Integer Operations

Overview

Modules
	AIE ID
	Digital Pre-Distortion
	Peak Cancellation Crest Factor Reduction (PC-CFR)

Functions
int	scalar_abs (int a)

Scalar select intrinsics
char	select (bool, char, char)
	Scalar select.
unsigned char	select (bool, unsigned char, unsigned char)
	Scalar select.
short	select (bool, short, short)
	Scalar select.
int	select (bool, int, int)
	Scalar select.
long	select (bool, long, long)
	Scalar select.
void *	select (bool, void *, void *)
	Scalar select.

Leading bit count
unsigned	clb (int a)
	Count number of leading consecutive bits. If a is positive, count leading 0's, otherwise count leading 1's.
unsigned	clb (unsigned a)
	Count number of leading consecutive 0's.
unsigned	clb (long long a)
	Count number of leading consecutive bits. If a is positive, count leading 0's, otherwise count leading 1's.
unsigned	clb (unsigned long long a)
	Count number of leading consecutive 0's.

Inserting nops in the micro-code
void	nop (unsigned delay)
	Insert pipeline bubbles in the micro-code.
void	nop ()
	Insert pipeline bubbles in the micro-code.

Bit extraction/update
unsigned	bitset (unsigned src, unsigned val, unsigned idx)
unsigned	bitget (unsigned src, unsigned idx)

Function Documentation

unsigned bitget	(	unsigned	src,
		unsigned	idx
	)

Returns the value (0 or 1) of the bit at idx in src.

return (src >> idx) & 1

Parameters

src	The value in which a bit is to be get.
idx	The index of the bit to get. This must be a compile time constant.

unsigned bitset	(	unsigned	src,
		unsigned	val,
		unsigned	idx
	)

Returns the updated src with the bit at idx set to val.

mask = ~(1 << idx)
newbit = ((val) ? 1 : 0) << idx
src = (src & mask) | newbit

Parameters

src	The value in which a bit is to be set.
val	The value to set the bit to. Treated as a 1 or 0 to be shifted into the correct position.
idx	The index of the bit to set. This must be a compile time constant.

unsigned clb

(

int

a

)

Count number of leading consecutive bits. If a is positive, count leading 0's, otherwise count leading 1's.

unsigned clb

(

unsigned

a

)

Count number of leading consecutive 0's.

unsigned clb

(

long long

a

)

Count number of leading consecutive bits. If a is positive, count leading 0's, otherwise count leading 1's.

unsigned clb

(

unsigned long long

a

)

Count number of leading consecutive 0's.

void nop

(

unsigned

delay

)

Insert pipeline bubbles in the micro-code.

Parameters

delay

Number of cycles between two instructions. (Default = 2)

Note

Delay must be a compile time constant.

Subsequent nops will be collapsed into a single nop (only the maximum value will be seen).

void nop

(

)

Insert pipeline bubbles in the micro-code.

Parameters

delay

Number of cycles between two instructions. (Default = 2)

Note

Delay must be a compile time constant.

Subsequent nops will be collapsed into a single nop (only the maximum value will be seen).

int scalar_abs

(

int

a

)

char select	(	bool	,
		char	,
		char
	)

Scalar select.

These instructions are equivalent to the ternary ?: operator and should normally be inferred by the compiler automatically.

unsigned long select	(	bool	,
		unsigned	char,
		unsigned	char
	)

Scalar select.

These instructions are equivalent to the ternary ?: operator and should normally be inferred by the compiler automatically.

short select	(	bool	,
		short	,
		short
	)

Scalar select.

These instructions are equivalent to the ternary ?: operator and should normally be inferred by the compiler automatically.

int select	(	bool	,
		int	,
		int
	)

Scalar select.

These instructions are equivalent to the ternary ?: operator and should normally be inferred by the compiler automatically.

long select	(	bool	,
		long	,
		long
	)

Scalar select.

These instructions are equivalent to the ternary ?: operator and should normally be inferred by the compiler automatically.

void* select	(	bool	,
		void *	,
		void *
	)

Scalar select.

These instructions are equivalent to the ternary ?: operator and should normally be inferred by the compiler automatically.