NRF51822 API efficiency

[scytulip]

Hi,

I'm reading and trying to understand the pstorage API. And I found a code snippet in pstorage_register (SDK 6.0.0).

do
{
    page_count++;
    if (total_size > PSTORAGE_FLASH_PAGE_SIZE)
    {
        total_size -= PSTORAGE_FLASH_PAGE_SIZE;
    }
    else
    {
        total_size = 0;
    }
    m_next_page_addr += PSTORAGE_FLASH_PAGE_SIZE;
}
while (total_size >= PSTORAGE_FLASH_PAGE_SIZE);

Why use iteration here? Does this style have higher efficiency than code like this?

if (total_size % PSTORAGE_FLASH_PAGE_SIZE == 0)
	page_count = total_size / PSTORAGE_FLASH_PAGE_SIZE;
else
	page_count = total_size / PSTORAGE_FLASH_PAGE_SIZE + 1;

[Nikita]

Division operations are very slow. In Cortex-M0 it'll take about 150 cycles as tested here.

[Clem Taylor]

Software division would not be used, because I'd assume that PSTORAGE_FLASH_PAGE_SIZE is a power of 2. The '%' becomes an '&' and the '/' becomes a '>>'

[Nikita]

Yes, that's right, in case of '%' though it'll be '<<' and '>>'.