Faster Quantized Neural Network Inference with XNNPack(blog.tensorflow.org) |
Faster Quantized Neural Network Inference with XNNPack(blog.tensorflow.org) |
But I imagine that story is changing with the advent of Pytorch Mobile, ONNX, and that Pytorch itself supports XNNPack.
If anyone has any tips or insights as to ease of mobile deployment using TF vs using Pytorch, please share!
Both training-aware and post training.
Should I consider XNNPACK for a modern mobile phone?
CPU is the default backend in TensorFlow Lite, and CPU inference always works and produce correct result. GPU/DSP/NPU inference can be faster, particularly for large models on high-end SoCs, but generally you need to make sure that the model is supported on the IP block, the result is correct and performance is better than the CPU baseline. And that quickly gets very complicated:
1. NN API, and TFLite GPU/DSP backends support a limited subset of all TensorFlow Lite operators, and if a model is only partially offloaded to GPU/DSP/NPU, part of it will still run on CPU, and commonly synchronization overhead kills all potential speedups of the specialized hardware. The situation is even worse in CoreML, as CoreML doesn't provide an API to even learn which operators failed to offload to GPU/NPU.
2. Bugs in GPU shader compilers and NN API drivers do happen, and unless your model is a standard MobileNet, you're likely to hit them at least on some mobile phones. Then you'd need an infrastructure to detect this situation and disable offloading the model to this IP block on particular phones.
3. Low-end SoCs usually completely lack DSP and NPU, and their GPU is often slower than CPU even in nominal peak performance. This happens because CPU cores in low-end SoCs are typically just downclocked versions of the CPU cores in high-end SoCs, but low-end GPUs have 8-16 times fewer GPU cores than their high-end counterparts.