/src/libhevc/common/ihevc_chroma_itrans_recon_8x8.c

Source
/******************************************************************************
*
* Copyright (C) 2012 Ittiam Systems Pvt Ltd, Bangalore
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at:
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*
******************************************************************************/
/**
 *******************************************************************************
 * @file
 *  ihevc_chroma_itrans_recon_8x8.c
 *
 * @brief
 *  Contains function definitions for 8x8 inverse transform  and reconstruction
 * of chroma interleaved data.
 *
 * @author
 *  100470
 *
 * @par List of Functions:
 *  - ihevc_chroma_itrans_recon_8x8()
 *
 * @remarks
 *  None
 *
 *******************************************************************************
 */

#include <stdio.h>
#include <string.h>
#include "ihevc_typedefs.h"
#include "ihevc_macros.h"
#include "ihevc_platform_macros.h"
#include "ihevc_defs.h"
#include "ihevc_trans_tables.h"
#include "ihevc_chroma_itrans_recon.h"
#include "ihevc_func_selector.h"
#include "ihevc_trans_macros.h"

/* All the functions work one component(U or V) of interleaved data depending upon pointers passed to it */
/* Data visualization */
/* U V U V U V U V */
/* U V U V U V U V */
/* U V U V U V U V */
/* U V U V U V U V */
/* If the pointer points to first byte of above stream (U) , functions will operate on U component */
/* If the pointer points to second byte of above stream (V) , functions will operate on V component */

/**
 *******************************************************************************
 *
 * @brief
 *  This function performs Inverse transform  and reconstruction for 8x8
 * input block
 *
 * @par Description:
 *  Performs inverse transform and adds the prediction  data and clips output
 * to 8 bit
 *
 * @param[in] pi2_src
 *  Input 8x8 coefficients
 *
 * @param[in] pi2_tmp
 *  Temporary 8x8 buffer for storing inverse transform
 *  1st stage output
 *
 * @param[in] pu1_pred
 *  Prediction 8x8 block
 *
 * @param[out] pu1_dst
 *  Output 8x8 block
 *
 * @param[in] src_strd
 *  Input stride
 *
 * @param[in] pred_strd
 *  Prediction stride
 *
 * @param[in] dst_strd
 *  Output Stride
 *
 * @param[in] shift
 *  Output shift
 *
 * @param[in] zero_cols
 *  Zero columns in pi2_src
 *
 * @returns  Void
 *
 * @remarks
 *  None
 *
 *******************************************************************************
 */


void ihevc_chroma_itrans_recon_8x8(WORD16 *pi2_src,
                                   WORD16 *pi2_tmp,
                                   UWORD8 *pu1_pred,
                                   UWORD8 *pu1_dst,
                                   WORD32 src_strd,
                                   WORD32 pred_strd,
                                   WORD32 dst_strd,
                                   WORD32 zero_cols,
                                   WORD32 zero_rows)
{
    WORD32 j, k;
    WORD32 e[4], o[4];
    WORD32 ee[2], eo[2];
    WORD32 add;
    WORD32 shift;
    WORD16 *pi2_tmp_orig;
    WORD32 trans_size;
    WORD32 zero_rows_2nd_stage = zero_cols;
    WORD32 row_limit_2nd_stage;
    UNUSED(zero_rows);
    trans_size = TRANS_SIZE_8;

    pi2_tmp_orig = pi2_tmp;

    if((zero_cols & 0xF0) == 0xF0)
        row_limit_2nd_stage = 4;
    else
        row_limit_2nd_stage = TRANS_SIZE_8;

