First remember to always use the python venv in python/.venv ( add this to AGENT.md so you rember) when you want to run python scrips or install python packages.
To track Apple Neural Engine (ANE) usage, you can use macmon with the pipe command. This is useful for verifying if models are running on the ANE.
macmon pipe -s 10 -i 500The -s 10 flag sets the number of samples, and -i 500 sets the interval in milliseconds.
Example output (JSON stream):
{"temp":{"cpu_temp_avg":40.835796,"gpu_temp_avg":44.810947},"memory":{"ram_total":25769803776,"ram_usage":14942404608,"swap_total":6442450944,"swap_usage":4665114624},"ecpu_usage":[1210,0.08278669],"pcpu_usage":[1260,0.002436177],"gpu_usage":[169,0.0027520815],"cpu_power":0.06117144,"gpu_power":0.0012522695,"ane_power":0.0,"all_power":0.06242371,"sys_power":10.269848,"ram_power":0.0812975,"gpu_ram_power":0.0}
{"temp":{"cpu_temp_avg":40.835796,"gpu_temp_avg":44.810947},"memory":{"ram_total":25769803776,"ram_usage":15089975296,"swap_total":6442450944,"swap_usage":4665114624},"ecpu_usage":[1692,0.21567217],"pcpu_usage":[1364,0.0064955307],"gpu_usage":[338,0.01348789],"cpu_power":0.15673351,"gpu_power":0.03071117,"ane_power":0.0,"all_power":0.18744469,"sys_power":10.272432,"ram_power":0.14051929,"gpu_ram_power":0.0}Look for ane_power > 0.0 to confirm usage.