| OLD | NEW |
|---|
| (Empty) | |
| 1 /* |
| 2 * Copyright (c) 2015 The WebM project authors. All Rights Reserved. |
| 3 * |
| 4 * Use of this source code is governed by a BSD-style license |
| 5 * that can be found in the LICENSE file in the root of the source |
| 6 * tree. An additional intellectual property rights grant can be found |
| 7 * in the file PATENTS. All contributing project authors may |
| 8 * be found in the AUTHORS file in the root of the source tree. |
| 9 */ |
| 10 |
| 11 #include "./vp9_rtcd.h" |
| 12 #include "vp9/common/mips/msa/vp9_convolve_msa.h" |
| 13 |
| 14 static void common_vt_8t_4w_msa(const uint8_t *src, int32_t src_stride, |
| 15 uint8_t *dst, int32_t dst_stride, |
| 16 int8_t *filter, int32_t height) { |
| 17 uint32_t loop_cnt; |
| 18 v16i8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10; |
| 19 v16i8 src10_r, src32_r, src54_r, src76_r, src98_r; |
| 20 v16i8 src21_r, src43_r, src65_r, src87_r, src109_r; |
| 21 v16i8 src2110, src4332, src6554, src8776, src10998; |
| 22 v8i16 filt, out10, out32; |
| 23 v16i8 filt0, filt1, filt2, filt3; |
| 24 |
| 25 src -= (3 * src_stride); |
| 26 |
| 27 filt = LOAD_SH(filter); |
| 28 filt0 = (v16i8)__msa_splati_h(filt, 0); |
| 29 filt1 = (v16i8)__msa_splati_h(filt, 1); |
| 30 filt2 = (v16i8)__msa_splati_h(filt, 2); |
| 31 filt3 = (v16i8)__msa_splati_h(filt, 3); |
| 32 |
| 33 LOAD_7VECS_SB(src, src_stride, src0, src1, src2, src3, src4, src5, src6); |
| 34 src += (7 * src_stride); |
| 35 |
| 36 ILVR_B_6VECS_SB(src0, src2, src4, src1, src3, src5, |
| 37 src1, src3, src5, src2, src4, src6, |
| 38 src10_r, src32_r, src54_r, src21_r, src43_r, src65_r); |
| 39 |
| 40 ILVR_D_3VECS_SB(src2110, src21_r, src10_r, src4332, src43_r, src32_r, |
| 41 src6554, src65_r, src54_r); |
| 42 |
| 43 XORI_B_3VECS_SB(src2110, src4332, src6554, src2110, src4332, src6554, 128); |
| 44 |
| 45 for (loop_cnt = (height >> 2); loop_cnt--;) { |
| 46 LOAD_4VECS_SB(src, src_stride, src7, src8, src9, src10); |
| 47 src += (4 * src_stride); |
| 48 |
| 49 ILVR_B_4VECS_SB(src6, src7, src8, src9, src7, src8, src9, src10, |
| 50 src76_r, src87_r, src98_r, src109_r); |
| 51 |
| 52 ILVR_D_2VECS_SB(src8776, src87_r, src76_r, src10998, src109_r, src98_r); |
| 53 |
| 54 XORI_B_2VECS_SB(src8776, src10998, src8776, src10998, 128); |
| 55 |
| 56 out10 = FILT_8TAP_DPADD_S_H(src2110, src4332, src6554, src8776, |
| 57 filt0, filt1, filt2, filt3); |
| 58 out32 = FILT_8TAP_DPADD_S_H(src4332, src6554, src8776, src10998, |
| 59 filt0, filt1, filt2, filt3); |
| 60 |
| 61 out10 = SRARI_SATURATE_SIGNED_H(out10, FILTER_BITS, 7); |
| 62 out32 = SRARI_SATURATE_SIGNED_H(out32, FILTER_BITS, 7); |
| 63 |
| 64 PCKEV_2B_XORI128_STORE_4_BYTES_4(out10, out32, dst, dst_stride); |
| 65 dst += (4 * dst_stride); |
| 66 |
| 67 src2110 = src6554; |
| 68 src4332 = src8776; |
| 69 src6554 = src10998; |
| 70 |
| 71 src6 = src10; |
| 72 } |
| 73 } |
| 74 |
| 75 static void common_vt_8t_8w_msa(const uint8_t *src, int32_t src_stride, |
