simdpp::to_mask
From libsimdpp-docs
< conv
| template<unsigned N, class E>
Ret<N, _DETAIL_> to_mask( Vec<N,E> a ); |
||
Converts a vector to a mask.
The values of the elements in the source vector must be either all ones or all zeros. Otherwise the behavior is undefined.
In instruction sets that support optimized mask types (e.g. AVX512), the operation results in a conversion instruction being executed. In instruction sets where masks are implemented as regular vectors this function is equivalent to a simple type conversion and does not result in any instructions being executed.
The implementation behaves as if the following set of overloads is provided:
Ret
|
Vec
|
|---|---|
| mask_int8 | int8 |
| mask_int8 | uint8 |
| mask_int16 | int16 |
| mask_int16 | uint16 |
| mask_int32 | int32 |
| mask_int32 | uint32 |
| mask_int64 | int64 |
| mask_int64 | uint64 |
| mask_float32 | float32 |
| mask_float64 | float64 |
The return type is a vector expression.
Parameters[edit]
| a | - | vector to convert |
Return value[edit]
A vector expression evaluating to vector converted to its mask equivalent.
Equivalent operation[edit]
r0 = a0 != 0 ? 1 : 0 ... rN = aN != 0 ? 1 : 0
See also[edit]
| compares vector elements for equality (function template) |