Data transfer¶

Data transfer instructions¶

template<class T1, class T2 = T1> inline simd_return_type<T1, T2> set_simd(const T1 &value)¶

Returns a batch with all values initialized to value.

Parameters: value – the scalar used to initialize the batch.
Returns: the batch wrapping the highest available instruction set.

template<class T1, class T2 = T1> inline simd_return_type<T1, T2> load_aligned(const T1 *src)¶

Loads the memory array pointed to by src into a batch and returns it.

src is required to be aligned.

Parameters: src – the pointer to the memory array to load.
Returns: the batch wrapping the highest available instruction set.

template<class T1, class T2 = T1> inline void load_aligned(const T1 *src, simd_type<T2> &dst)¶

Loads the memory array pointed to by src into the batch dst.

src is required to be aligned.

Parameters

src – the pointer to the memory array to load.
dst – the destination batch.

template<class T1, class T2> inline simd_return_type<T1, T2> load_aligned(const T1 *real_src, const T1 *imag_src)¶

Loads the memory arrays pointed to by real_src and imag_src into a batch of complex numbers and returns it.

real_src and imag_src are required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.

Returns

the batch of complex wrapping the highest available instruction set.

template<class T1, class T2> inline void load_aligned(const T1 *real_src, const T1 *imag_src, simd_type<T2> &dst)¶

Loads the memory arrays pointed to by real_src and imag_src into the batch dst.

real_src and imag_src are required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.
dst – the destination batch.

template<class T1, class T2 = T1> inline simd_return_type<T1, T2> load_unaligned(const T1 *src)¶

Loads the memory array pointed to by src into a batch and returns it.

src is not required to be aligned.

Parameters: src – the pointer to the memory array to load.
Returns: the batch wrapping the highest available instruction set.

template<class T1, class T2 = T1> inline void load_unaligned(const T1 *src, simd_type<T2> &dst)¶

Loads the memory array pointed to by src into the batch dst.

src is not required to be aligned.

Parameters

src – the pointer to the memory array to load.
dst – the destination batch.

template<class T1, class T2> inline simd_return_type<T1, T2> load_unaligned(const T1 *real_src, const T1 *imag_src)¶

Loads the memory arrays pointed to by real_src and imag_src into a batch of complex numbers and returns it.

real_src and imag_src are not required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.

Returns

the batch of complex wrapping the highest available instruction set.

template<class T1, class T2> inline void load_unaligned(const T1 *real_src, const T1 *imag_src, simd_type<T2> &dst)¶

Loads the memory arrays pointed to by real_src and imag_src into the batch dst.

real_src and imag_src are not required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.
dst – the destination batch.

template<class T1, class T2 = T1> inline void store_aligned(T1 *dst, const simd_type<T2> &src)¶

Stores the batch src into the memory array pointed to by dst.

dst is required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the batch to store.

template<class T1, class T2 = T1> inline void store_aligned(T1 *dst, const simd_bool_type<T2> &src)¶

Stores the boolean batch src into the memory array pointed to by dst.

dst is required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the boolean batch to store.

template<class T1, class T2 = T1> inline void store_unaligned(T1 *dst, const simd_type<T2> &src)¶

Stores the batch src into the memory array pointed to by dst.

dst is not required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the batch to store.

template<class T1, class T2 = T1> inline void store_unaligned(T1 *dst, const simd_bool_type<T2> &src)¶

Stores the boolean batch src into the memory array pointed to by dst.

dst is not required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the boolean batch to store.

template<class T1, class T2> inline void store_aligned(T1 *real_dst, T1 *imag_dst, const simd_type<T2> &src)¶

Stores the batch of complex numbers src into the memory arrays pointed to by real_dst and imag_dst.

real_dst and imag_dst are required to be aligned.

Parameters

real_dst – the pointer to the memory array of the real part.
imag_dst – the pointer to the memory array of the imaginary part.
src – the batch to store.

template<class T1, class T2> inline void store_unaligned(T1 *real_dst, T1 *imag_dst, const simd_type<T2> &src)¶

Stores the batch of complex numbers src into the memory arrays pointed to by real_dst and imag_dst.

real_dst and imag_dst are not required to be aligned.

Parameters

real_dst – the pointer to the memory array of the real part.
imag_dst – the pointer to the memory array of the imaginary part.
src – the batch to store.

