The table below describes the list of builtin math functions. These functions can take scalar or vector arguments. Click an item in the table for details about that function.
The table below describes the list of builtin math functions that are defined with the half_
or native_
prefix
Description
General information about builtin math functions: The builtin math functions are categorized into the following:

A list of builtin functions that have scalar or vector argument versions, and,

A list of builtin functions that only take scalar float arguments.
The vector versions of the math functions operate componentwise. The description is percomponent.
The builtin math functions are not affected by the prevailing rounding mode in the calling environment, and always return the same value as they would if called with the round to nearest even rounding mode.
The builtin math functions take scalar or vector arguments. For any specific use of these function, the actual type has to be the same for all arguments and the return type unless otherwise specified.
The generic type name gentype
is used to indicate that the function can take float
, float2
, float3
, float4
, float8
, float16
, double
, double2
, double3
, double4
, double8
, or double16
as the type for the arguments.
If extended with cl_khr_fp16
, generic type name gentype
may indicate half
and half{234816}
as arguments and return values.
The generic type name gentypef
is used to indicate that the function can take float
, float2
, float3
, float4
, float8
, or float16
as the type for the arguments.
The generic type name gentyped
is used to indicate that the function can take double
, double2
, double3
, double4
, double8
, or double16
as the type for the arguments.
The math functions that are defined with the half_
prefix are implemented with a minimum of 10bits of accuracy i.e.
an ULP value ⇐ 8192 ulp.
The math functions that are defined with the native_
prefix may map to one or more native device instructions and will typically have better performance compared to the corresponding functions (without the native_
prefix) described.
The accuracy (and in some cases the input range(s)) of these functions is implementationdefined.