Tensorflow 2.14 on Apple Silicon CPU (not GPU) - for performance comparison

[obriensystems]

GPU parallelism works out of the box - see #32
CPU parallelism seems to be defaulted at 33% of the cores
Attempting to optimize for M1 Max (10 core 8p/2e) and M4 Max (16 core 12p/4e)

set the threads to 4 x the total number of cores

M1Max = 10 (8p/2e) = 40 threads
M1Max = 16 (12p/4e) = 64 threads

M1 Max
default parallelism for
os.environ["OMP_NUM_THREADS"] = "10"
tf.config.threading.set_inter_op_parallelism_threads(0)
tf.config.threading.set_intra_op_parallelism_threads(0)

 Python	359.3	1:17.32	60	7	Apple	0.0	0.01	61795	michaelobrien		0 bytes	0 bytes	0 bytes	No	No	No	(null)	No	0 bytes	0	0 bytes	0	0 bytes	0 bytes	0 bytes	-	0 bytes	0 bytes	No	(null)