Im working on a Genio510 and i moved from IoT Yocto v24.1 to v25.0, and i having problems with tflite-runtime as a python3 module, because it doesnt exists for python 3.12 .
How i can use the GPU to accelerates inferences on models, or how i can introduce Armnn delegates support? previously i tested armnn delegates and it was perfect
Hi Gian,
Thanks for reaching out.
Please note that starting from IoT Yocto v25.0, the Arm NN delegate is no longer supported. However, you can still use the GPU delegate to accelerate your model inference.
As an example, here is the execution command to run the DenseNet model with the GPU delegate:
benchmark_model --graph=/usr/share/label_image/<your_tflite_model> --use_gpu=1 --gpu_precision_loss_allowed=1 --num_runs=10
You can find more details and context via this link:
Please let me know if you have any further questions.
Best,
Jun