fwd_txfm_msa.h 22.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 *  Copyright (c) 2015 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 VPX_DSP_MIPS_FWD_TXFM_MSA_H_
#define VPX_DSP_MIPS_FWD_TXFM_MSA_H_

#include "vpx_dsp/mips/txfm_macros_msa.h"
Jingning Han's avatar
Jingning Han committed
15
#include "vpx_dsp/txfm_common.h"
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275

#define VP9_FDCT4(in0, in1, in2, in3, out0, out1, out2, out3) {     \
  v8i16 cnst0_m, cnst1_m, cnst2_m, cnst3_m;                         \
  v8i16 vec0_m, vec1_m, vec2_m, vec3_m;                             \
  v4i32 vec4_m, vec5_m, vec6_m, vec7_m;                             \
  v8i16 coeff_m = { cospi_16_64, -cospi_16_64, cospi_8_64,          \
                    cospi_24_64, -cospi_8_64, 0, 0, 0 };            \
                                                                    \
  BUTTERFLY_4(in0, in1, in2, in3, vec0_m, vec1_m, vec2_m, vec3_m);  \
  ILVR_H2_SH(vec1_m, vec0_m, vec3_m, vec2_m, vec0_m, vec2_m);       \
  SPLATI_H2_SH(coeff_m, 0, 1, cnst0_m, cnst1_m);                    \
  cnst1_m = __msa_ilvev_h(cnst1_m, cnst0_m);                        \
  vec5_m = __msa_dotp_s_w(vec0_m, cnst1_m);                         \
                                                                    \
  SPLATI_H2_SH(coeff_m, 4, 3, cnst2_m, cnst3_m);                    \
  cnst2_m = __msa_ilvev_h(cnst3_m, cnst2_m);                        \
  vec7_m = __msa_dotp_s_w(vec2_m, cnst2_m);                         \
                                                                    \
  vec4_m = __msa_dotp_s_w(vec0_m, cnst0_m);                         \
  cnst2_m = __msa_splati_h(coeff_m, 2);                             \
  cnst2_m = __msa_ilvev_h(cnst2_m, cnst3_m);                        \
  vec6_m = __msa_dotp_s_w(vec2_m, cnst2_m);                         \
                                                                    \
  SRARI_W4_SW(vec4_m, vec5_m, vec6_m, vec7_m, DCT_CONST_BITS);      \
  PCKEV_H4_SH(vec4_m, vec4_m, vec5_m, vec5_m, vec6_m, vec6_m,       \
              vec7_m, vec7_m, out0, out2, out1, out3);              \
}

#define SRLI_AVE_S_4V_H(in0, in1, in2, in3, in4, in5, in6, in7) {        \
  v8i16 vec0_m, vec1_m, vec2_m, vec3_m, vec4_m, vec5_m, vec6_m, vec7_m;  \
                                                                         \
  SRLI_H4_SH(in0, in1, in2, in3, vec0_m, vec1_m, vec2_m, vec3_m, 15);    \
  SRLI_H4_SH(in4, in5, in6, in7, vec4_m, vec5_m, vec6_m, vec7_m, 15);    \
  AVE_SH4_SH(vec0_m, in0, vec1_m, in1, vec2_m, in2, vec3_m, in3,         \
             in0, in1, in2, in3);                                        \
  AVE_SH4_SH(vec4_m, in4, vec5_m, in5, vec6_m, in6, vec7_m, in7,         \
             in4, in5, in6, in7);                                        \
}