| 76 uint8_t *dst, int32_t dst_stride, |
| 77 int8_t *filter, int32_t height) { |
| 78 uint32_t loop_cnt; |
| 79 v16i8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10; |
| 80 v16i8 src10_r, src32_r, src54_r, src76_r, src98_r; |
| 81 v16i8 src21_r, src43_r, src65_r, src87_r, src109_r; |
| 82 v16i8 filt0, filt1, filt2, filt3; |
| 83 v8i16 filt, out0_r, out1_r, out2_r, out3_r; |
| 84 |
| 85 src -= (3 * src_stride); |
| 86 |
| 87 filt = LOAD_SH(filter); |
| 88 filt0 = (v16i8)__msa_splati_h(filt, 0); |
| 89 filt1 = (v16i8)__msa_splati_h(filt, 1); |
| 90 filt2 = (v16i8)__msa_splati_h(filt, 2); |
| 91 filt3 = (v16i8)__msa_splati_h(filt, 3); |
| 92 |
| 93 LOAD_7VECS_SB(src, src_stride, src0, src1, src2, src3, src4, src5, src6); |
| 94 src += (7 * src_stride); |
| 95 |
| 96 XORI_B_7VECS_SB(src0, src1, src2, src3, src4, src5, src6, |
| 97 src0, src1, src2, src3, src4, src5, src6, 128); |
| 98 |
| 99 ILVR_B_6VECS_SB(src0, src2, src4, src1, src3, src5, |
| 100 src1, src3, src5, src2, src4, src6, |
| 101 src10_r, src32_r, src54_r, src21_r, src43_r, src65_r); |
| 102 |
| 103 for (loop_cnt = (height >> 2); loop_cnt--;) { |
| 104 LOAD_4VECS_SB(src, src_stride, src7, src8, src9, src10); |
| 105 src += (4 * src_stride); |
| 106 |
| 107 XORI_B_4VECS_SB(src7, src8, src9, src10, src7, src8, src9, src10, 128); |
| 108 |
| 109 ILVR_B_4VECS_SB(src6, src7, src8, src9, src7, src8, src9, src10, |
| 110 src76_r, src87_r, src98_r, src109_r); |
| 111 |
| 112 out0_r = FILT_8TAP_DPADD_S_H(src10_r, src32_r, src54_r, src76_r, |
| 113 filt0, filt1, filt2, filt3); |
| 114 out1_r = FILT_8TAP_DPADD_S_H(src21_r, src43_r, src65_r, src87_r, |
| 115 filt0, filt1, filt2, filt3); |
| 116 out2_r = FILT_8TAP_DPADD_S_H(src32_r, src54_r, src76_r, src98_r, |
| 117 filt0, filt1, filt2, filt3); |
| 118 out3_r = FILT_8TAP_DPADD_S_H(src43_r, src65_r, src87_r, src109_r, |
| 119 filt0, filt1, filt2, filt3); |
| 120 |
| 121 out0_r = SRARI_SATURATE_SIGNED_H(out0_r, FILTER_BITS, 7); |
| 122 out1_r = SRARI_SATURATE_SIGNED_H(out1_r, FILTER_BITS, 7); |
| 123 out2_r = SRARI_SATURATE_SIGNED_H(out2_r, FILTER_BITS, 7); |
| 124 out3_r = SRARI_SATURATE_SIGNED_H(out3_r, FILTER_BITS, 7); |
| 125 |
| 126 PCKEV_B_4_XORI128_STORE_8_BYTES_4(out0_r, out1_r, out2_r, out3_r, |
| 127 dst, dst_stride); |
| 128 dst += (4 * dst_stride); |
| 129 |
| 130 src10_r = src54_r; |
| 131 src32_r = src76_r; |
| 132 src54_r = src98_r; |
| 133 src21_r = src65_r; |
| 134 src43_r = src87_r; |
| 135 src65_r = src109_r; |
| 136 |
| 137 src6 = src10; |
| 138 } |
| 139 } |
| 140 |
| 141 static void common_vt_8t_16w_mult_msa(const uint8_t *src, int32_t src_stride, |
| 142 uint8_t *dst, int32_t dst_stride, |
| 143 int8_t *filter, int32_t height, |
| 144 int32_t width) { |
| 145 const uint8_t *src_tmp; |
| 146 uint8_t *dst_tmp; |
| 147 uint32_t loop_cnt, cnt; |
| 148 v16i8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10; |
| 149 v16i8 filt0, filt1, filt2, filt3; |
| 150 v16i8 src10_r, src32_r, src54_r, src76_r, src98_r; |
