Fundamental Types

The following types are collectively called the fundamental types:

• The void type

• The scalar Boolean type

• The scalar integer types

• The scalar floating point types

Void Type {#void}

The type void contains no data and has a single unnamed value.

A function with return type void does not return a value.

Variables, arrays elements, or structure data members may not have type void.

Scalar Types {#scalar}

Boolean Type {#boolean}

Type bool is used to represent Boolean truth values: true and false.

The size of bool is target-defined. Similarly, the underlying bit patterns for true and false are target-defined. The use of bool should be avoided when a specific in-memory layout of a data structure is required. This includes data shared between different language targets even on the same device.

Integer Types {#integer}

The following integer types are defined:

Name	Description
int8_t	8-bit signed integer
int16_t	16-bit signed integer
int, int32_t	32-bit signed integer
int64_t	64-bit signed integer
uint8_t	8-bit unsigned integer
uint16_t	16-bit unsigned integer
uint, uint32_t	32-bit unsigned integer
uint64_t	64-bit unsigned integer

All arithmetic operations on signed and unsigned integers wrap on overflow.

All target platforms support the int/int32_t and uint/uint32_t types. The support for other types depends on the target and target capabilities. See target platforms for details.

All integer types are stored in memory with their natural size and alignment on all target that support them.

Floating-Point Types {#floating}

The following floating-point type are defined:

Name	Description	Precision (sign/exponent/significand bits)
half, float16_t	16-bit floating-point number	1/5/10
float, float32_t	32-bit floating-point number	1/8/23
double, float64_t	64-bit floating-point number	1/11/52

Rules for rounding, denormals, infinite values, and not-a-number (NaN) values are generally target-defined. IEEE 754 compliant targets adhere to the IEEE 754-2019 standard.

All targets support the float/float32_t type. Support for other types is target-defined. See target platforms for details.

Alignment and data layout

The size of a Boolean type is targed-defined. All other fundamental types have precisely defined sizes.

All fundamental types are naturally aligned. That is, their alignment is the same as their size.

All fundamental types use little-endian representation.

All signed integers use two’s complement representation.

📝 Remark: Fundamental types in other languages are not always naturally aligned. In particular, the alignment of C type uint64_t on x86-32 is typically 4 bytes.