How fast can be a model?

I am looking for some theoretical calculator on various models and their theoretical performance, so to speak its inference time, for declared theoretical accelerator HW OPs (e.g. TOPs, GFLOPs, etc…). So I am able roughly to estimate what can be a potential inference performance (inf/s).
If there are some calculation guidelines publicly available, ideally for TensorFlow Lite or Tensorflow Micro, please can you please point me?


You can check this thread

1 Like