What's a reasonable speed? Also, error message

[hullo]

I'm up and running now, no crashes [knock on wood]. It's going at about one iteration per second. Realface method, everything default iirc. Is that a reasonable rate or should it be going much faster?

I'm also getting a different error message at the start than last time, so if the above speed is abnormally slow, maybe there's a clue here:

Code: Select all

WARNING:tensorflow:Mixed precision compatibility check (mixed_float16): WARNING11/03/2023 04:09:20 INFO     No existing state file found. Generating.

Your GPU may run slowly with dtype policy mixed_float16 because it does not have compute capability of at least 7.0. Your GPU:11/03/2023 04:09:20 INFO     Storing Mixed Precision compatible layers. Please ignore any following warnings about using mixed precision.

METAL, no compute capability (probably not an Nvidia GPU)
11/03/2023 04:09:20 WARNING  Mixed precision compatibility check (mixed_float16): WARNINGSee https://developer.nvidia.com/cuda-gpus for a list of GPUs and their compute capabilities.

If you will use compatible GPU(s) not attached to this host, e.g. by running a multi-worker model, you can ignore this warning. This message will only be logged onceYour GPU may run slowly with dtype policy mixed_float16 because it does not have compute capability of at least 7.0. Your GPU:

  METAL, no compute capability (probably not an Nvidia GPU)
See https://developer.nvidia.com/cuda-gpus for a list of GPUs and their compute capabilities.
If you will use compatible GPU(s) not attached to this host, e.g. by running a multi-worker model, you can ignore this warning. This message will only be logged once

[torzdf]

Those message are standard, as we have to store mixed-precision compatible layers at start up. As it looks like you are using Apple Silicon, then that message is expected.

I can't speak for speed, as I don't have an Apple computer, however, if you run training with VERBOSE logging, it should tell you what device is being used.