| 151 v16i8 src21_r, src43_r, src65_r, src87_r, src109_r; |
| 152 v16i8 src10_l, src32_l, src54_l, src76_l, src98_l; |
| 153 v16i8 src21_l, src43_l, src65_l, src87_l, src109_l; |
| 154 v8i16 out0_r, out1_r, out2_r, out3_r, out0_l, out1_l, out2_l, out3_l; |
| 155 v8i16 filt; |
| 156 v16u8 tmp0, tmp1, tmp2, tmp3; |
| 157 |
| 158 src -= (3 * src_stride); |
| 159 |
| 160 filt = LOAD_SH(filter); |
| 161 filt0 = (v16i8)__msa_splati_h(filt, 0); |
| 162 filt1 = (v16i8)__msa_splati_h(filt, 1); |
| 163 filt2 = (v16i8)__msa_splati_h(filt, 2); |
| 164 filt3 = (v16i8)__msa_splati_h(filt, 3); |
| 165 |
| 166 for (cnt = (width >> 4); cnt--;) { |
| 167 src_tmp = src; |
| 168 dst_tmp = dst; |
| 169 |
| 170 LOAD_7VECS_SB(src_tmp, src_stride, |
| 171 src0, src1, src2, src3, src4, src5, src6); |
| 172 src_tmp += (7 * src_stride); |
| 173 |
| 174 XORI_B_7VECS_SB(src0, src1, src2, src3, src4, src5, src6, |
| 175 src0, src1, src2, src3, src4, src5, src6, 128); |
| 176 |
| 177 ILVR_B_6VECS_SB(src0, src2, src4, src1, src3, src5, |
| 178 src1, src3, src5, src2, src4, src6, |
| 179 src10_r, src32_r, src54_r, src21_r, src43_r, src65_r); |
| 180 |
| 181 ILVL_B_6VECS_SB(src0, src2, src4, src1, src3, src5, |
| 182 src1, src3, src5, src2, src4, src6, |
| 183 src10_l, src32_l, src54_l, src21_l, src43_l, src65_l); |
| 184 |
| 185 for (loop_cnt = (height >> 2); loop_cnt--;) { |
| 186 LOAD_4VECS_SB(src_tmp, src_stride, src7, src8, src9, src10); |
| 187 src_tmp += (4 * src_stride); |
| 188 |
| 189 XORI_B_4VECS_SB(src7, src8, src9, src10, src7, src8, src9, src10, 128); |
| 190 |
| 191 ILVR_B_4VECS_SB(src6, src7, src8, src9, src7, src8, src9, src10, |
| 192 src76_r, src87_r, src98_r, src109_r); |
| 193 |
| 194 ILVL_B_4VECS_SB(src6, src7, src8, src9, src7, src8, src9, src10, |
| 195 src76_l, src87_l, src98_l, src109_l); |
| 196 |
| 197 out0_r = FILT_8TAP_DPADD_S_H(src10_r, src32_r, src54_r, src76_r, |
| 198 filt0, filt1, filt2, filt3); |
| 199 out1_r = FILT_8TAP_DPADD_S_H(src21_r, src43_r, src65_r, src87_r, |
| 200 filt0, filt1, filt2, filt3); |
| 201 out2_r = FILT_8TAP_DPADD_S_H(src32_r, src54_r, src76_r, src98_r, |
| 202 filt0, filt1, filt2, filt3); |
| 203 out3_r = FILT_8TAP_DPADD_S_H(src43_r, src65_r, src87_r, src109_r, |
| 204 filt0, filt1, filt2, filt3); |
| 205 |
| 206 out0_l = FILT_8TAP_DPADD_S_H(src10_l, src32_l, src54_l, src76_l, |
| 207 filt0, filt1, filt2, filt3); |
| 208 out1_l = FILT_8TAP_DPADD_S_H(src21_l, src43_l, src65_l, src87_l, |
| 209 filt0, filt1, filt2, filt3); |
| 210 out2_l = FILT_8TAP_DPADD_S_H(src32_l, src54_l, src76_l, src98_l, |
| 211 filt0, filt1, filt2, filt3); |
| 212 out3_l = FILT_8TAP_DPADD_S_H(src43_l, src65_l, src87_l, src109_l, |
| 213 filt0, filt1, filt2, filt3); |
| 214 |
| 215 out0_r = SRARI_SATURATE_SIGNED_H(out0_r, FILTER_BITS, 7); |
| 216 out1_r = SRARI_SATURATE_SIGNED_H(out1_r, FILTER_BITS, 7); |
| 217 out2_r = SRARI_SATURATE_SIGNED_H(out2_r, FILTER_BITS, 7); |
| 218 out3_r = SRARI_SATURATE_SIGNED_H(out3_r, FILTER_BITS, 7); |
| 219 out0_l = SRARI_SATURATE_SIGNED_H(out0_l, FILTER_BITS, 7); |
