|
|
|
| Syntax C/C++ | #include <VFmath.h>
int VF_cos( fVector Y, fVector X, ui size );
int VFx_cos( fVector Y, fVector X, ui size, float A, float B, float C );
int VFr_cos( fVector Y, fVector X, ui size );
int VFrx_cos( fVector Y, fVector X, ui size, float A, float B, float C ); |
| C++ VecObj | #include <OptiVec.h>
int vector<T>::cos( const vector<T>& X );
int vector<T>::x_cos( const vector<T>& X, const T& A, const T& B, const T& C );
int vector<T>::r_cos( const vector<T>& X );
int vector<T>::rx_cos( const vector<T>& X, const T& A, const T& B, const T& C ); |
| Pascal/Delphi | uses VFmath;
function VF_cos( Y, X:fVector; size:UIntSize ): IntBool;
function VFx_cos( Y, X:fVector; size:UIntSize; A, B, C:Single ): IntBool;
function VFr_cos( Y, X:fVector; size:UIntSize ): IntBool;
function VFrx_cos( Y, X:fVector; size:UIntSize; A, B, C:Single ): IntBool; |
|
| CUDA function C/C++ | #include <cudaVFmath.h>
int cudaVF_cos( fVector d_Y, fVector d_X, ui size );
int cudaVFx_cos( fVector d_Y, fVector d_X, ui size, float A, float B, float C );
int cusdVFx_cos( fVector d_Y, fVector d_X, ui size, float *d_A, float *d_B, float *d_C );
int VFcu_cos( fVector h_Y, fVector h_X, ui size );
int VFxcu_cos( fVector h_Y, fVector h_X, ui size, float A, float B, float C );
|
| CUDA function Pascal/Delphi | uses VFmath;
function cudaVF_cos( d_Y, d_X:fVector; size:UIntSize ): IntBool;
function cudaVFx_cos( d_Y, d_X:fVector; size:UIntSize; A, B, C:Single ): IntBool;
function cusdVFx_cos( d_Y, d_X:fVector; size:UIntSize; d_A, d_B, d_C:PSingle ): IntBool;
function VFcu_cos( h_Y, h_X:fVector; size:UIntSize ): IntBool;
function VFxcu_cos( h_Y, h_X:fVector; size:UIntSize; A, B, C:Single ): IntBool;
|
|
| Description | simple versions: Yi = cos ( Xi )
expanded versions: Yi = C * cos ( A*Xi + B )
For large values of Xi, round-off error becomes very large; if the Xi values are representable as rational multiples of p, it is better to use VF_cosrpi than VF_cos.
If, on the other hand, one can be sure that all Xi are within a reasonable range one can employ the faster reduced-range versions with the prefixes VFr_ and VFrx_. The range requirements for the reduced-range versions are:
64-bit: |Xi| < 232 (roughly 4.2*109)
32-bit: |Xi| ≤ 2p).
The reduced-range versions are not available for CUDA. |
|
| Error handling | Precision errors in the real-value functions lead to a default result of 1.0 (as if the input were 0.0) and to a non-zero return value, but are otherwise ignored; _matherr is not called.
OVERFLOW errors can only occur in the complex versions and lead to a result of ±HUGE_VAL. |
|
| Return value | FALSE (0), if no error occurred, otherwise TRUE (non-zero). |
|
|