    /* Inverse Transform 1st stage */
    shift = IT_SHIFT_STAGE_1;
    add = 1 << (shift - 1);
    {
        /************************************************************************************************/
        /**********************************START - IT_RECON_8x8******************************************/
        /************************************************************************************************/

        for(j = 0; j < row_limit_2nd_stage; j++)
        {
            /* Checking for Zero Cols */
            if((zero_cols & 1) == 1)
            {
                memset(pi2_tmp, 0, trans_size * sizeof(WORD16));
            }
            else
            {
                /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
                for(k = 0; k < 4; k++)
                {
                    o[k] = g_ai2_ihevc_trans_8[1][k] * pi2_src[src_strd]
                                    + g_ai2_ihevc_trans_8[3][k]
                                                    * pi2_src[3 * src_strd]
                                    + g_ai2_ihevc_trans_8[5][k]
                                                    * pi2_src[5 * src_strd]
                                    + g_ai2_ihevc_trans_8[7][k]
                                                    * pi2_src[7 * src_strd];
                }

                eo[0] = g_ai2_ihevc_trans_8[2][0] * pi2_src[2 * src_strd]
                                + g_ai2_ihevc_trans_8[6][0] * pi2_src[6 * src_strd];
                eo[1] = g_ai2_ihevc_trans_8[2][1] * pi2_src[2 * src_strd]
                                + g_ai2_ihevc_trans_8[6][1] * pi2_src[6 * src_strd];
                ee[0] = g_ai2_ihevc_trans_8[0][0] * pi2_src[0]
                                + g_ai2_ihevc_trans_8[4][0] * pi2_src[4 * src_strd];
                ee[1] = g_ai2_ihevc_trans_8[0][1] * pi2_src[0]
                                + g_ai2_ihevc_trans_8[4][1] * pi2_src[4 * src_strd];

                /* Combining e and o terms at each hierarchy levels to calculate the final spatial domain vector */
                e[0] = ee[0] + eo[0];
                e[3] = ee[0] - eo[0];
                e[1] = ee[1] + eo[1];
                e[2] = ee[1] - eo[1];
                for(k = 0; k < 4; k++)
                {
                    pi2_tmp[k] =
                                    CLIP_S16(((e[k] + o[k] + add) >> shift));
                    pi2_tmp[k + 4] =
                                    CLIP_S16(((e[3 - k] - o[3 - k] + add) >> shift));
                }
            }
            pi2_src++;
            pi2_tmp += trans_size;
            zero_cols = zero_cols >> 1;
        }

        pi2_tmp = pi2_tmp_orig;

        /* Inverse Transform 2nd stage */
        shift = IT_SHIFT_STAGE_2;
        add = 1 << (shift - 1);

        if((zero_rows_2nd_stage & 0xF0) == 0xF0) /* First 4 rows of output of 1st stage are non-zero */
        {
            for(j = 0; j < trans_size; j++)
            {
                /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
                for(k = 0; k < 4; k++)
                {
                    o[k] = g_ai2_ihevc_trans_8[1][k] * pi2_tmp[trans_size]
                                    + g_ai2_ihevc_trans_8[3][k]
                                                    * pi2_tmp[3 * trans_size];
                }
                eo[0] = g_ai2_ihevc_trans_8[2][0] * pi2_tmp[2 * trans_size];
                eo[1] = g_ai2_ihevc_trans_8[2][1] * pi2_tmp[2 * trans_size];
                ee[0] = g_ai2_ihevc_trans_8[0][0] * pi2_tmp[0];
                ee[1] = g_ai2_ihevc_trans_8[0][1] * pi2_tmp[0];

