7186a2dd86
It in essence refactors the code for both the interpolation filtering and the convolution. This change includes the moving of all the files as well as the changing of the code from vp9_ prefix to vpx_ prefix accordingly, for underneath architectures: (1) x86; (2) arm/neon; and (3) mips/msa. The work on mips/drsp2 will be done in a separate change list. Change-Id: Ic3ce7fb7f81210db7628b373c73553db68793c46
159 lines
6.4 KiB
C
159 lines
6.4 KiB
C
/*
|
|
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
|
|
*
|
|
* Use of this source code is governed by a BSD-style license
|
|
* that can be found in the LICENSE file in the root of the source
|
|
* tree. An additional intellectual property rights grant can be found
|
|
* in the file PATENTS. All contributing project authors may
|
|
* be found in the AUTHORS file in the root of the source tree.
|
|
*/
|
|
|
|
#ifndef VP9_COMMON_VP9_IDCT_H_
|
|
#define VP9_COMMON_VP9_IDCT_H_
|
|
|
|
#include <assert.h>
|
|
|
|
#include "./vpx_config.h"
|
|
#include "vpx_dsp/txfm_common.h"
|
|
#if CONFIG_VP9_HIGHBITDEPTH
|
|
#include "vpx_dsp/vpx_dsp_common.h"
|
|
#endif // CONFIG_VP9_HIGHBITDEPTH
|
|
#include "vpx_ports/mem.h"
|
|
#include "vp9/common/vp9_common.h"
|
|
#include "vp9/common/vp9_enums.h"
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
static INLINE tran_low_t check_range(tran_high_t input) {
|
|
#if CONFIG_COEFFICIENT_RANGE_CHECKING
|
|
// For valid VP9 input streams, intermediate stage coefficients should always
|
|
// stay within the range of a signed 16 bit integer. Coefficients can go out
|
|
// of this range for invalid/corrupt VP9 streams. However, strictly checking
|
|
// this range for every intermediate coefficient can burdensome for a decoder,
|
|
// therefore the following assertion is only enabled when configured with
|
|
// --enable-coefficient-range-checking.
|
|
assert(INT16_MIN <= input);
|
|
assert(input <= INT16_MAX);
|
|
#endif // CONFIG_COEFFICIENT_RANGE_CHECKING
|
|
return (tran_low_t)input;
|
|
}
|
|
|
|
static INLINE tran_low_t dct_const_round_shift(tran_high_t input) {
|
|
tran_high_t rv = ROUND_POWER_OF_TWO(input, DCT_CONST_BITS);
|
|
return check_range(rv);
|
|
}
|
|
|
|
#if CONFIG_VP9_HIGHBITDEPTH
|
|
static INLINE tran_low_t highbd_check_range(tran_high_t input,
|
|
int bd) {
|
|
#if CONFIG_COEFFICIENT_RANGE_CHECKING
|
|
// For valid highbitdepth VP9 streams, intermediate stage coefficients will
|
|
// stay within the ranges:
|
|
// - 8 bit: signed 16 bit integer
|
|
// - 10 bit: signed 18 bit integer
|
|
// - 12 bit: signed 20 bit integer
|
|
const int32_t int_max = (1 << (7 + bd)) - 1;
|
|
const int32_t int_min = -int_max - 1;
|
|
assert(int_min <= input);
|
|
assert(input <= int_max);
|
|
(void) int_min;
|
|
#endif // CONFIG_COEFFICIENT_RANGE_CHECKING
|
|
(void) bd;
|
|
return (tran_low_t)input;
|
|
}
|
|
|
|
static INLINE tran_low_t highbd_dct_const_round_shift(tran_high_t input,
|
|
int bd) {
|
|
tran_high_t rv = ROUND_POWER_OF_TWO(input, DCT_CONST_BITS);
|
|
return highbd_check_range(rv, bd);
|
|
}
|
|
#endif // CONFIG_VP9_HIGHBITDEPTH
|
|
|
|
typedef void (*transform_1d)(const tran_low_t*, tran_low_t*);
|
|
|
|
typedef struct {
|
|
transform_1d cols, rows; // vertical and horizontal
|
|
} transform_2d;
|
|
|
|
#if CONFIG_VP9_HIGHBITDEPTH
|
|
typedef void (*highbd_transform_1d)(const tran_low_t*, tran_low_t*, int bd);
|
|
|
|
typedef struct {
|
|
highbd_transform_1d cols, rows; // vertical and horizontal
|
|
} highbd_transform_2d;
|
|
#endif // CONFIG_VP9_HIGHBITDEPTH
|
|
|
|
#if CONFIG_EMULATE_HARDWARE
|
|
// When CONFIG_EMULATE_HARDWARE is 1 the transform performs a
|
|
// non-normative method to handle overflows. A stream that causes
|
|
// overflows in the inverse transform is considered invalid in VP9,
|
|
// and a hardware implementer is free to choose any reasonable
|
|
// method to handle overflows. However to aid in hardware
|
|
// verification they can use a specific implementation of the
|
|
// WRAPLOW() macro below that is identical to their intended
|
|
// hardware implementation (and also use configure options to trigger
|
|
// the C-implementation of the transform).