#define VP9_FDCT8(in0, in1, in2, in3, in4, in5, in6, in7,            \
                  out0, out1, out2, out3, out4, out5, out6, out7) {  \
  v8i16 s0_m, s1_m, s2_m, s3_m, s4_m, s5_m, s6_m;                    \
  v8i16 s7_m, x0_m, x1_m, x2_m, x3_m;                                \
  v8i16 coeff_m = { cospi_16_64, -cospi_16_64, cospi_8_64,           \
                    cospi_24_64, cospi_4_64, cospi_28_64,            \
                    cospi_12_64, cospi_20_64 };                      \
                                                                     \
  /* FDCT stage1 */                                                  \
  BUTTERFLY_8(in0, in1, in2, in3, in4, in5, in6, in7,                \
              s0_m, s1_m, s2_m, s3_m, s4_m, s5_m, s6_m, s7_m);       \
  BUTTERFLY_4(s0_m, s1_m, s2_m, s3_m, x0_m, x1_m, x2_m, x3_m);       \
  ILVL_H2_SH(x1_m, x0_m, x3_m, x2_m, s0_m, s2_m);                    \
  ILVR_H2_SH(x1_m, x0_m, x3_m, x2_m, s1_m, s3_m);                    \
  SPLATI_H2_SH(coeff_m, 0, 1, x0_m, x1_m);                           \
  x1_m = __msa_ilvev_h(x1_m, x0_m);                                  \
  out4 = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x1_m);                    \
                                                                     \
  SPLATI_H2_SH(coeff_m, 2, 3, x2_m, x3_m);                           \
  x2_m = -x2_m;                                                      \
  x2_m = __msa_ilvev_h(x3_m, x2_m);                                  \
  out6 = DOT_SHIFT_RIGHT_PCK_H(s2_m, s3_m, x2_m);                    \
                                                                     \
  out0 = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x0_m);                    \
  x2_m = __msa_splati_h(coeff_m, 2);                                 \
  x2_m = __msa_ilvev_h(x2_m, x3_m);                                  \
  out2 = DOT_SHIFT_RIGHT_PCK_H(s2_m, s3_m, x2_m);                    \
                                                                     \
  /* stage2 */                                                       \
  ILVRL_H2_SH(s5_m, s6_m, s1_m, s0_m);                               \
                                                                     \
  s6_m = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x0_m);                    \
  s5_m = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x1_m);                    \
                                                                     \
  /* stage3 */                                                       \
  BUTTERFLY_4(s4_m, s7_m, s6_m, s5_m, x0_m, x3_m, x2_m, x1_m);       \
                                                                     \
  /* stage4 */                                                       \
  ILVL_H2_SH(x3_m, x0_m, x2_m, x1_m, s4_m, s6_m);                    \
  ILVR_H2_SH(x3_m, x0_m, x2_m, x1_m, s5_m, s7_m);                    \
                                                                     \
  SPLATI_H2_SH(coeff_m, 4, 5, x0_m, x1_m);                           \
  x1_m = __msa_ilvev_h(x0_m, x1_m);                                  \
  out1 = DOT_SHIFT_RIGHT_PCK_H(s4_m, s5_m, x1_m);                    \
                                                                     \
  SPLATI_H2_SH(coeff_m, 6, 7, x2_m, x3_m);                           \
  x2_m = __msa_ilvev_h(x3_m, x2_m);                                  \
  out5 = DOT_SHIFT_RIGHT_PCK_H(s6_m, s7_m, x2_m);                    \
                                                                     \
  x1_m = __msa_splati_h(coeff_m, 5);                                 \
  x0_m = -x0_m;                                                      \
  x0_m = __msa_ilvev_h(x1_m, x0_m);                                  \
  out7 = DOT_SHIFT_RIGHT_PCK_H(s4_m, s5_m, x0_m);                    \
                                                                     \
  x2_m = __msa_splati_h(coeff_m, 6);                                 \
  x3_m = -x3_m;                                                      \
  x2_m = __msa_ilvev_h(x2_m, x3_m);                                  \
  out3 = DOT_SHIFT_RIGHT_PCK_H(s6_m, s7_m, x2_m);                    \
}

