Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Float in C

  1. Aug 29, 2010 #1
    Hello,
    I want to know the answer of following program and the reason behind it..........as soon as possible

    void main()
    {
    float j;
    j=1000*1000;
    printf("%f",j);
    }
     
  2. jcsd
  3. Aug 29, 2010 #2

    Mark44

    Staff: Mentor

    What do you get when you attempt to compile this code?
     
  4. Aug 30, 2010 #3
    It answers as 16960.00000
    but i dunno y?????????
     
  5. Aug 30, 2010 #4

    Borek

    User Avatar

    Staff: Mentor

    Does it? It doesn't compile here without additional headers, and it displays 1000000.000000 as a result.
     
  6. Aug 30, 2010 #5

    Mark44

    Staff: Mentor

    OK, I see what's going on. I'm pretty sure that your compiler uses two-byte (signed, probably) ints, which can hold a largest (signed) value of 32,767. More modern compilers use four bytes for ints.

    The compiler does the multiplication using two-byte quantities, getting a value of 16960.

    This is easier to see if you work with hex (base-16) numbers. In hex, each pair of bytes is one byte, so a four-digit hex number fits in two bytes.
    1000 * 1000 in base 10 is the same as 3E8 * 3E8 in hex, which is F4240 in hex (or 1,000,000 in decimal).

    Since the multiplication is done using two-byte integers, there is no room for the F digit, so the result is truncated to the lower two bytes, leaving 4240 (hex), which is 16960 (decimal).

    This value is stored in the float variable j, and is later displayed by the printf statement. A float can store much larger numbers, but the damage has already been done, and you get the result that you did.

    You can force the multiplication to be done using long ints by appending L to the two constants, as in the following code.

    Also, I have #included "stdio.h" to provide the prototype for printf.

    Code (Text):

    #include "stdio.h"

    void main()
    {
       float j;
       j=1000L*1000L;
       printf("%f",j);
    }
     
    Last edited: Aug 30, 2010
  7. Aug 30, 2010 #6

    Borek

    User Avatar

    Staff: Mentor

    LOL, I was working with these compilers for many years (Z80, then 8086) but somehow I didn't catch what was going on.

    Shame on me :grumpy:
     
  8. Aug 31, 2010 #7

    Mark44

    Staff: Mentor

    Don't be so hard on yourself. We've been working with compilers that use 4-byte ints for what, 14 or 15 years?

    1000 * 1000 == 16960 is sort of a surprising result. It was something of a fluke that it occurred to me that that maybe truncation was causing this output. Once I started down that path, I discovered that 0xF4240 - 0xF0000 = 0x4240, which is 16960 in decimal.
     
  9. Aug 31, 2010 #8

    Borek

    User Avatar

    Staff: Mentor

    Once you posted about 2 bytes I started to think in terms of 1000*1000 mod 65536 - same result :smile:
     
  10. Aug 31, 2010 #9
    I think you guys are being too hard on yourselves. Off the top of my head I believe you will find a pragma or compiler option that controls implicit casts and addresses this.

    This is one of those things that really was not underscored until I played with C++ and the distinction of assignment versus initialization had to be made. The whole scope of operator overloading and the casting operator. ( I wrote a ton of printf debugging programs to understand how the c++ compiler treated expressions with implied casting ).

    In c++ it is not straightforward, and is compiler dependent.
     
  11. Aug 31, 2010 #10

    rcgldr

    User Avatar
    Homework Helper

    You could have done the multiplication using floating point by adding a period to each number:

    j = 1000. * 1000. ;
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook




Similar Discussions: Float in C
  1. Using float in C (Replies: 3)

  2. To C or not to C (Replies: 31)

  3. C or C++? (Replies: 8)

  4. Floating point numbers. (Replies: 18)

Loading...