4 ms·
> Yet Metal optimization resources imply 16-bit arithmetic should be significantly faster, in addition to a reduction of register usage leading to higher thread
by Veedrac 6y ago
> Yet Metal optimization resources imply 16-bit arithmetic should be significantly faster, in addition to a reduction of register usage leading to higher thread count (occupancy).
I believe this is a difference between the A14 GPU and the M1 GPU; the former's 32 bit throughput is half its 16 bit throughput, whereas on the latter they are equal.