| 220 out1_l = SRARI_SATURATE_SIGNED_H(out1_l, FILTER_BITS, 7); |
| 221 out2_l = SRARI_SATURATE_SIGNED_H(out2_l, FILTER_BITS, 7); |
| 222 out3_l = SRARI_SATURATE_SIGNED_H(out3_l, FILTER_BITS, 7); |
| 223 |
| 224 out0_r = (v8i16)__msa_pckev_b((v16i8)out0_l, (v16i8)out0_r); |
| 225 out1_r = (v8i16)__msa_pckev_b((v16i8)out1_l, (v16i8)out1_r); |
| 226 out2_r = (v8i16)__msa_pckev_b((v16i8)out2_l, (v16i8)out2_r); |
| 227 out3_r = (v8i16)__msa_pckev_b((v16i8)out3_l, (v16i8)out3_r); |
| 228 |
| 229 XORI_B_4VECS_UB(out0_r, out1_r, out2_r, out3_r, |
| 230 tmp0, tmp1, tmp2, tmp3, 128); |
| 231 |
| 232 STORE_4VECS_UB(dst_tmp, dst_stride, tmp0, tmp1, tmp2, tmp3); |
| 233 dst_tmp += (4 * dst_stride); |
| 234 |
| 235 src10_r = src54_r; |
| 236 src32_r = src76_r; |
| 237 src54_r = src98_r; |
| 238 src21_r = src65_r; |
| 239 src43_r = src87_r; |
| 240 src65_r = src109_r; |
| 241 |
| 242 src10_l = src54_l; |
| 243 src32_l = src76_l; |
| 244 src54_l = src98_l; |
| 245 src21_l = src65_l; |
| 246 src43_l = src87_l; |
| 247 src65_l = src109_l; |
| 248 |
| 249 src6 = src10; |
| 250 } |
| 251 |
| 252 src += 16; |
| 253 dst += 16; |
| 254 } |
| 255 } |
| 256 |
| 257 static void common_vt_8t_16w_msa(const uint8_t *src, int32_t src_stride, |
| 258 uint8_t *dst, int32_t dst_stride, |
| 259 int8_t *filter, int32_t height) { |
| 260 common_vt_8t_16w_mult_msa(src, src_stride, dst, dst_stride, |
| 261 filter, height, 16); |
| 262 } |
| 263 |
| 264 static void common_vt_8t_32w_msa(const uint8_t *src, int32_t src_stride, |
| 265 uint8_t *dst, int32_t dst_stride, |
| 266 int8_t *filter, int32_t height) { |
| 267 common_vt_8t_16w_mult_msa(src, src_stride, dst, dst_stride, |
| 268 filter, height, 32); |
| 269 } |
| 270 |
| 271 static void common_vt_8t_64w_msa(const uint8_t *src, int32_t src_stride, |
| 272 uint8_t *dst, int32_t dst_stride, |
| 273 int8_t *filter, int32_t height) { |
| 274 common_vt_8t_16w_mult_msa(src, src_stride, dst, dst_stride, |
| 275 filter, height, 64); |
| 276 } |
| 277 |
| 278 static void common_vt_2t_4x4_msa(const uint8_t *src, int32_t src_stride, |
| 279 uint8_t *dst, int32_t dst_stride, |
| 280 int8_t *filter) { |
| 281 uint32_t out0, out1, out2, out3; |
| 282 v16i8 src0, src1, src2, src3, src4; |
| 283 v16i8 src10_r, src32_r, src21_r, src43_r, src2110, src4332; |
| 284 v16i8 filt0; |
| 285 v8u16 filt; |
| 286 |
| 287 filt = LOAD_UH(filter); |
| 288 filt0 = (v16i8)__msa_splati_h((v8i16)filt, 0); |
| 289 |
| 290 LOAD_5VECS_SB(src, src_stride, src0, src1, src2, src3, src4); |
| 291 src += (5 * src_stride); |
| 292 |
| 293 ILVR_B_4VECS_SB(src0, src1, src2, src3, src1, src2, src3, src4, |
| 294 src10_r, src21_r, src32_r, src43_r); |
| 295 |
| 296 ILVR_D_2VECS_SB(src2110, src21_r, src10_r, src4332, src43_r, src32_r); |
| 297 |
| 298 src2110 = (v16i8)__msa_dotp_u_h((v16u8)src2110, (v16u8)filt0); |
| 299 src4332 = (v16i8)__msa_dotp_u_h((v16u8)src4332, (v16u8)filt0); |
| 300 |
| 301 src2110 = (v16i8)SRARI_SATURATE_UNSIGNED_H(src2110, FILTER_BITS, 7); |