#define FDCT8x16_EVEN(in0, in1, in2, in3, in4, in5, in6, in7,                \
                      out0, out1, out2, out3, out4, out5, out6, out7) {      \
  v8i16 s0_m, s1_m, s2_m, s3_m, s4_m, s5_m, s6_m, s7_m;                      \
  v8i16 x0_m, x1_m, x2_m, x3_m;                                              \
  v8i16 coeff_m = { cospi_16_64, -cospi_16_64, cospi_8_64, cospi_24_64,      \
                    cospi_4_64, cospi_28_64, cospi_12_64, cospi_20_64 };     \
                                                                             \
  /* FDCT stage1 */                                                          \
  BUTTERFLY_8(in0, in1, in2, in3, in4, in5, in6, in7,                        \
              s0_m, s1_m, s2_m, s3_m, s4_m, s5_m, s6_m, s7_m);               \
  BUTTERFLY_4(s0_m, s1_m, s2_m, s3_m, x0_m, x1_m, x2_m, x3_m);               \
  ILVL_H2_SH(x1_m, x0_m, x3_m, x2_m, s0_m, s2_m);                            \
  ILVR_H2_SH(x1_m, x0_m, x3_m, x2_m, s1_m, s3_m);                            \
  SPLATI_H2_SH(coeff_m, 0, 1, x0_m, x1_m);                                   \
  x1_m = __msa_ilvev_h(x1_m, x0_m);                                          \
  out4 = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x1_m);                            \
                                                                             \
  SPLATI_H2_SH(coeff_m, 2, 3, x2_m, x3_m);                                   \
  x2_m = -x2_m;                                                              \
  x2_m = __msa_ilvev_h(x3_m, x2_m);                                          \
  out6 = DOT_SHIFT_RIGHT_PCK_H(s2_m, s3_m, x2_m);                            \
                                                                             \
  out0 = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x0_m);                            \
  x2_m = __msa_splati_h(coeff_m, 2);                                         \
  x2_m = __msa_ilvev_h(x2_m, x3_m);                                          \
  out2 = DOT_SHIFT_RIGHT_PCK_H(s2_m, s3_m, x2_m);                            \
                                                                             \
  /* stage2 */                                                               \
  ILVRL_H2_SH(s5_m, s6_m, s1_m, s0_m);                                       \
                                                                             \
  s6_m = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x0_m);                            \
  s5_m = DOT_SHIFT_RIGHT_PCK_H(s0_m, s1_m, x1_m);                            \
                                                                             \
  /* stage3 */                                                               \
  BUTTERFLY_4(s4_m, s7_m, s6_m, s5_m, x0_m, x3_m, x2_m, x1_m);               \
                                                                             \
  /* stage4 */                                                               \
  ILVL_H2_SH(x3_m, x0_m, x2_m, x1_m, s4_m, s6_m);                            \
  ILVR_H2_SH(x3_m, x0_m, x2_m, x1_m, s5_m, s7_m);                            \
                                                                             \
  SPLATI_H2_SH(coeff_m, 4, 5, x0_m, x1_m);                                   \
  x1_m = __msa_ilvev_h(x0_m, x1_m);                                          \
  out1 = DOT_SHIFT_RIGHT_PCK_H(s4_m, s5_m, x1_m);                            \
                                                                             \
  SPLATI_H2_SH(coeff_m, 6, 7, x2_m, x3_m);                                   \
  x2_m = __msa_ilvev_h(x3_m, x2_m);                                          \
  out5 = DOT_SHIFT_RIGHT_PCK_H(s6_m, s7_m, x2_m);                            \
                                                                             \
  x1_m = __msa_splati_h(coeff_m, 5);                                         \
  x0_m = -x0_m;                                                              \
  x0_m = __msa_ilvev_h(x1_m, x0_m);                                          \
  out7 = DOT_SHIFT_RIGHT_PCK_H(s4_m, s5_m, x0_m);                            \
                                                                             \
  x2_m = __msa_splati_h(coeff_m, 6);                                         \
  x3_m = -x3_m;                                                              \
  x2_m = __msa_ilvev_h(x2_m, x3_m);                                          \
  out3 = DOT_SHIFT_RIGHT_PCK_H(s6_m, s7_m, x2_m);                            \
}

