float variable displaying in hexadecimal format using pointers

vlrk · 10-01-2008, 03:49 AM

Iam trying to display the flaot variable in terms of hexadecimal in byte by byte format.

Is this valid what iam doing here?

Code:

#include<stdio.h>

int main()
{
        float fVal = 10.1234;
        char *ptr;
        int i;

        ptr = &fVal;


        for(i = 0;i < 4;i++)
        {
                printf("%02x \n",*ptr);
                ptr++;
        }

        return 0;
}

output:

72
fffffff9
21
41

iam trying to find out the float data variable bit pattern , till now not with much success ( i.e how10.1234 will be converted into bits and kept in that 4 bytes).if any body can find link or tutorial that would be really benificial for me.

thanks for all the help friends

hk_mp5kpdw · 10-01-2008, 05:47 AM

Quote:

Originally Posted by vlrk

iam trying to find out the float data variable bit pattern , till now not with much success ( i.e how10.1234 will be converted into bits and kept in that 4 bytes).if any body can find link or tutorial that would be really benificial for me.

If you want to know the details of how a value, such as 10.1234 is converted into a pattern of 32 bits (4 bytes) then this may help:
http://en.wikipedia.org/wiki/IEEE_fl...point_standard

Viper187 · 10-01-2008, 07:03 AM

Someone I know showed me this trick a long time ago. I never entirely understood it though...

Code:

typedef unsigned int u32;
typedef unsigned long long int u64; //AKA int64

    float tmpfloat;
    u64 *castfloat=(u64*)&tmpfloat;
    double tmpdouble;
    u64 *castdouble=(u64*)&tmpdouble;
    u64 InputValue;
    tmpfloat = 100.0;
    InputValue = *castfloat;
    InputValue &= 0xFFFFFFFF;
    printf("%08X\n", InputValue);
    tmpdouble = 100.0;
    InputValue = *castdouble;
    printf("%I64X\n", InputValue);

nonoob · 10-01-2008, 06:04 PM

You might want to clean up the output a little:

Code:

unsigned char *ptr;
...
ptr = (unsigned char *)&fVal;
...

printf("%02X \n",*ptr);

'unsigned' so that negatives aren't, well, negative... And a capital 'X' in the format string so that you get nice capital hex digits. What I'm used to anyways.

So...
72 F9 21 41
is equivalent to
01110010111110010010000101000001

Now you can start to decode each field - sign, exponent, mantissa...

matsp · 10-01-2008, 06:14 PM

Quote:

Originally Posted by nonoob

So...
72 F9 21 41
is equivalent to
01110010111110010010000101000001

Now you can start to decode each field - sign, exponent, mantissa...

Or
01000001 00100001 11111001 01110010
depending on byte order of the machine.

Given the size of the number, I'd hazard a guess on the latter (1.0 I know is 3F800000, and 10.234 would be a bit bigger, but not dramatically so, since it's about 2^4 larger, which means 4121F972 is more likely than something starting with 72, near the limit of positive values, which makes the exponent large indeed).

--
Mats

Viper187 · 10-01-2008, 07:19 PM

Quote:

Originally Posted by matsp

Or
01000001 00100001 11111001 01110010
depending on byte order of the machine.

Given the size of the number, I'd hazard a guess on the latter (1.0 I know is 3F800000, and 10.234 would be a bit bigger, but not dramatically so, since it's about 2^4 larger, which means 4121F972 is more likely than something starting with 72, near the limit of positive values, which makes the exponent large indeed).

--
Mats

72...near the limit of positive values? 42C80000 is 100.0, 43480000 is 200.0

Nintendo 64 uses a lot of these. heheh

vlrk · 10-03-2008, 02:25 AM

These responses are indeed helping me a lot.

