| View previous topic :: View next topic |
| Author |
Message |
assaad
Joined: 09 Dec 2009
Posts: 37
|
| multiplication and division in ccs |
 Posted: Sat May 29, 2010 12:26 am |
 |
|
Hi
I am a little confused to use multiplication and division correctly in ccs.
For example:
| Code: |
unsigned int8 x;
unsigned int16 y;
unsigned int32 z;
x=200;
y=58000;
z=x*y;
|
I am using MPLAB to simulate the result, z gives wrong answer.
I tried to set x to unsigned int16, but still gives wrong answer !
Can anyone guide me to the correct multiplication, also for division ?
Also I use _mul(); but sometimes it gives correct answers sometimes not.
Thank you |
|
 |
Rohit de Sa
Joined: 09 Nov 2007
Posts: 282
Location: India
|
|
| Posted: Sat May 29, 2010 12:52 am |
|
|
This is a type conversion problem.
x*y is an (int8*int16). The result will be an int32. So x needs to be 'up-converted' into an int32 before multiplication. Try:For division, 'z' needs to be a float value, otherwise 200/58000 = 0. Use this: | Code: | float z;
z=(float)x/y; |
Rohit |
|
|
assaad
Joined: 09 Dec 2009
Posts: 37
|
|
| Posted: Sat May 29, 2010 1:12 am |
|
|
| Thank you I will try that now, but what about the usage of _mul() ? |
|
|
assaad
Joined: 09 Dec 2009
Posts: 37
|
|
| Posted: Sat May 29, 2010 2:16 am |
|
|
| Rohit de Sa wrote: | This is a type conversion problem.
x*y is an (int8*int16). The result will be an int32. So x needs to be 'up-converted' into an int32 before multiplication. Try:For division, 'z' needs to be a float value, otherwise 200/58000 = 0. Use this: | Code: | float z;
z=(float)x/y; |
Rohit |
also y is int16 , should I convert up to 32 ?
z=(int32)x*(int32)y ;
right ? |
|
|
Ttelmah
Joined: 11 Mar 2010
Posts: 19559
|
|
| Posted: Sat May 29, 2010 2:30 am |
|
|
If you have (say) two int16 values, and you multiply them together, using *, then the arithmetic is done using int16 arithmetic. So, if you take (say), 500*500, then you get 53392, which is the '16bit' result of this multiplication (low 16bits of the result).
To get the proper result, you have two choices:
1) Convert one of the values to int32 before the arithmetic. This forces int32 arithmetic to be used.
2) Use _mul. _mul performs 16bit arithmetic, but 'keeps' the overflow, to give the higher type result. Effectively converting 'up' as needed.
The later is faster than the former.
Best Wishes |
|
|
Rohit de Sa
Joined: 09 Nov 2007
Posts: 282
Location: India
|
|
| Posted: Sat May 29, 2010 10:25 am |
|
|
| Quote: | | also y is int16 , should I convert up to 32 |
No, you don't have to. As Ttelmah mentioned, int32 arithmetic is used even if one of the variables is an int32.
Rohit |
|
|
assaad
Joined: 09 Dec 2009
Posts: 37
|
|
| Posted: Mon May 31, 2010 2:58 am |
|
|
Thank you both  it is ok now .
I will be back with other questions  |
|
|
|
FAQ 
Search
Register
Profile 
Log in to check your private messages 
Log in