| 302 src4332 = (v16i8)SRARI_SATURATE_UNSIGNED_H(src4332, FILTER_BITS, 7); |
| 303 |
| 304 src2110 = (v16i8)__msa_pckev_b((v16i8)src4332, (v16i8)src2110); |
| 305 |
| 306 out0 = __msa_copy_u_w((v4i32)src2110, 0); |
| 307 out1 = __msa_copy_u_w((v4i32)src2110, 1); |
| 308 out2 = __msa_copy_u_w((v4i32)src2110, 2); |
| 309 out3 = __msa_copy_u_w((v4i32)src2110, 3); |
| 310 |
| 311 STORE_WORD(dst, out0); |
| 312 dst += dst_stride; |
| 313 STORE_WORD(dst, out1); |
| 314 dst += dst_stride; |
| 315 STORE_WORD(dst, out2); |
| 316 dst += dst_stride; |
| 317 STORE_WORD(dst, out3); |
| 318 } |
| 319 |
| 320 static void common_vt_2t_4x8_msa(const uint8_t *src, int32_t src_stride, |
| 321 uint8_t *dst, int32_t dst_stride, |
| 322 int8_t *filter) { |
| 323 uint32_t out0, out1, out2, out3, out4, out5, out6, out7; |
| 324 v16i8 src0, src1, src2, src3, src4, src5, src6, src7, src8; |
| 325 v16i8 src10_r, src32_r, src54_r, src76_r, src21_r, src43_r; |
| 326 v16i8 src65_r, src87_r, src2110, src4332, src6554, src8776; |
| 327 v16i8 filt0; |
| 328 v8u16 filt; |
| 329 |
| 330 filt = LOAD_UH(filter); |
| 331 filt0 = (v16i8)__msa_splati_h((v8i16)filt, 0); |
| 332 |
| 333 LOAD_8VECS_SB(src, src_stride, |
| 334 src0, src1, src2, src3, src4, src5, src6, src7); |
| 335 src += (8 * src_stride); |
| 336 |
| 337 src8 = LOAD_SB(src); |
| 338 src += src_stride; |
| 339 |
| 340 ILVR_B_8VECS_SB(src0, src1, src2, src3, src4, src5, src6, src7, |
| 341 src1, src2, src3, src4, src5, src6, src7, src8, |
| 342 src10_r, src21_r, src32_r, src43_r, |
| 343 src54_r, src65_r, src76_r, src87_r); |
| 344 |
| 345 ILVR_D_4VECS_SB(src2110, src21_r, src10_r, src4332, src43_r, src32_r, |
| 346 src6554, src65_r, src54_r, src8776, src87_r, src76_r); |
| 347 |
| 348 src2110 = (v16i8)__msa_dotp_u_h((v16u8)src2110, (v16u8)filt0); |
| 349 src4332 = (v16i8)__msa_dotp_u_h((v16u8)src4332, (v16u8)filt0); |
| 350 src6554 = (v16i8)__msa_dotp_u_h((v16u8)src6554, (v16u8)filt0); |
| 351 src8776 = (v16i8)__msa_dotp_u_h((v16u8)src8776, (v16u8)filt0); |
| 352 |
| 353 src2110 = (v16i8)SRARI_SATURATE_UNSIGNED_H(src2110, FILTER_BITS, 7); |
| 354 src4332 = (v16i8)SRARI_SATURATE_UNSIGNED_H(src4332, FILTER_BITS, 7); |
| 355 src6554 = (v16i8)SRARI_SATURATE_UNSIGNED_H(src6554, FILTER_BITS, 7); |
| 356 src8776 = (v16i8)SRARI_SATURATE_UNSIGNED_H(src8776, FILTER_BITS, 7); |
| 357 |
| 358 src2110 = (v16i8)__msa_pckev_b((v16i8)src4332, (v16i8)src2110); |
| 359 src4332 = (v16i8)__msa_pckev_b((v16i8)src8776, (v16i8)src6554); |
| 360 |
| 361 out0 = __msa_copy_u_w((v4i32)src2110, 0); |
| 362 out1 = __msa_copy_u_w((v4i32)src2110, 1); |
| 363 out2 = __msa_copy_u_w((v4i32)src2110, 2); |
| 364 out3 = __msa_copy_u_w((v4i32)src2110, 3); |
| 365 out4 = __msa_copy_u_w((v4i32)src4332, 0); |
| 366 out5 = __msa_copy_u_w((v4i32)src4332, 1); |
| 367 out6 = __msa_copy_u_w((v4i32)src4332, 2); |
| 368 out7 = __msa_copy_u_w((v4i32)src4332, 3); |
| 369 |
| 370 STORE_WORD(dst, out0); |
| 371 dst += dst_stride; |