|
|
//
|
|
// The particular WRAPLOW implementation below performs strict
|
|
// overflow wrapping to match common hardware implementations.
|
|
// bd of 8 uses trans_low with 16bits, need to remove 16bits
|
|
// bd of 10 uses trans_low with 18bits, need to remove 14bits
|
|
// bd of 12 uses trans_low with 20bits, need to remove 12bits
|
|
// bd of x uses trans_low with 8+x bits, need to remove 24-x bits
|
|
#define WRAPLOW(x, bd) ((((int32_t)(x)) << (24 - bd)) >> (24 - bd))
|
|
#else
|
|
#define WRAPLOW(x, bd) ((int32_t)(x))
|
|
#endif // CONFIG_EMULATE_HARDWARE
|
|
|
|
void vp9_iwht4x4_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob);
|
|
void vp9_idct4x4_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob);
|
|
void vp9_idct8x8_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob);
|
|
void vp9_idct16x16_add(const tran_low_t *input, uint8_t *dest, int stride, int
|
|
eob);
|
|
void vp9_idct32x32_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob);
|
|
|
|
void vp9_iht4x4_add(TX_TYPE tx_type, const tran_low_t *input, uint8_t *dest,
|
|
int stride, int eob);
|
|
void vp9_iht8x8_add(TX_TYPE tx_type, const tran_low_t *input, uint8_t *dest,
|
|
int stride, int eob);
|
|
void vp9_iht16x16_add(TX_TYPE tx_type, const tran_low_t *input, uint8_t *dest,
|
|
int stride, int eob);
|
|
|
|
#if CONFIG_VP9_HIGHBITDEPTH
|
|
void vp9_highbd_idct4(const tran_low_t *input, tran_low_t *output, int bd);
|
|
void vp9_highbd_idct8(const tran_low_t *input, tran_low_t *output, int bd);
|
|
void vp9_highbd_idct16(const tran_low_t *input, tran_low_t *output, int bd);
|
|
void vp9_highbd_iwht4x4_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob, int bd);
|
|
void vp9_highbd_idct4x4_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob, int bd);
|
|
void vp9_highbd_idct8x8_add(const tran_low_t *input, uint8_t *dest, int stride,
|
|
int eob, int bd);
|
|
void vp9_highbd_idct16x16_add(const tran_low_t *input, uint8_t *dest,
|
|
int stride, int eob, int bd);
|
|
void vp9_highbd_idct32x32_add(const tran_low_t *input, uint8_t *dest,
|
|
int stride, int eob, int bd);
|
|
void vp9_highbd_iht4x4_add(TX_TYPE tx_type, const tran_low_t *input,
|
|
uint8_t *dest, int stride, int eob, int bd);
|
|
void vp9_highbd_iht8x8_add(TX_TYPE tx_type, const tran_low_t *input,
|
|
uint8_t *dest, int stride, int eob, int bd);
|
|
void vp9_highbd_iht16x16_add(TX_TYPE tx_type, const tran_low_t *input,
|
|
uint8_t *dest, int stride, int eob, int bd);
|
|
static INLINE uint16_t highbd_clip_pixel_add(uint16_t dest, tran_high_t trans,
|
|
int bd) {
|
|
trans = WRAPLOW(trans, bd);
|
|
return clip_pixel_highbd(WRAPLOW(dest + trans, bd), bd);
|
|
}
|
|
#endif // CONFIG_VP9_HIGHBITDEPTH
|
|
#ifdef __cplusplus
|
|
} // extern "C"
|
|
#endif
|
|
|
|
#endif // VP9_COMMON_VP9_IDCT_H_
|