#define FDCT8x16_ODD(input0, input1, input2, input3,               \
                     input4, input5, input6, input7,               \
                     out1, out3, out5, out7,                       \
                     out9, out11, out13, out15) {                  \
  v8i16 stp21_m, stp22_m, stp23_m, stp24_m, stp25_m, stp26_m;      \
  v8i16 stp30_m, stp31_m, stp32_m, stp33_m, stp34_m, stp35_m;      \
  v8i16 stp36_m, stp37_m, vec0_m, vec1_m;                          \
  v8i16 vec2_m, vec3_m, vec4_m, vec5_m, vec6_m;                    \
  v8i16 cnst0_m, cnst1_m, cnst4_m, cnst5_m;                        \
  v8i16 coeff_m = { cospi_16_64, -cospi_16_64, cospi_8_64,         \
                    cospi_24_64, -cospi_8_64, -cospi_24_64,        \
                    cospi_12_64, cospi_20_64 };                    \
  v8i16 coeff1_m = { cospi_2_64, cospi_30_64, cospi_14_64,         \
                     cospi_18_64, cospi_10_64, cospi_22_64,        \
                     cospi_6_64, cospi_26_64 };                    \
  v8i16 coeff2_m = { -cospi_2_64, -cospi_10_64, -cospi_18_64,      \
                     -cospi_26_64, 0, 0, 0, 0 };                   \
                                                                   \
  /* stp 1 */                                                      \
  ILVL_H2_SH(input2, input5, input3, input4, vec2_m, vec4_m);      \
  ILVR_H2_SH(input2, input5, input3, input4, vec3_m, vec5_m);      \
                                                                   \
  cnst4_m = __msa_splati_h(coeff_m, 0);                            \
  stp25_m = DOT_SHIFT_RIGHT_PCK_H(vec2_m, vec3_m, cnst4_m);        \
                                                                   \
  cnst5_m = __msa_splati_h(coeff_m, 1);                            \
  cnst5_m = __msa_ilvev_h(cnst5_m, cnst4_m);                       \
  stp22_m = DOT_SHIFT_RIGHT_PCK_H(vec2_m, vec3_m, cnst5_m);        \
  stp24_m = DOT_SHIFT_RIGHT_PCK_H(vec4_m, vec5_m, cnst4_m);        \
  stp23_m = DOT_SHIFT_RIGHT_PCK_H(vec4_m, vec5_m, cnst5_m);        \
                                                                   \
  /* stp2 */                                                       \
  BUTTERFLY_4(input0, input1, stp22_m, stp23_m,                    \
              stp30_m, stp31_m, stp32_m, stp33_m);                 \
  BUTTERFLY_4(input7, input6, stp25_m, stp24_m,                    \
              stp37_m, stp36_m, stp35_m, stp34_m);                 \
                                                                   \
  ILVL_H2_SH(stp36_m, stp31_m, stp35_m, stp32_m, vec2_m, vec4_m);  \
  ILVR_H2_SH(stp36_m, stp31_m, stp35_m, stp32_m, vec3_m, vec5_m);  \
                                                                   \
  SPLATI_H2_SH(coeff_m, 2, 3, cnst0_m, cnst1_m);                   \
  cnst0_m = __msa_ilvev_h(cnst0_m, cnst1_m);                       \
  stp26_m = DOT_SHIFT_RIGHT_PCK_H(vec2_m, vec3_m, cnst0_m);        \
                                                                   \
  cnst0_m = __msa_splati_h(coeff_m, 4);                            \
  cnst1_m = __msa_ilvev_h(cnst1_m, cnst0_m);                       \
  stp21_m = DOT_SHIFT_RIGHT_PCK_H(vec2_m, vec3_m, cnst1_m);        \
                                                                   \
  SPLATI_H2_SH(coeff_m, 5, 2, cnst0_m, cnst1_m);                   \
  cnst1_m = __msa_ilvev_h(cnst0_m, cnst1_m);                       \
  stp25_m = DOT_SHIFT_RIGHT_PCK_H(vec4_m, vec5_m, cnst1_m);        \