                /* Combining e and o terms at each hierarchy levels to calculate the final spatial domain vector */
                e[0] = ee[0] + eo[0];
                e[3] = ee[0] - eo[0];
                e[1] = ee[1] + eo[1];
                e[2] = ee[1] - eo[1];
                for(k = 0; k < 4; k++)
                {
                    WORD32 itrans_out;
                    itrans_out =
                                    CLIP_S16(((e[k] + o[k] + add) >> shift));
                    pu1_dst[k * 2] = CLIP_U8((itrans_out + pu1_pred[k * 2]));
                    itrans_out =
                                    CLIP_S16(((e[3 - k] - o[3 - k] + add) >> shift));
                    pu1_dst[(k + 4) * 2] =
                                    CLIP_U8((itrans_out + pu1_pred[(k + 4) * 2]));
                }
                pi2_tmp++;
                pu1_pred += pred_strd;
                pu1_dst += dst_strd;
            }
        }
        else /* All rows of output of 1st stage are non-zero */
        {
            for(j = 0; j < trans_size; j++)
            {
                /* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
                for(k = 0; k < 4; k++)
                {
                    o[k] = g_ai2_ihevc_trans_8[1][k] * pi2_tmp[trans_size]
                                    + g_ai2_ihevc_trans_8[3][k]
                                                    * pi2_tmp[3 * trans_size]
                                    + g_ai2_ihevc_trans_8[5][k]
                                                    * pi2_tmp[5 * trans_size]
                                    + g_ai2_ihevc_trans_8[7][k]
                                                    * pi2_tmp[7 * trans_size];
                }

                eo[0] = g_ai2_ihevc_trans_8[2][0] * pi2_tmp[2 * trans_size]
                                + g_ai2_ihevc_trans_8[6][0] * pi2_tmp[6 * trans_size];
                eo[1] = g_ai2_ihevc_trans_8[2][1] * pi2_tmp[2 * trans_size]
                                + g_ai2_ihevc_trans_8[6][1] * pi2_tmp[6 * trans_size];
                ee[0] = g_ai2_ihevc_trans_8[0][0] * pi2_tmp[0]
                                + g_ai2_ihevc_trans_8[4][0] * pi2_tmp[4 * trans_size];
                ee[1] = g_ai2_ihevc_trans_8[0][1] * pi2_tmp[0]
                                + g_ai2_ihevc_trans_8[4][1] * pi2_tmp[4 * trans_size];

                /* Combining e and o terms at each hierarchy levels to calculate the final spatial domain vector */
                e[0] = ee[0] + eo[0];
                e[3] = ee[0] - eo[0];
                e[1] = ee[1] + eo[1];
                e[2] = ee[1] - eo[1];
                for(k = 0; k < 4; k++)
                {
                    WORD32 itrans_out;
                    itrans_out =
                                    CLIP_S16(((e[k] + o[k] + add) >> shift));
                    pu1_dst[k * 2] = CLIP_U8((itrans_out + pu1_pred[k * 2]));
                    itrans_out =
                                    CLIP_S16(((e[3 - k] - o[3 - k] + add) >> shift));
                    pu1_dst[(k + 4) * 2] =
                                    CLIP_U8((itrans_out + pu1_pred[(k + 4) * 2]));
                }
                pi2_tmp++;
                pu1_pred += pred_strd;
                pu1_dst += dst_strd;
            }
        }
        /************************************************************************************************/
        /************************************END - IT_RECON_8x8******************************************/
        /************************************************************************************************/
    }
}

