model from optimum-intel openvino on NPU, we are getting below error. Could anyone please help to resolve this error 

Error:

(openvino_archive_env) C:\Users\tm_qc\Qualcomm\openvino_notebooks\notebooks>python .\stable_diffusion_intel_inferencing.py --model_version '2' --device NPU --batch_size 1 --prompt "red car in snowy forest, epic vista, beautiful landscape, 4k, 8k"
INFO:nncf:NNCF initialized successfully. Supported frameworks detected: torch, tensorflow, onnx, openvino
Fetching 13 files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 13/13 [00:00<?, ?it/s]
C:\Users\tm_qc\Qualcomm\openvino_archive_env\lib\site-packages\transformers\models\clip\feature_extraction_clip.py:28: FutureWarning: The class CLIPFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please use CLIPImageProcessor instead.
warnings.warn(
Compiling the vae_decoder to NPU ...
error: TilingStrategyAssignment Pass failed : Failed to tile VPU.MVN at 'loc(fused["MVN_1268523", "t_MVN"])'
An error occurred: Exception from src\inference\src\core.cpp:102:
[ GENERAL_ERROR ] Exception from src\vpux_plugin\src\plugin.cpp:432:
LevelZeroCompilerInDriver: Failed to compile network. Error code: 2013265924. Compilation failed
Failed to create executable

Thanks,

Shravanthi J

[Peh_Intel]

Hi Shravanthi J,

Please check that you are running on Windows 11 (64-bit), Intel® NPU driver is being installed and detected and NPU plugin (openvino_intel_npu_plugin.dll) is available.

·      The Intel® NPU driver for Windows is available through Windows Update but it may also be installed manually by downloading the NPU driver package and following the Windows driver installation guide.

·      If a driver has already been installed, you should be able to find ‘Intel(R) NPU Accelerator’ in Windows Device Manager. If you cannot find such a device, the NPU is most likely listed in “Other devices” as “Multimedia Video Controller.”

·      NPU plugin is currently available only with the Archive distribution of OpenVINO™ (https://storage.openvinotoolkit.org/repositories/openvino/packages/2023.3/windows/)

Next, please validate the NPU installation by running Hello Classification Python Sample with googlenet-v1 model. I attached googlenet-v1 model (FP32) for your ease.

Lastly, please share your demo script on running stable diffusion model (stable_diffusion_intel_inferencing.py) or point to us where you are referring to for duplication purpose.

Regards,

Peh

[Shravanthi]

Hi,

I am using the system Intel(R) Core(TM) Ultra 7 155H and openvino version 2023.3.0 (archive for NPU), we are able to run InceptionV4 on NPU, below are the screenshots 

I have attached stable diffusion script, Please tell me if we have some work around to run gen AI models on NPU

Thanks,

Shravanthi J

[Aznie_Intel]

Hi Shravanthi J,

You may refer to our Generative AI Optimization and Deployment for running Generative AI Models using Native OpenVINO APIs. There are some examples of using Optimum-Intel for model conversion and inference.

However, please note the NPU support in OpenVINO is still under active development and may offer a limited set of supported OpenVINO features. NPU device is currently supported by AUTO and MULTI inference modes (HETERO execution is partially supported, for certain models). You may check more information for NPU plugin features here.

Regards,

Aznie

[Aznie_Intel]

Hi Shravanthi J,

This thread will no longer be monitored since we have provided a information. If you need any additional information from Intel, please submit a new question.

Regards,

Aznie