4 ms·

I tried it on a fairly big repo today (~500k loc) and once it was done with the initial processing performance was fine.