                                                                   \
  cnst0_m = __msa_splati_h(coeff_m, 3);                            \
  cnst1_m = __msa_ilvev_h(cnst1_m, cnst0_m);                       \
  stp22_m = DOT_SHIFT_RIGHT_PCK_H(vec4_m, vec5_m, cnst1_m);        \
                                                                   \
  /* stp4 */                                                       \
  BUTTERFLY_4(stp30_m, stp37_m, stp26_m, stp21_m,                  \
              vec6_m, vec2_m, vec4_m, vec5_m);                     \
  BUTTERFLY_4(stp33_m, stp34_m, stp25_m, stp22_m,                  \
              stp21_m, stp23_m, stp24_m, stp31_m);                 \
                                                                   \
  ILVRL_H2_SH(vec2_m, vec6_m, vec1_m, vec0_m);                     \
  SPLATI_H2_SH(coeff1_m, 0, 1, cnst0_m, cnst1_m);                  \
  cnst0_m = __msa_ilvev_h(cnst0_m, cnst1_m);                       \
                                                                   \
  out1 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst0_m);           \
                                                                   \
  cnst0_m = __msa_splati_h(coeff2_m, 0);                           \
  cnst0_m = __msa_ilvev_h(cnst1_m, cnst0_m);                       \
  out15 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst0_m);          \
                                                                   \
  ILVRL_H2_SH(vec4_m, vec5_m, vec1_m, vec0_m);                     \
  SPLATI_H2_SH(coeff1_m, 2, 3, cnst0_m, cnst1_m);                  \
  cnst1_m = __msa_ilvev_h(cnst1_m, cnst0_m);                       \
                                                                   \
  out9 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst1_m);           \
                                                                   \
  cnst1_m = __msa_splati_h(coeff2_m, 2);                           \
  cnst0_m = __msa_ilvev_h(cnst0_m, cnst1_m);                       \
  out7 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst0_m);           \
                                                                   \
  ILVRL_H2_SH(stp23_m, stp21_m, vec1_m, vec0_m);                   \
  SPLATI_H2_SH(coeff1_m, 4, 5, cnst0_m, cnst1_m);                  \
  cnst0_m = __msa_ilvev_h(cnst0_m, cnst1_m);                       \
  out5 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst0_m);           \
                                                                   \
  cnst0_m = __msa_splati_h(coeff2_m, 1);                           \
  cnst0_m = __msa_ilvev_h(cnst1_m, cnst0_m);                       \
  out11 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst0_m);          \
                                                                   \
  ILVRL_H2_SH(stp24_m, stp31_m, vec1_m, vec0_m);                   \
  SPLATI_H2_SH(coeff1_m, 6, 7, cnst0_m, cnst1_m);                  \
  cnst1_m = __msa_ilvev_h(cnst1_m, cnst0_m);                       \
                                                                   \
  out13 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst1_m);          \
                                                                   \
  cnst1_m = __msa_splati_h(coeff2_m, 3);                           \
  cnst0_m = __msa_ilvev_h(cnst0_m, cnst1_m);                       \
  out3 = DOT_SHIFT_RIGHT_PCK_H(vec0_m, vec1_m, cnst0_m);           \
}

