Open shawnyxf opened 4 months ago
I wonder if anyone have tested the efficiency of ONNX with QNN EQ on cpu or htp?
Or how much faster for inference using the ONNX with QNN EQ than just ONNX alone?
It's so high level. It depends on which Qualcomm hardware you are using. You can add some code around the session run to get the number for your device.
I wonder if anyone have tested the efficiency of ONNX with QNN EQ on cpu or htp?
Or how much faster for inference using the ONNX with QNN EQ than just ONNX alone?