3ac0fc7c90
speeds up decoding of a 64kbps test file by 2MHz on h300 Change-Id: I437d05278fe1c495715cf0e3477f9960d1df9d3a
181 lines
7.5 KiB
C
181 lines
7.5 KiB
C
/* Copyright (C) 2007-2009 Xiph.Org Foundation
|
|
Copyright (C) 2003-2008 Jean-Marc Valin
|
|
Copyright (C) 2007-2008 CSIRO */
|
|
/**
|
|
@file fixed_generic.h
|
|
@brief Generic fixed-point operations
|
|
*/
|
|
/*
|
|
Redistribution and use in source and binary forms, with or without
|
|
modification, are permitted provided that the following conditions
|
|
are met:
|
|
|
|
- Redistributions of source code must retain the above copyright
|
|
notice, this list of conditions and the following disclaimer.
|
|
|
|
- Redistributions in binary form must reproduce the above copyright
|
|
notice, this list of conditions and the following disclaimer in the
|
|
documentation and/or other materials provided with the distribution.
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
|
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER
|
|
OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
|
|
EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
|
|
PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
|
|
PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
|
|
LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
|
|
NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
|
|
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
*/
|
|
|
|
#ifndef FIXED_GENERIC_H
|
|
#define FIXED_GENERIC_H
|
|
|
|
/** Multiply a 16-bit signed value by a 16-bit unsigned value. The result is a 32-bit signed value */
|
|
#define MULT16_16SU(a,b) ((opus_val32)(opus_val16)(a)*(opus_val32)(opus_uint16)(b))
|
|
|
|
/** 16x32 multiplication, followed by a 16-bit shift right. Results fits in 32 bits */
|
|
#define MULT16_32_Q16(a,b) ADD32(MULT16_16((a),SHR((b),16)), SHR(MULT16_16SU((a),((b)&0x0000ffff)),16))
|
|
|
|
/** 16x32 multiplication, followed by a 16-bit shift right (round-to-nearest). Results fits in 32 bits */
|
|
#define MULT16_32_P16(a,b) ADD32(MULT16_16((a),SHR((b),16)), PSHR(MULT16_16((a),((b)&0x0000ffff)),16))
|
|
|
|
#if defined(CPU_COLDFIRE)
|
|
static inline int32_t MULT16_32_Q15(int32_t a, int32_t b)
|
|
{
|
|
int32_t r;
|
|
asm volatile ("mac.l %[a], %[b], %%acc0;"
|
|
"movclr.l %%acc0, %[r];"
|
|
: [r] "=r" (r)
|
|
: [a] "r" (a<<16), [b] "r" (b)
|
|
: "cc");
|
|
return r;
|
|
}
|
|
|
|
#elif defined(CPU_ARM)
|
|
static inline int32_t MULT16_32_Q15(int32_t a, int32_t b)
|
|
{
|
|
int32_t lo, hi;
|
|
asm volatile("smull %[lo], %[hi], %[b], %[a] \n\t"
|
|
"mov %[lo], %[lo], lsr #15 \n\t"
|
|
"orr %[hi], %[lo], %[hi], lsl #17 \n\t"
|
|
: [lo] "=&r" (lo), [hi] "=&r" (hi)
|
|
: [a] "r" (a), [b] "r" (b) );
|
|
return(hi);
|
|
}
|
|
|
|
#else
|
|
/** 16x32 multiplication, followed by a 15-bit shift right. Results fits in 32 bits */
|
|
#define MULT16_32_Q15(a,b) ADD32(SHL(MULT16_16((a),SHR((b),16)),1), SHR(MULT16_16SU((a),((b)&0x0000ffff)),15))
|
|
#endif
|
|
|
|
#if defined(CPU_COLDFIRE)
|
|
static inline int32_t MULT32_32_Q31(int32_t a, int32_t b)
|
|
{
|
|
int32_t r;
|
|
asm volatile ("mac.l %[a], %[b], %%acc0;"
|
|
"movclr.l %%acc0, %[r];"
|
|
: [r] "=r" (r)
|
|
: [a] "r" (a), [b] "r" (b)
|
|
: "cc");
|
|
return r;
|
|
}
|
|
|
|
#elif defined(CPU_ARM)
|
|
static inline int32_t MULT32_32_Q31(int32_t a, int32_t b)
|
|
{
|
|
int32_t lo, hi;
|
|
asm volatile("smull %[lo], %[hi], %[a], %[b] \n\t"
|
|
"mov %[lo], %[lo], lsr #31 \n\t"
|
|
"orr %[hi], %[lo], %[hi], lsl #1 \n\t"
|
|
: [lo] "=&r" (lo), [hi] "=&r" (hi)
|
|
: [a] "r" (a), [b] "r" (b) );
|
|
return(hi);
|
|
}
|
|
|
|
#else
|
|
/** 32x32 multiplication, followed by a 31-bit shift right. Results fits in 32 bits */
|
|
//#define MULT32_32_Q31(a,b) ADD32(ADD32(SHL(MULT16_16(SHR((a),16),SHR((b),16)),1), SHR(MULT16_16SU(SHR((a),16),((b)&0x0000ffff)),15)), SHR(MULT16_16SU(SHR((b),16),((a)&0x0000ffff)),15))
|
|
#define MULT32_32_Q31(a,b) (opus_val32)((((int64_t)(a)) * ((int64_t)(b)))>>31)