Coverage Report

Created: 2026-02-14 06:43

Line	Count	Source
1		/******************************************************************************
2		*
3		* Copyright (C) 2012 Ittiam Systems Pvt Ltd, Bangalore
4		*
5		* Licensed under the Apache License, Version 2.0 (the "License");
6		* you may not use this file except in compliance with the License.
7		* You may obtain a copy of the License at:
8		*
9		* http://www.apache.org/licenses/LICENSE-2.0
10		*
11		* Unless required by applicable law or agreed to in writing, software
12		* distributed under the License is distributed on an "AS IS" BASIS,
13		* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14		* See the License for the specific language governing permissions and
15		* limitations under the License.
16		*
17		******************************************************************************/
18		/**
19		*******************************************************************************
20		* @file
21		* ihevc_chroma_itrans_recon_8x8.c
22		*
23		* @brief
24		* Contains function definitions for 8x8 inverse transform and reconstruction
25		* of chroma interleaved data.
26		*
27		* @author
28		* 100470
29		*
30		* @par List of Functions:
31		* - ihevc_chroma_itrans_recon_8x8()
32		*
33		* @remarks
34		* None
35		*
36		*******************************************************************************
37		*/
38
39		#include <stdio.h>
40		#include <string.h>
41		#include "ihevc_typedefs.h"
42		#include "ihevc_macros.h"
43		#include "ihevc_platform_macros.h"
44		#include "ihevc_defs.h"
45		#include "ihevc_trans_tables.h"
46		#include "ihevc_chroma_itrans_recon.h"
47		#include "ihevc_func_selector.h"
48		#include "ihevc_trans_macros.h"
49
50		/* All the functions work one component(U or V) of interleaved data depending upon pointers passed to it */
51		/* Data visualization */
52		/* U V U V U V U V */
53		/* U V U V U V U V */
54		/* U V U V U V U V */
55		/* U V U V U V U V */
56		/* If the pointer points to first byte of above stream (U) , functions will operate on U component */
57		/* If the pointer points to second byte of above stream (V) , functions will operate on V component */
58
59		/**
60		*******************************************************************************
61		*
62		* @brief
63		* This function performs Inverse transform and reconstruction for 8x8
64		* input block
65		*
66		* @par Description:
67		* Performs inverse transform and adds the prediction data and clips output
68		* to 8 bit
69		*
70		* @param[in] pi2_src
71		* Input 8x8 coefficients
72		*
73		* @param[in] pi2_tmp
74		* Temporary 8x8 buffer for storing inverse transform
75		* 1st stage output
76		*
77		* @param[in] pu1_pred
78		* Prediction 8x8 block
79		*
80		* @param[out] pu1_dst
81		* Output 8x8 block
82		*
83		* @param[in] src_strd
84		* Input stride
85		*
86		* @param[in] pred_strd
87		* Prediction stride
88		*
89		* @param[in] dst_strd
90		* Output Stride
91		*
92		* @param[in] shift
93		* Output shift
94		*
95		* @param[in] zero_cols
96		* Zero columns in pi2_src
97		*
98		* @returns Void
99		*
100		* @remarks
101		* None
102		*
103		*******************************************************************************
104		*/
105
106
107		void ihevc_chroma_itrans_recon_8x8(WORD16 *pi2_src,
