How can I declare "New Datatype" in C?

haxpor · 07-27-2007, 01:24 AM

Some situation, involved calculation that is very large where you must allowcate some bits of memory to be larger than the standard type that C has.
So i want to know "How can I declare new type (for my arbitrary numbers of bits, says 1024 bits or something)?".

Or any ideas for that could be great, thanks.

excel28 · 07-27-2007, 02:00 AM

You can look at OpenSSL. They deal with large numbers too. One of the big numbers structures is called BIGNUM.

haxpor · 07-27-2007, 03:14 AM

excel28, thanks for pointing me to the right source BUT i am not yet know the answer.

I am now looking at BIGNUM structure (BIGNUM is name for bignum_st, real structure) in include/bn.h.
It defined as follow,

Code:

struct bignum_st
{
  BN_ULONG *d;	/* Pointer to an array of 'BN_BITS2' bit chunks. */
  int top;	/* Index of last used d +1. */
  /* The next are internal book keeping for bn_expand. */
  int dmax;	/* Size of the d array. */
  int neg;	/* one if the number is negative */
  int flags;
};

Also I see from the comment that lead to bn_expand in the same file, but now I just trying to look at it.

I keep trying...

exvor · 07-27-2007, 05:07 AM

Well C can only store data types that are big enough to be stored into a register on the computer so in essence you can only have one large enough to fit into one of them. Of course this is not a limitation as you can do some "magic" and get larger numbers to fit. I think the largest you can have is a unsigned long.

haxpor · 08-01-2007, 03:54 AM

Hmm,
If so, then in science situation in real world. How can they manage to calculate that thing? (I dont know really)

bigearsbilly · 08-01-2007, 05:34 AM

by very clever manipulation of memory and registers

psisquare · 08-01-2007, 08:00 AM

As you can see from the OpenSSL source, arbitrary precision calculations in C are highly non-trivial. Fortunately, there are libraries to ease this task, for example GMP.

ta0kira · 08-01-2007, 12:10 PM

Well, if it makes things a little more clear: In general, all of the arithmetic operations performed on built-in types are defined in asm, and therefore directly correlate to single processor operations. To go outside of those boundaries requires functions written in terms of operations on built-in types. Would you know how to calculate 1234 / 4321 not allowing any portion of the calculation to exceed 2 digits? I have no idea how to do that. Ever wonder why it takes forever to make extremely accurate scientific calculations? I think GMP is your best bet as psisquare said.
ta0kira

haxpor · 08-04-2007, 01:35 AM

Ok, thanks everyone. I will try that.