By "rather low limit to how many rows" do you mean that related to a performance bottleneck for your data? Roughly how many rows do you have? In one recent test, I found it converted 0.5-5M strings per second, depending on format, which could be considered fast or slow depending on the context.
What date format(s) are you looking to convert to Date format? ymd is designed to be very flexible and parses many possible format variations like "21 Jan 20" and "2021.1.20" and "21 January-20," so if you know the format(s) ahead of time, another option may be faster.