| 372 STORE_WORD(dst, out1); |
| 373 dst += dst_stride; |
| 374 STORE_WORD(dst, out2); |
| 375 dst += dst_stride; |
| 376 STORE_WORD(dst, out3); |
| 377 dst += dst_stride; |
| 378 STORE_WORD(dst, out4); |
| 379 dst += dst_stride; |
| 380 STORE_WORD(dst, out5); |
| 381 dst += dst_stride; |
| 382 STORE_WORD(dst, out6); |
| 383 dst += dst_stride; |
| 384 STORE_WORD(dst, out7); |
| 385 } |
| 386 |
| 387 static void common_vt_2t_4w_msa(const uint8_t *src, int32_t src_stride, |
| 388 uint8_t *dst, int32_t dst_stride, |
| 389 int8_t *filter, int32_t height) { |
| 390 if (4 == height) { |
| 391 common_vt_2t_4x4_msa(src, src_stride, dst, dst_stride, filter); |
| 392 } else if (8 == height) { |
| 393 common_vt_2t_4x8_msa(src, src_stride, dst, dst_stride, filter); |
| 394 } |
| 395 } |
| 396 |
| 397 static void common_vt_2t_8x4_msa(const uint8_t *src, int32_t src_stride, |
| 398 uint8_t *dst, int32_t dst_stride, |
| 399 int8_t *filter) { |
| 400 v16u8 src0, src1, src2, src3, src4; |
| 401 v16u8 vec0, vec1, vec2, vec3, filt0; |
| 402 v8u16 tmp0, tmp1, tmp2, tmp3; |
| 403 v8u16 filt; |
| 404 |
| 405 /* rearranging filter_y */ |
| 406 filt = LOAD_UH(filter); |
| 407 filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0); |
| 408 |
| 409 LOAD_5VECS_UB(src, src_stride, src0, src1, src2, src3, src4); |
| 410 |
| 411 ILVR_B_2VECS_UB(src0, src1, src1, src2, vec0, vec1); |
| 412 ILVR_B_2VECS_UB(src2, src3, src3, src4, vec2, vec3); |
| 413 |
| 414 /* filter calc */ |
| 415 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 416 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 417 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 418 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 419 |
| 420 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 421 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 422 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 423 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 424 |
| 425 PCKEV_B_STORE_8_BYTES_4(tmp0, tmp1, tmp2, tmp3, dst, dst_stride); |
| 426 } |
| 427 |
| 428 static void common_vt_2t_8x8mult_msa(const uint8_t *src, int32_t src_stride, |
| 429 uint8_t *dst, int32_t dst_stride, |
| 430 int8_t *filter, int32_t height) { |
| 431 uint32_t loop_cnt; |
| 432 v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8; |
| 433 v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7, filt0; |
| 434 v8u16 tmp0, tmp1, tmp2, tmp3; |
| 435 v8u16 filt; |
| 436 |
| 437 /* rearranging filter_y */ |
| 438 filt = LOAD_UH(filter); |
| 439 filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0); |
| 440 |
| 441 src0 = LOAD_UB(src); |
| 442 src += src_stride; |
| 443 |
| 444 for (loop_cnt = (height >> 3); loop_cnt--;) { |
| 445 LOAD_8VECS_UB(src, src_stride, |
| 446 src1, src2, src3, src4, src5, src6, src7, src8); |
| 447 src += (8 * src_stride); |
| 448 |
| 449 ILVR_B_4VECS_UB(src0, src1, src2, src3, src1, src2, src3, src4, |
| 450 vec0, vec1, vec2, vec3); |
| 451 |
| 452 ILVR_B_4VECS_UB(src4, src5, src6, src7, src5, src6, src7, src8, |
