Good morning 
I have a problem with multiplication in TS201.
I'm trying to evaluate the following expression

unsigned int a = 0xABABABAB;
unsigned int b = 0xACACACAC;
 
unsigned long long c = a*b;

However c doesn't store the correct result but it stores only the 32 least significant bits of
the 32x32 bit multiplication.
What's wrong?
I obtain the correct result only if I cast a and b to unsigned long long but acting by this way
the compiler emulates 64x64 bit operation slowing down the whole process.
Any hint will be appreciated.
Gio

Hi Gio,

What you want to do is to look up 'intrinsic functions' in the TigerSharc
compiler manual (starts on p1-126).  These functions map directly to the
machine code instruction that does the specific operation you want.  For
instance, here is the intrinsic for returning a 64 bit result from a
multiply of two 32 bit integer numbers and the corresponding assembly
instruction it maps to.

long long int __builtin_mult_i1x32_wide(int Rm, int Rn);

Rsd = Rm * Rn (I);

Hope this helps.

Cheers,

George Kadziolka

Kaztek Systems

www.kaztek.com <http://www.kaztek.com/>

  _____  

From: a...@yahoogroups.com [mailto:a...@yahoogroups.com] On Behalf Of
Giovanni Parodi
Sent: Tuesday, August 07, 2007 6:07 AM
To: a...@yahoogroups.com
Subject: [adsp] 32x32 bit multiplication on TS201

Good morning 
I have a problem with multiplication in TS201.
I'm trying to evaluate the following expression

unsigned int a = 0xABABABAB;
unsigned int b = 0xACACACAC;

unsigned long long c = a*b;

However c doesn't store the correct result but it stores only the 32 least
significant bits of the 32x32 bit multiplication.
What's wrong?
I obtain the correct result only if I cast a and b to unsigned long long but
acting by this way the compiler emulates 64x64 bit operation slowing down
the whole process.
Any hint will be appreciated.
Gio

  
Good morning 

I have a problem with multiplication in TS201.

I'm trying to evaluate the following expression

  

unsigned int a = 0xABABABAB;

unsigned int b = 0xACACACAC;

  

unsigned long long c = a*b;

  

However c doesn't store the correct result but it stores only the 32
least
significant bits of the 32x32 bit multiplication.

What's wrong?

I obtain the correct result only if I cast a and b to unsigned long
long but
acting by this way the compiler emulates 64x64 bit operation slowing
down the
whole process.

Any hint will be appreciated.

Gio