Division 2 16Bits Integer with remainder?

neurus · Joined: 31 Mar 2004 Posts: 23 Location: Switzerland

Hi

I want to divide two 16Bits Integer digits and shown it on the display with 2 digits after the comma ( remainder ). F. e.:

Result = int1 * 100 / int2

The division has to be as exactly as possible. Wich method can you recommd me?

I'm working with CCS and pic18f252

Thank you

Pablo

newguy · Joined: 24 Jun 2004 Posts: 1907

Here is what I'd do.

ckielstra · Joined: 18 Mar 2004 Posts: 3680 Location: The Netherlands

Nice trick!! I forgot about that one.

Just a small remark: the intermediate answer will not always fit in a 16-bit variable (see the 80500 from the example). You can solve this by adding a cast to int32. Most other compilers would insert the cast to int32 automatically, but CCS can not always be trusted on this.

newguy · Joined: 24 Jun 2004 Posts: 1907

Doh!!! I didn't catch that (80500 being too big for an int16.) Thanks!

ckielstra · Joined: 18 Mar 2004 Posts: 3680 Location: The Netherlands

If you want to save code space by avoiding the 32-bit integer division, then you can multiply by 10 i.s.o. 100 and apply the same trick multiple times.
In fact, by placing it in a loop you can then generate as many digits as you want.... This is how we learned to do divisions at primary school. Cool

neurus · Joined: 31 Mar 2004 Posts: 23 Location: Switzerland

hi

thank you very much for your answers. I will test your recommendations.
What about floating point? What do you mean? Is it better, if I calculate the division with integer or float digits?

Aloha Pablo

newguy · Joined: 24 Jun 2004 Posts: 1907

Floats have their place, but it's usually best to avoid them. They take more time to calculate when compared to doing the same thing with integers. I believe that they also consume more memory (both RAM & program ROM) too.

I had an application where I tried to do a task using floats. My ROM usage (18F452) went ballistic when I tried to implement the routines. The ROM usage was ridiculous.....I remember something like 30% of the available ROM evaporated when I compiled my routine. That's when I switched to ints, and saved a lot of ROM, but at the expense of much more coding time.

Guest · Posted: Tue Apr 19, 2005 2:07 am

hi newguy

thanks for your answer. I tested the division with integer and I'm not satisfied with the result. I'm fortunate that my application is a slower application and I've enough memory to program. Here is my code:

neurus · Joined: 31 Mar 2004 Posts: 23 Location: Switzerland

hi

I'm the guest!!

Humberto · Posted: Tue Apr 19, 2005 6:28 am

This is a recurrent issue in this forum. Use of float has the advantage of
high arithmethic resolution at expenses of PIC memory resources.

Micromega Corporation has just released a 32-bit floating point
coprocessor, the uM-FPU that can be interfaced with microcontrollers
using I2C or SPI. It is a 8 PIN device and can do most of the arithmetic
operations either using float or long integers with the capabilities
of cross conversion between them.

I don�t tested yet, it�s very new and I�m trying to get one sample (in
Argentina this is not a trivial task)
I would like to know some comments from the forum�s gurues.

Humberto

Guest · Posted: Tue Apr 19, 2005 10:16 pm