| 453 vec4, vec5, vec6, vec7); |
| 454 |
| 455 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 456 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 457 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 458 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 459 |
| 460 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 461 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 462 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 463 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 464 |
| 465 PCKEV_B_STORE_8_BYTES_4(tmp0, tmp1, tmp2, tmp3, dst, dst_stride); |
| 466 dst += (4 * dst_stride); |
| 467 |
| 468 tmp0 = __msa_dotp_u_h(vec4, filt0); |
| 469 tmp1 = __msa_dotp_u_h(vec5, filt0); |
| 470 tmp2 = __msa_dotp_u_h(vec6, filt0); |
| 471 tmp3 = __msa_dotp_u_h(vec7, filt0); |
| 472 |
| 473 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 474 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 475 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 476 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 477 |
| 478 PCKEV_B_STORE_8_BYTES_4(tmp0, tmp1, tmp2, tmp3, dst, dst_stride); |
| 479 dst += (4 * dst_stride); |
| 480 |
| 481 src0 = src8; |
| 482 } |
| 483 } |
| 484 |
| 485 static void common_vt_2t_8w_msa(const uint8_t *src, int32_t src_stride, |
| 486 uint8_t *dst, int32_t dst_stride, |
| 487 int8_t *filter, int32_t height) { |
| 488 if (4 == height) { |
| 489 common_vt_2t_8x4_msa(src, src_stride, dst, dst_stride, filter); |
| 490 } else { |
| 491 common_vt_2t_8x8mult_msa(src, src_stride, dst, dst_stride, filter, height); |
| 492 } |
| 493 } |
| 494 |
| 495 static void common_vt_2t_16w_msa(const uint8_t *src, int32_t src_stride, |
| 496 uint8_t *dst, int32_t dst_stride, |
| 497 int8_t *filter, int32_t height) { |
| 498 uint32_t loop_cnt; |
| 499 v16u8 src0, src1, src2, src3, src4; |
| 500 v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7, filt0; |
| 501 v8u16 tmp0, tmp1, tmp2, tmp3; |
| 502 v8u16 filt; |
| 503 |
| 504 /* rearranging filter_y */ |
| 505 filt = LOAD_UH(filter); |
| 506 filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0); |
| 507 |
| 508 src0 = LOAD_UB(src); |
| 509 src += src_stride; |
| 510 |
| 511 for (loop_cnt = (height >> 2); loop_cnt--;) { |
| 512 LOAD_4VECS_UB(src, src_stride, src1, src2, src3, src4); |
| 513 src += (4 * src_stride); |
| 514 |
| 515 ILV_B_LRLR_UB(src0, src1, src1, src2, vec1, vec0, vec3, vec2); |
| 516 |
| 517 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 518 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 519 |
| 520 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 521 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 522 |
| 523 PCKEV_B_STORE_VEC(tmp1, tmp0, dst); |
| 524 dst += dst_stride; |
| 525 |
| 526 ILV_B_LRLR_UB(src2, src3, src3, src4, vec5, vec4, vec7, vec6); |
| 527 |
| 528 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 529 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 530 |
