efficiency match.

[Aman JIANG]

THE GAME :

Write a C function to swap the bits of a char so
that its bits would become the mirror image of the
char.MSBs become its LSBs etc. E.g. 11001100
binary would become 00110011 binary.

---------------------------------------

[Skarmander]

Aman JIANG wrote:

THE GAME :

Write a C function to swap the bits of a char so
that its bits would become the mirror image of the
char.MSBs become its LSBs etc. E.g. 11001100
binary would become 00110011 binary.

This is exciting! Here's my attempt. I hope it's of use to you. I haven't
really tested it thoroughly, so I hope there are no mistakes.

#include <stdlib.h>
#include <stdio.h>
#include <limits.h>

int swap_bits(int b) {
int result = 0;
char digits[CHAR_BIT + 1];
char digits_rev[CHAR_BIT + 1] = {0};
int n;
for (n = 0; n != CHAR_BIT; ++n) {
sprintf(digits + n, "%i", !!((1 << n) & b));
}
for (n = 0; n != CHAR_BIT; ++n) {
digits_rev[n] = digits[CHAR_BIT - n - 1];
}
for (n = 0; n != CHAR_BIT; ++n) {
result |= (digits_rev[n] - '0') << n;
}
return result;
}

S.

[Frederick Gotham]

Aman JIANG posted:

unsigned char RevChar(unsigne d char Byte)
{
typedef unsigned long tpname;
assert(sizeof(t pname) == sizeof(void*));
tpname byte = (tpname)Byte;
return (unsigned char)
(((byte & 128) >7)
| ((byte & 64) >5)
| ((byte & 32) >3)
| ((byte & 16) >1)
| ((byte & 8) << 1)
| ((byte & 4) << 3)
| ((byte & 2) << 5)
| ((byte & 1) << 7));
}

Since when is a byte eight bits?

If I wanted to mirror-image an unsigned char, I'd probably start off with
something like:

#include <limits.h>

char unsigned Mirror(char unsigned const val_byte)
{
unsigned const val = val_byte;

unsigned mirror = 0;

unsigned single_bit_val = 1U << CHAR_BIT-1;
unsigned single_bit_mirr or = 1;

do
{
if(val & single_bit_val) mirror |= single_bit_mirr or;
else mirror &= ~single_bit_mir ror;
}
while(single_bi t_mirror <<= 1,
(single_bit_val >>= 1) != (unsigned)UCHAR _MAX+1);

return mirror;
}

--

Frederick Gotham

[Aman JIANG]

Frederick Gotham wrote:

Since when is a byte eight bits?

If I wanted to mirror-image an unsigned char, I'd probably start off with
something like:

Yes, eight bits.
and... was ur code right ?

[Eric Sosman]

Aman JIANG wrote:

unsigned char RevChar(unsigne d char Byte)
{
typedef unsigned long tpname;
assert(sizeof(t pname) == sizeof(void*));
tpname byte = (tpname)Byte;
return (unsigned char)
(((byte & 128) >7)
| ((byte & 64) >5)
| ((byte & 32) >3)
| ((byte & 16) >1)
| ((byte & 8) << 1)
| ((byte & 4) << 3)
| ((byte & 2) << 5)
| ((byte & 1) << 7));
}

Yes... Yes... and ALL jeerer here,
can you overtake me on performance ?

Possibly. The C Standard says nothing about the speeds
of operations on different data types, which are inherently
implementation-dependent. On some machines, though, it will
turn out that `unsigned long' arithmetic is slower than the
ordinary `int' arithmetic you would get if you were to
work directly with `byte' instead of with `Byte'.[*]

Here's an easy optimization, though: Get rid of the
completely pointless assert() call. It asserts nothing that
is of any importance to the code, and if it takes any time
that time is pure waste.
[*] On some systems, arithmetic on `byte' will be done
in `unsigned int' rather than in `int'. However, such systems
use a character that's at least sixteen bits wide, so your
code would be wrong anyhow.

if you can, this is my homework.
BUT if you can't, it's just ur homework, okay ?

PS: donnot use TABLE plz.

Why not? Did your teacher forbid it?

--
Eric Sosman
es*****@acm-dot-org.invalid

[CBFalconer]

Frederick Gotham wrote:

>

.... snip ...

>
Since when is a byte eight bits?

Since an octal 7400 series buffer cost one dollar.

--
"The French have no word for entrepreneur." - George W. Bush
"Those who enter the country illegally violate the law."
- George W. Bush in Tucson, Ariz., Nov. 28, 2005
"I hear the voices". - G W Bush, 2006-04-18

[Richard Heathfield]

jacob navia said:

Richard Heathfield a écrit :

>ravichoudhar i said:

>>>it is just complement operation, can be done using ~ operator.

For the example posted, that would work. But it would not give a true
mirror image on many inputs. Using ~, 10000000 would become 01111111
rather than 00000001.

Yes but that's this guy problem. If he asks other people to do his/her
homework he can't ask also for correct answers :-)

I agree. I did not provide a correct answer. I merely commented on an
incorrect answer.

--
Richard Heathfield
"Usenet is a strange place" - dmr 29/7/1999
http://www.cpax.org.uk
email: rjh at above domain (but drop the www, obviously)

[Frederick Gotham]

>Since when is a byte eight bits?

CBFalconer posted:

Since an octal 7400 series buffer cost one dollar.

Aman JIANG posted:

Yes, eight bits.

The quantity of bits which comprises a byte is implementation-defined and can
be determined from the macro CHAR_BIT.

Even if 99.9999% of machines have 8-Bit bytes, that has no bearing whatsover
on the definition of the specification of the C Programming Language.

I would expect your original function to check for 8-Bit bytes:

#include <limits.h>

void Func(void)
{
#if CHAR_BIT != 8
#error "This code is not Standard C code -- bytes must be 8-Bit."
#endif
}

--

Frederick Gotham

[Aman JIANG]

Eric Sosman wrote:

Why not? Did your teacher forbid it?

1. I am NOT a Student. OK?
2. TABLE is NOT a true algorithm for this.

[Clark S. Cox III]

Aman JIANG wrote:

unsigned char RevChar(unsigne d char Byte)
{
typedef unsigned long tpname;
assert(sizeof(t pname) == sizeof(void*));

What is the point of this assert?

tpname byte = (tpname)Byte;

What is the point of this typecast?

return (unsigned char)
(((byte & 128) >7)
| ((byte & 64) >5)
| ((byte & 32) >3)
| ((byte & 16) >1)
| ((byte & 8) << 1)
| ((byte & 4) << 3)
| ((byte & 2) << 5)
| ((byte & 1) << 7));
}

Yes... Yes... and ALL jeerer here,
can you overtake me on performance ?

Quite possibly. On many platforms, a lookup table can be much faster
(especially a small one that only needs 255 elements).

if you can, this is my homework.
BUT if you can't, it's just ur homework, okay ?

PS: donnot use TABLE plz.

Why not? What if the most optimal solution uses a table?

--
Clark S. Cox III
cl*******@gmail .com

[Flash Gordon]

Aman JIANG wrote:

<snip>

PS: donnot use TABLE plz.

That means I can't do it since all the computers here are on tables.
--
Flash Gordon