Intrinsics

Hardware Intrinsics for the SH4 fast-math instructions More...

Files
file	fmath.h
	Inline functions for the DC's special math instructions.
file	fmath_base.h
	Base definitions for the DC's special math instructions.

Macros
#define	__FMINLINE   static
#define	F_PI   3.1415926f
	PI constant (if you don't want full math.h)

Functions
__FMINLINE float	fipr (float x, float y, float z, float w, float a, float b, float c, float d)
	Floating point inner product.
__FMINLINE float	fipr_magnitude_sqr (float x, float y, float z, float w)
	Floating point inner product w/self (square of vector magnitude)
__FMINLINE float	fsin (float r)
	Floating point sine.
__FMINLINE float	fcos (float r)
	Floating point cosine.
__FMINLINE float	ftan (float r)
	Floating point tangent.
__FMINLINE float	fisin (int d)
	Integer sine.
__FMINLINE float	ficos (int d)
	Integer cosine.
__FMINLINE float	fitan (int d)
	Integer tangent.
__FMINLINE float	fsqrt (float f)
	Floating point square root.
__FMINLINE float	frsqrt (float f)
__FMINLINE void	fsincos (float f, float *s, float *c)
	Calculate the sine and cosine of a value in degrees.
__FMINLINE void	fsincosr (float f, float *s, float *c)
	Calculate the sine and cosine of a value in radians.
__FMINLINE uint32	pvr_pack_bump (float h, float t, float q)
	Calculate the offset color value for a set of bumpmap parameters.

Detailed Description

Hardware Intrinsics for the SH4 fast-math instructions

Macro Definition Documentation

__FMINLINE

#define __FMINLINE   static

F_PI

#define F_PI   3.1415926f

PI constant (if you don't want full math.h)

Referenced by pvr_pack_bump().

Function Documentation

fcos()

__FMINLINE float fcos

(

float

r

)

Floating point cosine.

Parameters

r	a floating point number between 0 and 2*PI

Returns

cos(r), where r is [0..2*PI]

ficos()

__FMINLINE float ficos

(

int

d

)

Integer cosine.

Parameters

d	an integer between 0 and 65535

Returns

cos(d), where d is [0..65535]

fipr()

__FMINLINE float fipr	(	float	x,
		float	y,
		float	z,
		float	w,
		float	a,
		float	b,
		float	c,
		float	d )

Floating point inner product.

Returns

v1 dot v2 (inner product)

fipr_magnitude_sqr()

__FMINLINE float fipr_magnitude_sqr	(	float	x,
		float	y,
		float	z,
		float	w )

Floating point inner product w/self (square of vector magnitude)

Returns

v1 dot v1 (square of magnitude)

fisin()

__FMINLINE float fisin

(

int

d

)

Integer sine.

Parameters

d	an integer between 0 and 65535

Returns

sin(d), where d is [0..65535]

fitan()

__FMINLINE float fitan

(

int

d

)

Integer tangent.

Parameters

d	an integer between 0 and 65535

Returns

tan(d), where d is [0..65535]

frsqrt()

__FMINLINE float frsqrt

(

float

f

)

Returns

1.0f / sqrt(f)

fsin()

__FMINLINE float fsin

(

float

r

)

Floating point sine.

Parameters

r	a floating point number between 0 and 2*PI

Returns

sin(r), where r is [0..2*PI]

fsincos()

__FMINLINE void fsincos	(	float	f,
		float *	s,
		float *	c )

Calculate the sine and cosine of a value in degrees.

This function uses the fsca instruction to calculate an approximation of the sine and cosine of the input value.

Parameters

f	The value to calculate the sine and cosine of.
s	Storage for the returned sine value.
c	Storage for the returned cosine value.

fsincosr()

__FMINLINE void fsincosr	(	float	f,
		float *	s,
		float *	c )

Calculate the sine and cosine of a value in radians.

This function uses the fsca instruction to calculate an approximation of the sine and cosine of the input value.

Parameters

f	The value to calculate the sine and cosine of.
s	Storage for the returned sine value.
c	Storage for the returned cosine value.

fsqrt()

__FMINLINE float fsqrt

(

float

f

)

Floating point square root.

Returns

sqrt(f)

ftan()

__FMINLINE float ftan

(

float

r

)

Floating point tangent.

Parameters

r	a floating point number between 0 and 2*PI

Returns

tan(r), where r is [0..2*PI]

pvr_pack_bump()

__FMINLINE uint32 pvr_pack_bump	(	float	h,
		float	t,
		float	q )

Calculate the offset color value for a set of bumpmap parameters.

This function calculates the value to be placed into the oargb value for the use of bumpmapping on a polygon. The angles specified should be expressed in radians and within the limits specified for the individual parameter.

Parameters

h	Weighting value in the range [0, 1] for how defined the bumpiness of the surface should be.
t	Spherical elevation angle in the range [0, pi/2] between the surface and the lighting source. A value of pi/2 implies that the light is directly overhead.
q	Spherical rotation angle in the range [0, 2*pi] between the surface and the lighting source.

Returns

32-bit packed value to be used as an offset color on the surface to be bump mapped.

Note

For more information about how bumpmapping on the PVR works, refer to US Patent 6,819,319, which describes the algorithm implemented in the hardware (specifically look at Figures 2 and 3, along with the description in the Detailed Description section).

Thanks to Fredrik Ehnbom for figuring this stuff out and posting it to the mailing list back in 2005!

References F_PI.