Is there any reason this simple GPIO timing should be much faster with SDK 11.0 ?

[Russ Mirov]

I've got an application that does a lot of nRF51822 GPIO wiggling and the timing needs to be controlled. In one spot I simply set a pin high then low like this:

nrf_gpio_cfg_output(5);
nrf_gpio_pin_set(5);
nrf_gpio_pin_clear(5);

I've ensured that the 16MHz clock is running like this:

// Start 16 MHz crystal oscillator.
NRF_CLOCK->EVENTS_HFCLKSTARTED = 0;
NRF_CLOCK->TASKS_HFCLKSTART    = 1;
// Wait for the external oscillator to start up.
while (NRF_CLOCK->EVENTS_HFCLKSTARTED == 0) {  };  // Spin waiting for HFclock to start 

OK, so the strange thing is that with the older SDK the pulse high time was about 1uSec. With the new SDK it is about 186nSec. That sounds like it is now taking three 16MHz clock cycles to run. And it seemed to take a lot longer before. Anyone have an explanation ? Did "nrf_gpio_pin_set/clear" get a lot faster ?

Thanks, -Russ

[RK]

why don't you look at the generated assembler and see for yourself. You're not going to explain a difference of a couple of clock cycles without doing so.

[Ole Bauck]

These functions are inline functions. Inline instructs the compiler to attempt to embed the function content into the calling code instead of executing an actual call, see this thread. Depending on your project and compiler optimization, the functions may be three lines of assembly, or several. You should look at the generated assembly code as RK suggested. You can for example do this in the debug window.