C PROGRAMMING
Data Types
| Date Published: | |
| Last Modified: |
The Basic Data Types
char | The smallest type on an architecture. Usually 8-bits wide. Sign depends on implementation. |
int | The only type that the modulus operation can be applied on. At least 16-bits in size. Usually 32-bits on 32-bit systems. |
float | Single precision floating point number. Typically 32-bits wide. |
double | Double precision floating point number. Typically 64-bits wide, although can be analogous to float (i.e. 32-bits wide) on smaller systems such as 8-bit microcontrollers. |
long double | Not normally supported on embedded platforms. 128-bits. |
void | Special case data-type. |
Using sizeof()
sizeof() can be used to return the number of bytes each type uses.
“There was a young man named Wight, Who invented the thirteen bit byte. You’ll get so much more, from your memory, I’m sure. But sadly your sizeof ain’t right.”
Fixed-width Integral Types
The problem with using int and all of it’s derivatives (short int, long int, long long int, e.t.c) is that the width of the integer is platform specific. It is normally the same width as the platforms bus, but at least 16-bits. It is also called the natural width. For example, on an 8-bit system, an int will be 16 bits wide (remember, the C standard specifies it can’t be less than 16 bits). On a 16-bit platform, it will usually be 16 bits, 32 bits for a 32-bit platform, 64 bits for a 64-bit platform, and so on, you get the idea!
To write portable code, it is usually better to use fixed-width integral types.
Fixed-width integral types also need special symbols for printf() statements. These are specified in <cinttypes.h>. They begin with the letters PRI.
Floating Point Support
Most higher-end microcontrollers and CPUs will have a hardware floating-point unit (FPU) inside them, which allows the CPU to do fast floating-point arithmetic. If you have a lower-end, cheaper microcontroller, it may not contain a FPU. In this case, you really have two options if you want to manipulate numbers with decimal precision:
- Use a software-based floating point library (it does the floating-point arithmetic using the CPU). This is usually much lower than when there is a FPU. Berkeley Softfloat is one of the most powerful, widely available software-based floating-point libraries.
- Use fixed-point numbers instead! See /programming/general/fixed-point-mathematics for more info, or check out my C++ fixed-point library for embedded systems at https://github.com/gbmhunter/MFixedPoint/.
Like this page? Upvote with shurikens!