Question about unit_cost of UntypedBroadcastTwo

[JAEWOOKe]

Dear all,

I tried to use open_mp for element_wise add and the cost function kept making num_threads = 1 that prohibited running in parallel.

d_of_p is 4.

I am running it on ARM board.

Do I have to change the unit_cost?

If so, how should I change it?

Thanks,
Jaewook