108		WORD16 *pi2_tmp,
109		UWORD8 *pu1_pred,
110		UWORD8 *pu1_dst,
111		WORD32 src_strd,
112		WORD32 pred_strd,
113		WORD32 dst_strd,
114		WORD32 zero_cols,
115		WORD32 zero_rows)
116	114k	{
117	114k	WORD32 j, k;
118	114k	WORD32 e[4], o[4];
119	114k	WORD32 ee[2], eo[2];
120	114k	WORD32 add;
121	114k	WORD32 shift;
122	114k	WORD16 *pi2_tmp_orig;
123	114k	WORD32 trans_size;
124	114k	WORD32 zero_rows_2nd_stage = zero_cols;
125	114k	WORD32 row_limit_2nd_stage;
126	114k	UNUSED(zero_rows);
127	114k	trans_size = TRANS_SIZE_8;
128
129	114k	pi2_tmp_orig = pi2_tmp;
130
131	114k	if((zero_cols & 0xF0) == 0xF0)
132	58.1k	row_limit_2nd_stage = 4;
133	56.7k	else
134	56.7k	row_limit_2nd_stage = TRANS_SIZE_8;
135
136		/* Inverse Transform 1st stage */
137	114k	shift = IT_SHIFT_STAGE_1;
138	114k	add = 1 << (shift - 1);
139	114k	{
140		/************************************************************************************************/
141		/********************************START - IT_RECON_8x8****************************************/
142		/************************************************************************************************/
143
144	801k	for(j = 0; j < row_limit_2nd_stage; j++)
145	686k	{
146		/* Checking for Zero Cols */
147	686k	if((zero_cols & 1) == 1)
148	144k	{
149	144k	memset(pi2_tmp, 0, trans_size * sizeof(WORD16));
150	144k	}
151	542k	else
152	542k	{
153		/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
154	2.71M	for(k = 0; k < 4; k++)
155	2.16M	{
156	2.16M	o[k] = g_ai2_ihevc_trans_8[1][k] * pi2_src[src_strd]
157	2.16M	+ g_ai2_ihevc_trans_8[3][k]
158	2.16M	* pi2_src[3 * src_strd]
159	2.16M	+ g_ai2_ihevc_trans_8[5][k]
160	2.16M	* pi2_src[5 * src_strd]
161	2.16M	+ g_ai2_ihevc_trans_8[7][k]
162	2.16M	* pi2_src[7 * src_strd];
163	2.16M	}
164
165	542k	eo[0] = g_ai2_ihevc_trans_8[2][0] * pi2_src[2 * src_strd]
166	542k	+ g_ai2_ihevc_trans_8[6][0] * pi2_src[6 * src_strd];
167	542k	eo[1] = g_ai2_ihevc_trans_8[2][1] * pi2_src[2 * src_strd]
168	542k	+ g_ai2_ihevc_trans_8[6][1] * pi2_src[6 * src_strd];
169	542k	ee[0] = g_ai2_ihevc_trans_8[0][0] * pi2_src[0]
170	542k	+ g_ai2_ihevc_trans_8[4][0] * pi2_src[4 * src_strd];
171	542k	ee[1] = g_ai2_ihevc_trans_8[0][1] * pi2_src[0]
172	542k	+ g_ai2_ihevc_trans_8[4][1] * pi2_src[4 * src_strd];
173
174		/* Combining e and o terms at each hierarchy levels to calculate the final spatial domain vector */
175	542k	e[0] = ee[0] + eo[0];
176	542k	e[3] = ee[0] - eo[0];
177	542k	e[1] = ee[1] + eo[1];
178	542k	e[2] = ee[1] - eo[1];
179	2.71M	for(k = 0; k < 4; k++)
180	2.16M	{
181	2.16M	pi2_tmp[k] =
182	2.16M	CLIP_S16(((e[k] + o[k] + add) >> shift));
183	2.16M	pi2_tmp[k + 4] =
184	2.16M	CLIP_S16(((e[3 - k] - o[3 - k] + add) >> shift));
185	2.16M	}
186	542k	}
187	686k	pi2_src++;
188	686k	pi2_tmp += trans_size;
189	686k	zero_cols = zero_cols >> 1;
190	686k	}
191
192	114k	pi2_tmp = pi2_tmp_orig;
193
194		/* Inverse Transform 2nd stage */
195	114k	shift = IT_SHIFT_STAGE_2;
196	114k	add = 1 << (shift - 1);
197
198	114k	if((zero_rows_2nd_stage & 0xF0) == 0xF0) /* First 4 rows of output of 1st stage are non-zero */
199	58.1k	{
200	523k	for(j = 0; j < trans_size; j++)
201	465k	{
202		/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
203	2.32M	for(k = 0; k < 4; k++)
204	1.86M	{
