I tried to run the inference of our network with snpe-net-run on different devices, on CPU mode.
I tried four devices. The first one is PC, the second one is snapdragon 636, the third one is snapdragon 865, the fourth one is snapdragon 888. I have the same result on PC, snapdragon 636 and snapdragon 888. But the result on snapdragon 865 is very different. And I also check the output of every layer by adding --debug. I found the result of the first layer is not right.
Our first layer is as following:
self.first_conv = nn.Sequential(
nn.Conv2d(3, 24, 3, 2, 1, bias=False),
Dear customer,
What's target you implmented for the platform you mentioned? It works well if model accuracy exists little difference if run on different hardware. The X86 , ARM, DSP consisits of different hareware architecture.
BR.
Wei