I request you to explain me the below lines which i found in "http://en.wikipedia.org/wiki/IEEE_floating-point_standard"

Code:

Next, the number (without the sign; i.e., unsigned, no two's complement) is converted to binary notation, giving 1110110.101. The 101 after the binary point has the value 0.625 because it is the sum of:

   1. (2 to the power of −1) × 1, from the first digit after the binary point
   2. (2 to the power of −2) × 0, from the second digit
   3. (2 to the power of −3) × 1, from the third digit.

iam trying to understand how 101 will become 625.

friends please give your thoughts.

thanks for all your inputs .

**laserlight** · 10-03-2008, 02:30 AM

Quote:

iam trying to understand how 101 will become 625.

Rather, you are trying to understand how 0.101 in base 2 is 0.625 in base 10. But before you do that explain why 101 in base 2 is 5 in base 10.

vlrk · 10-03-2008, 02:41 AM

101 = (2 power of 2) * 1+ (2 power of 1) * 0 + (2 power of 0) * 1 = 4+0+1=5

In this way

**laserlight** · 10-03-2008, 02:48 AM

Good. Notice that in converting 101 in base 2 to base 10, you began by multiplying with 2^2 (2 to the power of 2, not bitwise xor), then 2^1, and finally 2^0. Now, for 0.101 in base 2, do the same process, but multiply with 2^-1, 2^-2 and 2^-3, respectively.

vlrk · 10-03-2008, 02:55 AM

Now i got that , we have to take the fractional part if iam right here.

i.e 2^-1= 1/2 = 0.5 * 1 = 0.5
2^-2 = 1/4 = 0.25 * 0 = 0
2^-3 = 1/8 = 0.125 * 1 = 0.125

0.5+0+0.125=0.625

is it correct .. ? or some other way

**laserlight** · 10-03-2008, 03:01 AM

Yes, that is correct.

vlrk · 10-03-2008, 03:14 AM

0.625 = 2^-4 also so why cannot it be 1000.

where as to find the binary notation for any integer we do a 2 division and the remainder becomes binary notation.

similarly for this fraction part what is the method used to know the binary notation.

thanks friends

**laserlight** · 10-03-2008, 03:20 AM

Quote:

0.625 = 2^-4 also so why cannot it be 1000.

2^-4 = 0.0625 != 0.625

nonoob · 10-03-2008, 06:21 AM

I don't know where your 0.625 comes from...
Let's start again.

10.1234 = 1.265425 * 2^3
The computer scales the number down by whatever power of 2 it takes to make the number 1.nnnnnn. In this case, divided it by 8.
The integer '1' is tossed since it's always present after this scaling, so we don't need to waste a bit.

The binary representation of 1.265425 is
1.01000011111100101110010
but you didn't see the first '1'

1 + (implicit)
0.25 +
etc.

10-01-2008, 03:49 AM	#1
vlrk Registered User Join Date: Mar 2008 Location: India Posts: 72	float variable displaying in hexadecimal format using pointers Iam trying to display the flaot variable in terms of hexadecimal in byte by byte format. Is this valid what iam doing here? Code: #include<stdio.h> int main() { float fVal = 10.1234; char ptr; int i; ptr = &fVal; for(i = 0;i < 4;i++) { printf("%02x \n",ptr); ptr++; } return 0; } output: 72 fffffff9 21 41 iam trying to find out the float data variable bit pattern , till now not with much success ( i.e how10.1234 will be converted into bits and kept in that 4 bytes).if any body can find link or tutorial that would be really benificial for me. thanks for all the help friends
vlrk is offline

10-01-2008, 06:04 PM	#4
nonoob Registered User Join Date: Sep 2008 Location: Toronto, Canada Posts: 521	You might want to clean up the output a little: Code: unsigned char ptr; ... ptr = (unsigned char )&fVal; ... printf("%02X \n",*ptr); 'unsigned' so that negatives aren't, well, negative... And a capital 'X' in the format string so that you get nice capital hex digits. What I'm used to anyways. So... 72 F9 21 41 is equivalent to 01110010111110010010000101000001 Now you can start to decode each field - sign, exponent, mantissa...
nonoob is offline

