Float variable in register storage class

Hi

i ran the following code on QNX platform using gcc compiler:

register float value;
value =1.2;
printf("Value= %f",value);
Console output: 1.2
I read somewhere that if the above snippet is run, the integer part of the float variable is returned (due to the register storage class) but isn't true in the above case.
How is a float variable stored inside the memory in case of the register storage class?
Thanx

You can declare all sorts of things as being 'register', but that doesn't actually mean a great deal (the compiler is free to ignore it).

The only consistent and provable thing that will happen though is that it prevents you from creating pointers to things declared register.

Any compiler that did such a thing would be a faulty compiler.


However the compiler chooses. All the register keyword does is provide a hint to the compiler. The compiler is perfectly free to ignore that hint.

In most cases on most architectures the compiler will ignore it (I've never come across one that didn't). The compiler knows better how to optimise code for speed than you do, such manual hints are unnecessary and doomed to have little or no effect. Consult the compiler documentation for what it actually does.

GCC does at least allow you to force the use of specific registers (http://gcc.gnu.org/onlinedocs/gcc/Explicit-Reg-Vars.html#index-explicit-register-variables-2673), but in normal ISO C usage, I believe ignores it.


Most processors have precious few general purpose registers (especially x86), hence the directive is ignored. The ARM ABI uses the first four GP registers to pass parameters.

For a floating point variable, register storage (whether directed or not) may be affected by the presence or otherwise of an FPU.

A single precision floating-point value is 32 bits, on a 32 bit processor it will fit entirely in a register. FPU registers are larger if the FPU supports double precision.