3 ms·
That difference (at least in the math->numpy case) is entirely down to conversions (or at least it was when I profiled it ~7 years ago). If you're careful with
by aragilar 3y ago
That difference (at least in the math->numpy case) is entirely down to conversions (or at least it was when I profiled it ~7 years ago). If you're careful with types (i.e. avoiding converting from float to numpy scalar values), then the difference disappears.