1 /* ------------------------------------------------------------------
2 * Copyright (C) 1998-2009 PacketVideo
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either
14 * See the License for the specific language governing permissions
15 * and limitations under the License.
16 * -------------------------------------------------------------------
21 ------------------------------------------------------------------------------
28 ------------------------------------------------------------------------------
29 INPUT AND OUTPUT DEFINITIONS
31 Int32 x 32-bit integer input length 16
33 Int32 flag 1 forward dct16, 0 modified dct-16
35 ------------------------------------------------------------------------------
38 Implement dct of lenght 16
40 ------------------------------------------------------------------------------
44 ------------------------------------------------------------------------------
47 ------------------------------------------------------------------------------
50 ------------------------------------------------------------------------------
54 /*----------------------------------------------------------------------------
56 ----------------------------------------------------------------------------*/
61 #include "fxp_mul32.h"
63 /*----------------------------------------------------------------------------
65 ; Define module specific macros here
66 ----------------------------------------------------------------------------*/
69 /*----------------------------------------------------------------------------
71 ; Include all pre-processor statements here. Include conditional
72 ; compile variables also.
73 ----------------------------------------------------------------------------*/
75 #define Qfmt_31(a) (Int32)(a*0x7FFFFFFF + (a>=0?0.5F:-0.5F))
77 #define Qfmt15(x) (Int16)(x*((Int32)1<<15) + (x>=0?0.5F:-0.5F))
79 /*----------------------------------------------------------------------------
80 ; LOCAL FUNCTION DEFINITIONS
81 ; Function Prototype declaration
82 ----------------------------------------------------------------------------*/
84 /*----------------------------------------------------------------------------
85 ; LOCAL STORE/BUFFER/POINTER DEFINITIONS
86 ; Variable declaration - defined here and used outside this module
87 ----------------------------------------------------------------------------*/
89 /*----------------------------------------------------------------------------
90 ; EXTERNAL FUNCTION REFERENCES
91 ; Declare functions defined elsewhere and referenced in this module
92 ----------------------------------------------------------------------------*/
94 /*----------------------------------------------------------------------------
95 ; EXTERNAL GLOBAL STORE/BUFFER/POINTER REFERENCES
96 ; Declare variables used in this module but defined elsewhere
97 ----------------------------------------------------------------------------*/
99 /*----------------------------------------------------------------------------
101 ----------------------------------------------------------------------------*/
104 void dct_16(Int32 vec[], Int flag)
126 /* split input vector */
129 tmp_o0 = fxp_mul32_by_16((vec[ 0] - vec[15]), Qfmt15(0.50241928618816F));
130 tmp0 = vec[ 0] + vec[15];
