/tinyml

Primary LanguageC++

Tiny ML benchmark

1. Inference time

Inference time is shown to the serial monitor. This time is captured every time a new command is spoken.

img_5.png

For the sample use case the tame taken for each inference is ~72 ms.

2. Static Size and code size

To capture the code size and static

  1. First Enable verbose ouput in the Preferences Pane.

img_1.png

  1. Then after compliling the sketch we can see the command to be executed to check the code size and static data size.

img_2.png

  1. Running this command will give the code size and static size.

img_3.png

  1. Dynamic memory consumption is also captured during sketch compilation.

img_4.png