
XPeRiPheRaLMoDuLes8(CaLC):CaLCuLationMoDuLe
s1C33L17teChniCaLManuaL
ePson
X-1-1
X
CALC
X.1 Calculation Module
X.1.1overview
The S1C33L17 is equipped with a hardware calculation accelerator to reduce multimedia processing loads. The
calculation module serves as the C language interface required to use the hardware calculation accelerator. The
hardware calculation accelerator allows high-speed processing of product-sum calculations, matrix calculations,
affine transformations, and butterfly calculations.
This module reduces processing loads for image processing tasks such as coordinate transformations and color
conversion, as well as audio and video decoder processing tasks.
X.1.2FunctionsProvided
This module provides the following functions.
nameofaPi
Function
CalcOpen
This API enables use of the calculation module.
CalcClose
This API terminates use of the calculation module.
CalcMac4
This API performs 4-element product-sum calculations.
CalcMac8
This API performs 8-element product-sum calculations.
CalcMac16
This API performs 16-element product-sum calculations.
CalcMatrix2
This API performs second-order matrix multiplication ((2 x 2) x (2 x 1)).
CalcMatrix3
This API performs third-order matrix multiplication ((3 x 3) x (3 x 1)).
CalcMatrix4
This API performs fourth-order matrix multiplication ((4 x 4) x (4 x 1)).
CalcAffine2
This API performs second-order affine transformation ((2 x 2) x (2 x 1) + (2 x 1)).
CalcAffine3
This API performs third-order affine transformations ((3 x 3) x (3 x 1) + (3 x 1)).
CalcButterfly
This API performs butterfly calculations.
The module can handle the following numerical formats.
nameofmode
numericalformat
note
CALC_UINT32
32-bit unsigned integer (unsigned long)
Without saturation processing
CALC_SINT32
32-bit signed integer (signed long)
Without saturation processing
CALC_FIXED16_Q13
Q13 notation 16-bit signed fixed point
With saturation processing
CALC_FIXED16_Q14
Q14 notation 16-bit signed fixed point
With saturation processing
This module can be activated for use by including the header file described at the end of this manual.
Note that one of the multiplicands must be arranged in DSTRAM due to the restrictions imposed by the hardware
calculation accelerator.