bits comparison

[vib]

Hi there,

Below is the code that compares bit by bit of 32-bits word. Two
errors(bits) are allowed. I am not sure if this is the right way to
check for x number of error bits. Advice and suggestion are all
welcome.

/* not compiler tested */

unsigned int BitMask;
unsigned int inWord;
unsigned int pattern;
unsigned int errCnt;
unsigned int temp;

Mask = 0x1;
errCnt = 0;
for ( i= 0; i< 32; i++)
{
temp = (inWord&BitMask ) & (pattern&BitMask ) )
if(!temp)
{
if(++errCnt > 2)
break;
}
BitMask = BitMask << 1;
}

Thanks in advance.

vib

[vib]

Oops! The corrected code

for ( i= 0; i< 32; i++)
{
if (inWord&BitMask ) != (pattern&BitMask ) )
{
if(++errCnt > 2)
break;
}
BitMask = BitMask << 1;
}

vib

[pete]

vib wrote:

Hi there,

Below is the code that compares bit by bit of 32-bits word. Two
errors(bits) are allowed. I am not sure if this is the right way to
check for x number of error bits. Advice and suggestion are all
welcome.

/* not compiler tested */

unsigned int inWord;
unsigned int pattern;
unsigned int errCnt;

errCnt = bit_count(inWord ^ pattern);

unsigned bit_count(unsigned n)
{
unsigned count;

for (count = 0; n != 0; n &= n - 1) {
++count;
}
return count;
}

--
pete

[vib]

pete wrote:

errCnt = bit_count(inWord ^ pattern);

unsigned bit_count(unsigned n)
{
unsigned count;

for (count = 0; n != 0; n &= n - 1) {
++count;
}
return count;
}

Brilliant! the n &= n -1;

Many thanks.
vib

[pete]

vib wrote:

pete wrote:

errCnt = bit_count(inWord ^ pattern);

unsigned bit_count(unsigned n)
{
unsigned count;

for (count = 0; n != 0; n &= n - 1) {
++count;
}
return count;
}

Brilliant! the n &= n -1;

I'm glad you like it. It's not original.

Also note that the solution works for any size unsigned int
with any number of bits.

--
pete

[vib]

Yup, I know. That is indeed efficient for my resource limited
situations.

[Jason Curl]

vib wrote:

Hi there,

Below is the code that compares bit by bit of 32-bits word. Two
errors(bits) are allowed. I am not sure if this is the right way to
check for x number of error bits. Advice and suggestion are all
welcome.
There is never a "right" way to do something and if it works, then it
could be considered "right". Some hints:
- C doesn't have bit-test operators, you will need to do everything with
bit manipulation operators like what you've done.
- The "XOR" operation could be used to check what bits are the same, and
what bits are different, e.g.

errCnt = 0;
temp = ~inWord ^ pattern;
while(temp && errCnt < 2) {
errCnt+=temp & 0x01;
temp >>= 1;
}

I haven't tested it, but it should give you a general idea. It's a
little faster, but not by much.

/* not compiler tested */

unsigned int BitMask;
unsigned int inWord;
unsigned int pattern;
unsigned int errCnt;
unsigned int temp;

Mask = 0x1;
errCnt = 0;
for ( i= 0; i< 32; i++)
{
temp = (inWord&BitMask ) & (pattern&BitMask ) )
Here you don't need &BitMask twice.
if(!temp)
{
if(++errCnt > 2)
break;
}
BitMask = BitMask << 1;
}

Thanks in advance.

vib

[pete]

Jason Curl wrote:

vib wrote:

Hi there,

errCnt = 0;
temp = ~inWord ^ pattern;
I don't think that the ~ belongs there.
while(temp && errCnt < 2) {
errCnt+=temp & 0x01;
temp >>= 1;
}

I haven't tested it, but it should give you a general idea. It's a
little faster, but not by much.

/* not compiler tested */

unsigned int BitMask;
unsigned int inWord;
unsigned int pattern;
unsigned int errCnt;
unsigned int temp;

Mask = 0x1;
errCnt = 0;
for ( i= 0; i< 32; i++)
{
temp = (inWord&BitMask ) & (pattern&BitMask ) )

Here you don't need &BitMask twice.

if(!temp)
{
if(++errCnt > 2)
break;
}
BitMask = BitMask << 1;
}

--
pete

[vib]

Jason Curl wrote:

There is never a "right" way to do something and if it works, then it
could be considered "right". Some hints:
- C doesn't have bit-test operators, you will need to do everything with
bit manipulation operators like what you've done.
- The "XOR" operation could be used to check what bits are the same, and
what bits are different, e.g.

Thanks for the valuable advice.

vib