Generic load and store¶

template<class T1, class T2 = T1> inline simd_return_type<T1, T2> load_simd(const T1 *src, aligned_mode)¶

Loads the memory array pointed to by src into a batch and returns it.

src is required to be aligned.

Parameters: src – the pointer to the memory array to load.
Returns: the batch wrapping the highest available instruction set.

template<class T1, class T2 = T1> inline void load_simd(const T1 *src, simd_type<T2> &dst, aligned_mode)¶

Loads the memory array pointed to by src into the batch dst.

src is required to be aligned.

Parameters

src – the pointer to the memory array to load.
dst – the destination batch.

template<class T1, class T2> inline simd_return_type<T1, T2> load_simd(const T1 *real_src, const T1 *imag_src, aligned_mode)¶

Loads the memory arrays pointed to by real_src and imag_src into a batch of complex numbers and returns it.

real_src and imag_src are required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.

Returns

the batch of complex wrapping the highest available instruction set.

template<class T1, class T2> inline void load_simd(const T1 *real_src, const T1 *imag_src, simd_type<T2> &dst, aligned_mode)¶

Loads the memory arrays pointed to by real_src and imag_src into the batch dst.

real_src and imag_src are required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.
dst – the destination batch.

template<class T1, class T2 = T1> inline simd_return_type<T1, T2> load_simd(const T1 *src, unaligned_mode)¶

Loads the memory array pointed to by src into a batch and returns it.

src is not required to be aligned.

Parameters: src – the pointer to the memory array to load.
Returns: the batch wrapping the highest available instruction set.

template<class T1, class T2 = T1> inline void load_simd(const T1 *src, simd_type<T2> &dst, unaligned_mode)¶

Loads the memory array pointed to by src into the batch dst.

src is not required to be aligned.

Parameters

src – the pointer to the memory array to load.
dst – the destination batch.

template<class T1, class T2> inline simd_return_type<T1, T2> load_simd(const T1 *real_src, const T1 *imag_src, unaligned_mode)¶

Loads the memory arrays pointed to by real_src and imag_src into a batch of complex numbers and returns it.

real_src and imag_src are not required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.

Returns

the batch of complex wrapping the highest available instruction set.

template<class T1, class T2> inline void load_simd(const T1 *real_src, const T1 *imag_src, simd_type<T2> &dst, unaligned_mode)¶

Loads the memory arrays pointed to by real_src and imag_src into the batch dst.

real_src and imag_src are not required to be aligned.

Parameters

real_src – the pointer to the memory array containing the real part.
imag_src – the pointer to the memory array containing the imaginary part.
dst – the destination batch.

template<class T1, class T2 = T1> inline void store_simd(T1 *dst, const simd_type<T2> &src, aligned_mode)¶

Stores the batch src into the memory array pointed to by dst.

dst is required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the batch to store.

template<class T1, class T2 = T1> inline void store_simd(T1 *dst, const simd_bool_type<T2> &src, aligned_mode)¶

Stores the boolean batch src into the memory array pointed to by dst.

dst is required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the boolean batch to store.

template<class T1, class T2 = T1> inline void store_simd(T1 *dst, const simd_type<T2> &src, unaligned_mode)¶

Stores the batch src into the memory array pointed to by dst.

dst is not required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the batch to store.

template<class T1, class T2 = T1> inline void store_simd(T1 *dst, const simd_bool_type<T2> &src, unaligned_mode)¶

Stores the boolean batch src into the memory array pointed to by dst.

dst is not required to be aligned.

Parameters

dst – the pointer to the memory array.
src – the boolean batch to store.

template<class T1, class T2> inline void store_simd(T1 *real_dst, T1 *imag_dst, const simd_type<T2> &src, aligned_mode)¶

Stores the batch of complex numbers src into the memory arrays pointed to by real_dst and imag_dst.

real_dst and imag_dst are required to be aligned.

Parameters

real_dst – the pointer to the memory array of the real part.
imag_dst – the pointer to the memory array of the imaginary part.
src – the batch to store.

template<class T1, class T2> inline void store_simd(T1 *real_dst, T1 *imag_dst, const simd_type<T2> &src, unaligned_mode)¶

Stores the batch of complex numbers src into the memory arrays pointed to by real_dst and imag_dst.

real_dst and imag_dst are not required to be aligned.

Parameters

real_dst – the pointer to the memory array of the real part.
imag_dst – the pointer to the memory array of the imaginary part.
src – the batch to store.