floating point computation

[alex051]

Hi,

I want to do floating point operations like multiplication, division and square root to calculate PWM duty ratio in dsPIC. I want to do these operations at high frequency.

But I am encountering problems because my dsPIC is not able to do all the floating point computations at high frequency rate. I know that fixed point operations are much faster.

So is there any method to convert floating point operations to fixed point operation?

Thanks

 

[fatalc]

Hi. Maybe you could try scaling the values?

e.g. 1.234 * 4000 can be replaced by 1234 * 4000000 and scale down the final result

I am not sure if this is what you wanted.

 

[amraldo]

You have to do it by hand, using shifting operations for multiplication and division.
I had a similar situation for color conversion functions. I replaced floating point operations to fixed point operations. Target multiplication and division operations first.
--
Amr

 

[chanchal.chauhan]

Use left and right shift operator.

Regards
Chanchal

 

[alex051]

Thanks..

i know shifting operation is faster to make division and multiplication.

But shifiting divides and multiplies only by factor of 2,4,8,16,32..etc

So what if I want to diviide by any other random number..like 10 or 100

 

[amraldo]

you have to convert it to nearest multiplication and division
--
Amr