[SOLVED] need max int when using stdint.h

bkelly · 04-30-2018, 01:30 PM

I am using stdint.h to name integers of a specific size and to let the compiler pick the fastest size for fastest program speed. For example

Code:

uint_fast16_t  x;
uint_fast32_t  y;

INT_MAX may no longer be correct. I looked in limits.h but did not find the name uint_fast_16_t and subsequently the max int value as it might really be a 32 or 64 bit integer.

Is there a DEFINE or macro somewhere that specifies the size of uint_fast16_t at compile time?
If not, then I presume the only I must do something like:

Code:

this_size = sizeof( x );

followed by a couple of IF statements to declare/set a maximum value.

dugan · 04-30-2018, 01:36 PM

Would this work?

Code:

x = -1;
y = -1;

NevemTeve · 04-30-2018, 02:22 PM

http://pubs.opengroup.org/onlinepubs...ag_13_48_03_02

bkelly · 04-30-2018, 05:02 PM

NevemTeve: I checked that link but do not understand part of what it is trying to tell me.

Quote:

The macro INTN_C( value) shall expand to an integer constant expression corresponding to the type int_least N _t.

I tried a few variations and the best I got was that it returned the value of what I placed in "value".

However, going to dugan's post, I did a simple test and that is the way to go. When I first through about it, subtracting one from zero in an unsigned int is not legal. But that is not correct thinking. The answer is not mathematically accurate. But it does provide right value for my needs. It is up to the programmer to be aware of the limitations of unsigned ints. Thanks for the kind tip that I overlooked.

NevemTeve · 05-01-2018, 12:46 AM

That text says there are macros like INT_FAST16_MIN, INT_FAST32_MAX, etc in stdint.h

Edit: On my PC (x86_64):
gcc/clang -m32:

Code:

INT_FAST16_MIN=-2147483648
INT_FAST16_MAX= 2147483647
INT_FAST32_MIN=-2147483648
INT_FAST32_MAX= 2147483647

gcc/clang -m64:

Code:

INT_FAST16_MIN=-9223372036854775808
INT_FAST16_MAX= 9223372036854775807
INT_FAST32_MIN=-9223372036854775808
INT_FAST32_MAX= 9223372036854775807

ntubski · 05-01-2018, 08:06 AM

Quote:

Originally Posted by bkelly

When I first through about it, subtracting one from zero in an unsigned int is not legal. But that is not correct thinking. The answer is not mathematically accurate.

By the way, the C standard defines unsigned integers to work in 2's complement, so you get the mathetically accurate result modulo 2^n. Signed integers are not defined that way, so if you go out of range, anything can happen.