205	1.86M	o[k] = g_ai2_ihevc_trans_8[1][k] * pi2_tmp[trans_size]
206	1.86M	+ g_ai2_ihevc_trans_8[3][k]
207	1.86M	* pi2_tmp[3 * trans_size];
208	1.86M	}
209	465k	eo[0] = g_ai2_ihevc_trans_8[2][0] * pi2_tmp[2 * trans_size];
210	465k	eo[1] = g_ai2_ihevc_trans_8[2][1] * pi2_tmp[2 * trans_size];
211	465k	ee[0] = g_ai2_ihevc_trans_8[0][0] * pi2_tmp[0];
212	465k	ee[1] = g_ai2_ihevc_trans_8[0][1] * pi2_tmp[0];
213
214		/* Combining e and o terms at each hierarchy levels to calculate the final spatial domain vector */
215	465k	e[0] = ee[0] + eo[0];
216	465k	e[3] = ee[0] - eo[0];
217	465k	e[1] = ee[1] + eo[1];
218	465k	e[2] = ee[1] - eo[1];
219	2.32M	for(k = 0; k < 4; k++)
220	1.85M	{
221	1.85M	WORD32 itrans_out;
222	1.85M	itrans_out =
223	1.85M	CLIP_S16(((e[k] + o[k] + add) >> shift));
224	1.85M	pu1_dst[k * 2] = CLIP_U8((itrans_out + pu1_pred[k * 2]));
225	1.85M	itrans_out =
226	1.85M	CLIP_S16(((e[3 - k] - o[3 - k] + add) >> shift));
227	1.85M	pu1_dst[(k + 4) * 2] =
228	1.85M	CLIP_U8((itrans_out + pu1_pred[(k + 4) * 2]));
229	1.85M	}
230	465k	pi2_tmp++;
231	465k	pu1_pred += pred_strd;
232	465k	pu1_dst += dst_strd;
233	465k	}
234	58.1k	}
235	56.7k	else /* All rows of output of 1st stage are non-zero */
236	56.7k	{
237	510k	for(j = 0; j < trans_size; j++)
238	454k	{
239		/* Utilizing symmetry properties to the maximum to minimize the number of multiplications */
240	2.27M	for(k = 0; k < 4; k++)
241	1.81M	{
242	1.81M	o[k] = g_ai2_ihevc_trans_8[1][k] * pi2_tmp[trans_size]
243	1.81M	+ g_ai2_ihevc_trans_8[3][k]
244	1.81M	* pi2_tmp[3 * trans_size]
245	1.81M	+ g_ai2_ihevc_trans_8[5][k]
246	1.81M	* pi2_tmp[5 * trans_size]
247	1.81M	+ g_ai2_ihevc_trans_8[7][k]
248	1.81M	* pi2_tmp[7 * trans_size];
249	1.81M	}
250
251	454k	eo[0] = g_ai2_ihevc_trans_8[2][0] * pi2_tmp[2 * trans_size]
252	454k	+ g_ai2_ihevc_trans_8[6][0] * pi2_tmp[6 * trans_size];
253	454k	eo[1] = g_ai2_ihevc_trans_8[2][1] * pi2_tmp[2 * trans_size]
254	454k	+ g_ai2_ihevc_trans_8[6][1] * pi2_tmp[6 * trans_size];
255	454k	ee[0] = g_ai2_ihevc_trans_8[0][0] * pi2_tmp[0]
256	454k	+ g_ai2_ihevc_trans_8[4][0] * pi2_tmp[4 * trans_size];
257	454k	ee[1] = g_ai2_ihevc_trans_8[0][1] * pi2_tmp[0]
258	454k	+ g_ai2_ihevc_trans_8[4][1] * pi2_tmp[4 * trans_size];
259
260		/* Combining e and o terms at each hierarchy levels to calculate the final spatial domain vector */
261	454k	e[0] = ee[0] + eo[0];
262	454k	e[3] = ee[0] - eo[0];
263	454k	e[1] = ee[1] + eo[1];
264	454k	e[2] = ee[1] - eo[1];
265	2.27M	for(k = 0; k < 4; k++)
266	1.81M	{
267	1.81M	WORD32 itrans_out;
268	1.81M	itrans_out =
269	1.81M	CLIP_S16(((e[k] + o[k] + add) >> shift));
270	1.81M	pu1_dst[k * 2] = CLIP_U8((itrans_out + pu1_pred[k * 2]));
271	1.81M	itrans_out =
272	1.81M	CLIP_S16(((e[3 - k] - o[3 - k] + add) >> shift));
273	1.81M	pu1_dst[(k + 4) * 2] =
274	1.81M	CLIP_U8((itrans_out + pu1_pred[(k + 4) * 2]));
275	1.81M	}
276	454k	pi2_tmp++;
277	454k	pu1_pred += pred_strd;
278	454k	pu1_dst += dst_strd;
279	454k	}
280	56.7k	}
281		/************************************************************************************************/
282		/**********************************END - IT_RECON_8x8****************************************/
283		/************************************************************************************************/
284	114k	}
285	114k	}