|
|
#endif
|
|
/** Compile-time conversion of float constant to 16-bit value */
|
|
#define QCONST16(x,bits) ((opus_val16)(.5+(x)*(((opus_val32)1)<<(bits))))
|
|
|
|
/** Compile-time conversion of float constant to 32-bit value */
|
|
#define QCONST32(x,bits) ((opus_val32)(.5+(x)*(((opus_val32)1)<<(bits))))
|
|
|
|
/** Negate a 16-bit value */
|
|
#define NEG16(x) (-(x))
|
|
/** Negate a 32-bit value */
|
|
#define NEG32(x) (-(x))
|
|
|
|
/** Change a 32-bit value into a 16-bit value. The value is assumed to fit in 16-bit, otherwise the result is undefined */
|
|
#define EXTRACT16(x) ((opus_val16)(x))
|
|
/** Change a 16-bit value into a 32-bit value */
|
|
#define EXTEND32(x) ((opus_val32)(x))
|
|
|
|
/** Arithmetic shift-right of a 16-bit value */
|
|
#define SHR16(a,shift) ((a) >> (shift))
|
|
/** Arithmetic shift-left of a 16-bit value */
|
|
#define SHL16(a,shift) ((opus_int16)((opus_uint16)(a)<<(shift)))
|
|
/** Arithmetic shift-right of a 32-bit value */
|
|
#define SHR32(a,shift) ((a) >> (shift))
|
|
/** Arithmetic shift-left of a 32-bit value */
|
|
#define SHL32(a,shift) ((opus_int32)((opus_uint32)(a)<<(shift)))
|
|
|
|
/** 32-bit arithmetic shift right with rounding-to-nearest instead of rounding down */
|
|
#define PSHR32(a,shift) (SHR32((a)+((EXTEND32(1)<<((shift))>>1)),shift))
|
|
/** 32-bit arithmetic shift right where the argument can be negative */
|
|
#define VSHR32(a, shift) (((shift)>0) ? SHR32(a, shift) : SHL32(a, -(shift)))
|
|
|
|
/** "RAW" macros, should not be used outside of this header file */
|
|
#define SHR(a,shift) ((a) >> (shift))
|
|
#define SHL(a,shift) SHL32(a,shift)
|
|
#define PSHR(a,shift) (SHR((a)+((EXTEND32(1)<<((shift))>>1)),shift))
|
|
#define SATURATE(x,a) (((x)>(a) ? (a) : (x)<-(a) ? -(a) : (x)))
|
|
|
|
/** Shift by a and round-to-neareast 32-bit value. Result is a 16-bit value */
|
|
#define ROUND16(x,a) (EXTRACT16(PSHR32((x),(a))))
|
|
/** Divide by two */
|
|
#define HALF16(x) (SHR16(x,1))
|
|
#define HALF32(x) (SHR32(x,1))
|
|
|
|
/** Add two 16-bit values */
|
|
#define ADD16(a,b) ((opus_val16)((opus_val16)(a)+(opus_val16)(b)))
|
|
/** Subtract two 16-bit values */
|
|
#define SUB16(a,b) ((opus_val16)(a)-(opus_val16)(b))
|
|
/** Add two 32-bit values */
|
|
#define ADD32(a,b) ((opus_val32)(a)+(opus_val32)(b))
|
|
/** Subtract two 32-bit values */
|
|
#define SUB32(a,b) ((opus_val32)(a)-(opus_val32)(b))
|
|
|
|
/** 16x16 multiplication where the result fits in 16 bits */
|
|
#define MULT16_16_16(a,b) ((((opus_val16)(a))*((opus_val16)(b))))
|
|
|
|
/* (opus_val32)(opus_val16) gives TI compiler a hint that it's 16x16->32 multiply */
|
|
/** 16x16 multiplication where the result fits in 32 bits */
|
|
#define MULT16_16(a,b) (((opus_val32)(opus_val16)(a))*((opus_val32)(opus_val16)(b)))
|
|
|
|
/** 16x16 multiply-add where the result fits in 32 bits */
|
|
#define MAC16_16(c,a,b) (ADD32((c),MULT16_16((a),(b))))
|
|
/** 16x32 multiply-add, followed by a 15-bit shift right. Results fits in 32 bits */
|
|
#define MAC16_32_Q15(c,a,b) ADD32(c,ADD32(MULT16_16((a),SHR((b),15)), SHR(MULT16_16((a),((b)&0x00007fff)),15)))
|
|
|
|
#define MULT16_16_Q11_32(a,b) (SHR(MULT16_16((a),(b)),11))
|
|
#define MULT16_16_Q13(a,b) (SHR(MULT16_16((a),(b)),13))
|
|
#define MULT16_16_Q14(a,b) (SHR(MULT16_16((a),(b)),14))
|
|
#define MULT16_16_Q15(a,b) (SHR(MULT16_16((a),(b)),15))
|
|
|
|
#define MULT16_16_P13(a,b) (SHR(ADD32(4096,MULT16_16((a),(b))),13))
|
|
#define MULT16_16_P14(a,b) (SHR(ADD32(8192,MULT16_16((a),(b))),14))
|
|
#define MULT16_16_P15(a,b) (SHR(ADD32(16384,MULT16_16((a),(b))),15))
|
|
|
|
/** Divide a 32-bit value by a 16-bit value. Result fits in 16 bits */
|
|
#define DIV32_16(a,b) ((opus_val16)(((opus_val32)(a))/((opus_val16)(b))))
|
|
|
|
/** Divide a 32-bit value by a 32-bit value. Result fits in 32 bits */
|
|
#define DIV32(a,b) (((opus_val32)(a))/((opus_val32)(b)))
|
|
|
|
#endif
|