On nRF52, I see that 100 kHz gives a SCL period of 10 us which is correct but for 250 kHz, I get 8 us instead of expected 4 us. Similarly, for 400 kHz I only see 5 us instead of 2.5 us. These measurements are done on an oscilloscope.
Any ideas what could be causing the problem for the higher clock frequencies? We are using SDK12.