276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
#define FDCT_POSTPROC_2V_NEG_H(vec0, vec1) {      \
  v8i16 tp0_m, tp1_m;                             \
  v8i16 one_m = __msa_ldi_h(1);                   \
                                                  \
  tp0_m = __msa_clti_s_h(vec0, 0);                \
  tp1_m = __msa_clti_s_h(vec1, 0);                \
  vec0 += 1;                                      \
  vec1 += 1;                                      \
  tp0_m = one_m & tp0_m;                          \
  tp1_m = one_m & tp1_m;                          \
  vec0 += tp0_m;                                  \
  vec1 += tp1_m;                                  \
  vec0 >>= 2;                                     \
  vec1 >>= 2;                                     \
}

#define FDCT32_POSTPROC_NEG_W(vec) {      \
  v4i32 temp_m;                           \
  v4i32 one_m = __msa_ldi_w(1);           \
                                          \
  temp_m = __msa_clti_s_w(vec, 0);        \
  vec += 1;                               \
  temp_m = one_m & temp_m;                \
  vec += temp_m;                          \
  vec >>= 2;                              \
}

#define FDCT32_POSTPROC_2V_POS_H(vec0, vec1) {      \
  v8i16 tp0_m, tp1_m;                               \
  v8i16 one = __msa_ldi_h(1);                       \
                                                    \
  tp0_m = __msa_clei_s_h(vec0, 0);                  \
  tp1_m = __msa_clei_s_h(vec1, 0);                  \
  tp0_m = (v8i16)__msa_xori_b((v16u8)tp0_m, 255);   \
  tp1_m = (v8i16)__msa_xori_b((v16u8)tp1_m, 255);   \
  vec0 += 1;                                        \
  vec1 += 1;                                        \
  tp0_m = one & tp0_m;                              \
  tp1_m = one & tp1_m;                              \
  vec0 += tp0_m;                                    \
  vec1 += tp1_m;                                    \
  vec0 >>= 2;                                       \
  vec1 >>= 2;                                       \
}

#define DOTP_CONST_PAIR_W(reg0_left, reg1_left, reg0_right,      \
                          reg1_right, const0, const1,            \
                          out0, out1, out2, out3) {              \
  v4i32 s0_m, s1_m, s2_m, s3_m, s4_m, s5_m, s6_m, s7_m;          \
  v2i64 tp0_m, tp1_m, tp2_m, tp3_m;                              \
  v4i32 k0_m = __msa_fill_w((int32_t) const0);                   \
                                                                 \
  s0_m = __msa_fill_w((int32_t) const1);                         \
  k0_m = __msa_ilvev_w(s0_m, k0_m);                              \
                                                                 \
  ILVRL_W2_SW(-reg1_left, reg0_left, s1_m, s0_m);                \
  ILVRL_W2_SW(reg0_left, reg1_left, s3_m, s2_m);                 \
  ILVRL_W2_SW(-reg1_right, reg0_right, s5_m, s4_m);              \
  ILVRL_W2_SW(reg0_right, reg1_right, s7_m, s6_m);               \
                                                                 \
  DOTP_SW2_SD(s0_m, s1_m, k0_m, k0_m, tp0_m, tp1_m);             \
  DOTP_SW2_SD(s4_m, s5_m, k0_m, k0_m, tp2_m, tp3_m);             \
  tp0_m = __msa_srari_d(tp0_m, DCT_CONST_BITS);                  \
  tp1_m = __msa_srari_d(tp1_m, DCT_CONST_BITS);                  \
  tp2_m = __msa_srari_d(tp2_m, DCT_CONST_BITS);                  \
  tp3_m = __msa_srari_d(tp3_m, DCT_CONST_BITS);                  \
  out0 = __msa_pckev_w((v4i32)tp0_m, (v4i32)tp1_m);              \
  out1 = __msa_pckev_w((v4i32)tp2_m, (v4i32)tp3_m);              \
                                                                 \
  DOTP_SW2_SD(s2_m, s3_m, k0_m, k0_m, tp0_m, tp1_m);             \
  DOTP_SW2_SD(s6_m, s7_m, k0_m, k0_m, tp2_m, tp3_m);             \
  tp0_m = __msa_srari_d(tp0_m, DCT_CONST_BITS);                  \
  tp1_m = __msa_srari_d(tp1_m, DCT_CONST_BITS);                  \
  tp2_m = __msa_srari_d(tp2_m, DCT_CONST_BITS);                  \
  tp3_m = __msa_srari_d(tp3_m, DCT_CONST_BITS);                  \
  out2 = __msa_pckev_w((v4i32)tp0_m, (v4i32)tp1_m);              \
  out3 = __msa_pckev_w((v4i32)tp2_m, (v4i32)tp3_m);              \
}

355 356 357 358
void fdct8x16_1d_column(const int16_t *input, int16_t *tmp_ptr,
                        int32_t src_stride);
void fdct16x8_1d_row(int16_t *input, int16_t *output);
#endif  // VPX_DSP_MIPS_FWD_TXFM_MSA_H_