10-03-2008, 02:25 AM	#7
vlrk Registered User Join Date: Mar 2008 Location: India Posts: 72	These responses are indeed helping me a lot. I request you to explain me the below lines which i found in "http://en.wikipedia.org/wiki/IEEE_floating-point_standard" Code: Next, the number (without the sign; i.e., unsigned, no two's complement) is converted to binary notation, giving 1110110.101. The 101 after the binary point has the value 0.625 because it is the sum of: 1. (2 to the power of −1) × 1, from the first digit after the binary point 2. (2 to the power of −2) × 0, from the second digit 3. (2 to the power of −3) × 1, from the third digit. iam trying to understand how 101 will become 625. friends please give your thoughts. thanks for all your inputs .
vlrk is offline

10-03-2008, 02:41 AM	#9
vlrk Registered User Join Date: Mar 2008 Location: India Posts: 72	101 = (2 power of 2) * 1+ (2 power of 1) * 0 + (2 power of 0) * 1 = 4+0+1=5 In this way
vlrk is offline

10-03-2008, 02:48 AM	#10
laserlight C++ Witch Join Date: Oct 2003 Location: Singapore Posts: 11,314	Good. Notice that in converting 101 in base 2 to base 10, you began by multiplying with 2^2 (2 to the power of 2, not bitwise xor), then 2^1, and finally 2^0. Now, for 0.101 in base 2, do the same process, but multiply with 2^-1, 2^-2 and 2^-3, respectively. __________________ C + C++ Compiler: MinGW port of GCC Build + Version Control System: SCons + Bazaar Look up a C/C++ Reference and learn How To Ask Questions The Smart Way
laserlight is offline

10-03-2008, 02:55 AM	#11
vlrk Registered User Join Date: Mar 2008 Location: India Posts: 72	Now i got that , we have to take the fractional part if iam right here. i.e 2^-1= 1/2 = 0.5 * 1 = 0.5 2^-2 = 1/4 = 0.25 * 0 = 0 2^-3 = 1/8 = 0.125 * 1 = 0.125 0.5+0+0.125=0.625 is it correct .. ? or some other way
vlrk is offline

10-03-2008, 03:01 AM	#12
laserlight C++ Witch Join Date: Oct 2003 Location: Singapore Posts: 11,314	Yes, that is correct. __________________ C + C++ Compiler: MinGW port of GCC Build + Version Control System: SCons + Bazaar Look up a C/C++ Reference and learn How To Ask Questions The Smart Way
laserlight is offline

10-03-2008, 03:14 AM	#13
vlrk Registered User Join Date: Mar 2008 Location: India Posts: 72	0.625 = 2^-4 also so why cannot it be 1000. where as to find the binary notation for any integer we do a 2 division and the remainder becomes binary notation. similarly for this fraction part what is the method used to know the binary notation. thanks friends
vlrk is offline

10-03-2008, 06:21 AM	#15
nonoob Registered User Join Date: Sep 2008 Location: Toronto, Canada Posts: 521	I don't know where your 0.625 comes from... Let's start again. 10.1234 = 1.265425 * 2^3 The computer scales the number down by whatever power of 2 it takes to make the number 1.nnnnnn. In this case, divided it by 8. The integer '1' is tossed since it's always present after this scaling, so we don't need to waste a bit. The binary representation of 1.265425 is 1.01000011111100101110010 but you didn't see the first '1' 1 + (implicit) 0.25 + etc.
nonoob is offline

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
sorting number	Leslie	C Programming	8	05-20-2009 04:23 AM
C in UNIX environment; external bubble sort doesn't work	stn0091	C Programming	2	05-13-2009 03:25 PM
Reflective Factory Design	Shamino	Game Programming	4	12-16-2005 06:50 PM
help w/another program!!!	edshaft	C++ Programming	2	12-17-2001 11:34 AM
How do you search & sort an array?	sketchit	C Programming	30	11-03-2001 05:26 PM