SK-TDA4VM: Setting up the environment for custom model implementation

Zafar Iqbal

Part Number: SK-TDA4VM
Other Parts Discussed in Thread: TDA4VM

Hi, I am trying to setup the environment for custom model implementation on TDA4VM. I get the error as shown in the attached screenshot, while trying to run ./scripts/run_python_examples.sh by following the steps here, https://github.com/TexasInstruments/edgeai-tidl-tools/blob/master/README.md#setup

The process doesn't exit and simply pauses for an indefinite time. How to resolve this issue?

over 1 year ago

0 Takuma Fujiwara over 1 year ago

TI__Mastermind 25400 points

Hi Zafar,

Could you try uninstalling open source runtimes like ONNX RT, TFlite RT, TVM RT, and reinstall TI's fork of the runtimes that are listed in the requirements_pc.txt file within edgeai-tidl-tools?

Issue is probably related to this thread: https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1094514/tda4vm-i-try-to-use-edgeai-tidl-tools-to-compile-the-model-but-i-have-a-lot-of-problems?tisearch=e2e-sitesearch&keymatch=get_TI_benchmark_data#

Regards,

Takuma

0 Zafar Iqbal over 1 year ago in reply to Takuma Fujiwara

Prodigy 40 points

I'm getting an error when trying to install these two on the Ubuntu 20.04,
github.com/.../tvm-0.8.dev0-cp36-cp36m-linux_x86_64.whl
https://github.com/TexasInstruments/edgeai-tidl-tools/releases/download/08_02_00_01-rc1/onnxruntime_tidl-1.7.0-cp36-cp36m-linux_x86_64.whl

The error is,

onnxruntime_tidl-1.7.0-cp36-cp36m-linux_x86_64.whl is not a supported wheel on this platform.

tvm-0.8.dev0-cp36-cp36m-linux_x86_64.whl is not a supported wheel on this platform.

How can we resolve this?

0 Zafar Iqbal over 1 year ago in reply to Takuma Fujiwara

Prodigy 40 points

Also, I tried to run the python examples even though the two errors were not resolved. I cam across these errors,

1. ImportError: Python version mismatch: module was compiled for Python 3.6, but the interpreter version is incompatible: 3.8.10 (default, Jun 22 2022, 20:18:18)

2. /home/user/.local/lib/python3.8/site-packages/onnxruntime/capi/onnxruntime_inference_collection.py:55: UserWarning: Specified provider 'TIDLCompilationProvider' is not in available provider names.Available providers: 'CPUExecutionProvider'
warnings.warn("Specified provider '{}' is not in available provider names."

Please see the screenshot for details. How can we resolve this?

0 Takuma Fujiwara over 1 year ago in reply to Zafar Iqbal

TI__Mastermind 25400 points

Hi Zafar,

1. Python versions can be switched using update-alternatives on Linux. Could you try switching your Python version to 3.6?

2. This is related to the initial issue. It should be fixed if the runtime libraries from TI is installed. Since the requirements_pc.txt did not work, could you try installing the libraries from source: https://github.com/TexasInstruments

Regards,

Takuma

0 Zafar Iqbal over 1 year ago in reply to Takuma Fujiwara

Prodigy 40 points

The error is,

onnxruntime_tidl-1.7.0-cp36-cp36m-linux_x86_64.whl is not a supported wheel on this platform.

tvm-0.8.dev0-cp36-cp36m-linux_x86_64.whl is not a supported wheel on this platform.

This is from the TI libraries. How can we resolve this?

0 Takuma Fujiwara over 1 year ago in reply to Zafar Iqbal

TI__Mastermind 25400 points

Hi Zafar,

edgeai-tidl-tools is only validated on Ubuntu 18.04, and not much validation has been done with Ubuntu 20.04. The repository includes a Dockerfile for Ubuntu 18.04, so could you try building a Docker image referencing the included Dockerfile and run the examples inside a Docker container for Ubuntu 18.04?

Regards,

Takuma

Processors

Processors forum

SK-TDA4VM: Setting up the environment for custom model implementation