132 tmp_o7 = fxp_mul32_Q31((vec[ 7] - vec[ 8]) << 3, Qfmt_31(0.63764357733614F));
133 tmp7 = vec[ 7] + vec[ 8];
135 itmp_e0 = (tmp0 + tmp7);
136 tmp7 = fxp_mul32_by_16((tmp0 - tmp7), Qfmt15(0.50979557910416F));
138 tmp_o1 = fxp_mul32_by_16((vec[ 1] - vec[14]), Qfmt15(0.52249861493969F));
139 tmp1 = vec[ 1] + vec[14];
140 tmp_o6 = fxp_mul32_by_16((vec[ 6] - vec[ 9]) << 1, Qfmt15(0.86122354911916F));
141 tmp6 = vec[ 6] + vec[ 9];
143 itmp_e1 = (tmp1 + tmp6);
144 tmp6 = fxp_mul32_by_16((tmp1 - tmp6), Qfmt15(0.60134488693505F));
146 tmp_o2 = fxp_mul32_by_16((vec[ 2] - vec[13]), Qfmt15(0.56694403481636F));
147 tmp2 = vec[ 2] + vec[13];
148 tmp_o5 = fxp_mul32_by_16((vec[ 5] - vec[10]) << 1, Qfmt15(0.53033884299517F));
149 tmp5 = vec[ 5] + vec[10];
151 itmp_e2 = (tmp2 + tmp5);
152 tmp5 = fxp_mul32_by_16((tmp2 - tmp5), Qfmt15(0.89997622313642F));
154 tmp_o3 = fxp_mul32_by_16((vec[ 3] - vec[12]), Qfmt15(0.64682178335999F));
155 tmp3 = vec[ 3] + vec[12];
156 tmp_o4 = fxp_mul32_by_16((vec[ 4] - vec[11]), Qfmt15(0.78815462345125F));
157 tmp4 = vec[ 4] + vec[11];
159 tmp1 = (tmp3 + tmp4);
160 tmp4 = fxp_mul32_Q31((tmp3 - tmp4) << 2, Qfmt_31(0.64072886193538F));
162 /* split even part of tmp_e */
164 tmp0 = (itmp_e0 + tmp1);
165 tmp1 = fxp_mul32_by_16((itmp_e0 - tmp1), Qfmt15(0.54119610014620F));
168 tmp3 = fxp_mul32_by_16((itmp_e1 - itmp_e2) << 1, Qfmt15(0.65328148243819F));
169 tmp2 = (itmp_e1 + itmp_e2);
171 vec[ 0] = (tmp0 + tmp2) >> 1;
172 vec[ 8] = fxp_mul32_by_16((tmp0 - tmp2), Qfmt15(0.70710678118655F));
173 vec[12] = fxp_mul32_by_16((tmp1 - tmp3) << 1, Qfmt15(0.70710678118655F));
174 vec[ 4] = tmp1 + tmp3;
177 /* split odd part of tmp_e */
179 tmp1 = fxp_mul32_by_16((tmp7 - tmp4) << 1, Qfmt15(0.54119610014620F));
181 tmp3 = fxp_mul32_Q31((tmp6 - tmp5) << 2, Qfmt_31(0.65328148243819F));
185 vec[10] = fxp_mul32_by_16((tmp7 - tmp6) << 1, Qfmt15(0.70710678118655F));
186 vec[ 2] = tmp7 + tmp6;
187 vec[14] = fxp_mul32_by_16((tmp1 - tmp3) << 1, Qfmt15(0.70710678118655F));
189 tmp1 += tmp3 + vec[14];
191 vec[ 6] = tmp1 + vec[10];
198 tmp7 = tmp_o0 + tmp_o7;
199 tmp_o7 = fxp_mul32_by_16((tmp_o0 - tmp_o7) << 1, Qfmt15(0.50979557910416F));
201 tmp6 = tmp_o1 + tmp_o6;
202 tmp_o1 = fxp_mul32_by_16((tmp_o1 - tmp_o6) << 1, Qfmt15(0.60134488693505F));
204 tmp5 = tmp_o2 + tmp_o5;
205 tmp_o5 = fxp_mul32_by_16((tmp_o2 - tmp_o5) << 1, Qfmt15(0.89997622313642F));
207 tmp4 = tmp_o3 + tmp_o4;
209 tmp_o3 = fxp_mul32_Q31((tmp_o3 - tmp_o4) << 3, Qfmt_31(0.6407288619354F));
225 tmp1 = fxp_mul32_by_16((tmp7 - tmp4) << 1, Qfmt15(0.54119610014620F));
227 tmp3 = fxp_mul32_Q31((tmp6 - tmp5) << 2, Qfmt_31(0.65328148243819F));
230 vec[ 9] = fxp_mul32_Q31((tmp0 - tmp2) << 1, Qfmt_31(0.70710678118655F));
231 vec[ 1] = tmp0 + tmp2;
232 vec[13] = fxp_mul32_Q31((tmp1 - tmp3) << 1, Qfmt_31(0.70710678118655F));
234 vec[ 5] = tmp1 + tmp3 + vec[13];
238 tmp0 = tmp_o7 + tmp_o3;
239 tmp1 = fxp_mul32_by_16((tmp_o7 - tmp_o3) << 1, Qfmt15(0.54119610014620F));
240 tmp2 = tmp_o1 + tmp_o5;
241 tmp3 = fxp_mul32_Q31((tmp_o1 - tmp_o5) << 2, Qfmt_31(0.65328148243819F));
243 vec[11] = fxp_mul32_Q31((tmp0 - tmp2) << 1, Qfmt_31(0.70710678118655F));
244 vec[ 3] = tmp0 + tmp2;
245 vec[15] = fxp_mul32_Q31((tmp1 - tmp3) << 1, Qfmt_31(0.70710678118655F));
246 vec[ 7] = tmp1 + tmp3 + vec[15];