On 8/10/07, Juerg Haefliger wrote: > Hi Dave,, Hi Juerg, > > > Hi Dave, > > > > Hi Jeurg, > > > > > > Hello, > > > > > > > > I don't understand why the IN_TO_REG(val) is defined by > > > > (SENSORS_LIMIT((((val) + 8)/16),0,255)) > > > > > > (val+8)/16 is equal to val/16 with rounding, i.e. round(val/16), and > > > then the result is clipped to stay within the range 0-255. > > > > Thanks. But I just don't understand why the variable val plus 8? How > > do you know? > > The kernel can't do floating point arithmetic, it truncates the > results. So we have to add (divisor/2) before doing the division to > achieve rounding. For example if val=8, (8/16) would result in 0 (in > the kernel). But round(8/16) is actually 1, which is what we want. We > achieve this by adding 8 (which is 16/2) before dividing by 16. > > Bottom line, you have to add (divisor/2) before dividing to get > correct rounding of the result. > > > > Another example I see is IN_TO_REG(val) defined: > > > > (SENSORS_LIMIT((((val) + 4 / 8),0,255)), why is val plus 4? why not 8? Thanks. > > Same here 4 is divisor/2 (8/2). Thank you for your clear explanation. Now I completely understand, thank you very much! I'd really appreciate! > > ...juerg > > > > > > > > Hope this helps > > > ...juerg > > > > > > > > > > and IN_FROM_REG(val) is > > > > defined by ((val) * 16) > > > > > > > > Would you teach me? Thank you. > > > > > > > > BR, > > > > Dave. > > > > > > > > BR,Dave. > > >