| 531 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 532 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 533 |
| 534 PCKEV_B_STORE_VEC(tmp3, tmp2, dst); |
| 535 dst += dst_stride; |
| 536 |
| 537 tmp0 = __msa_dotp_u_h(vec4, filt0); |
| 538 tmp1 = __msa_dotp_u_h(vec5, filt0); |
| 539 |
| 540 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 541 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 542 |
| 543 PCKEV_B_STORE_VEC(tmp1, tmp0, dst); |
| 544 dst += dst_stride; |
| 545 |
| 546 tmp2 = __msa_dotp_u_h(vec6, filt0); |
| 547 tmp3 = __msa_dotp_u_h(vec7, filt0); |
| 548 |
| 549 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 550 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 551 |
| 552 PCKEV_B_STORE_VEC(tmp3, tmp2, dst); |
| 553 dst += dst_stride; |
| 554 |
| 555 src0 = src4; |
| 556 } |
| 557 } |
| 558 |
| 559 static void common_vt_2t_32w_msa(const uint8_t *src, int32_t src_stride, |
| 560 uint8_t *dst, int32_t dst_stride, |
| 561 int8_t *filter, int32_t height) { |
| 562 uint32_t loop_cnt; |
| 563 v16u8 src0, src1, src2, src3, src4, src5, src6, src7, src8, src9; |
| 564 v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7, filt0; |
| 565 v8u16 tmp0, tmp1, tmp2, tmp3; |
| 566 v8u16 filt; |
| 567 |
| 568 /* rearranging filter_y */ |
| 569 filt = LOAD_UH(filter); |
| 570 filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0); |
| 571 |
| 572 src0 = LOAD_UB(src); |
| 573 src5 = LOAD_UB(src + 16); |
| 574 src += src_stride; |
| 575 |
| 576 for (loop_cnt = (height >> 2); loop_cnt--;) { |
| 577 LOAD_4VECS_UB(src, src_stride, src1, src2, src3, src4); |
| 578 |
| 579 ILV_B_LRLR_UB(src0, src1, src1, src2, vec1, vec0, vec3, vec2); |
| 580 |
| 581 LOAD_4VECS_UB(src + 16, src_stride, src6, src7, src8, src9); |
| 582 src += (4 * src_stride); |
| 583 |
| 584 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 585 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 586 |
| 587 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 588 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 589 |
| 590 PCKEV_B_STORE_VEC(tmp1, tmp0, dst); |
| 591 |
| 592 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 593 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 594 |
| 595 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 596 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 597 |
| 598 PCKEV_B_STORE_VEC(tmp3, tmp2, dst + dst_stride); |
| 599 |
| 600 ILV_B_LRLR_UB(src2, src3, src3, src4, vec5, vec4, vec7, vec6); |
| 601 |
| 602 tmp0 = __msa_dotp_u_h(vec4, filt0); |
| 603 tmp1 = __msa_dotp_u_h(vec5, filt0); |
| 604 |
| 605 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 606 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 607 |
| 608 PCKEV_B_STORE_VEC(tmp1, tmp0, dst + 2 * dst_stride); |
| 609 |
| 610 tmp2 = __msa_dotp_u_h(vec6, filt0); |
| 611 tmp3 = __msa_dotp_u_h(vec7, filt0); |
| 612 |
| 613 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 614 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 615 |
| 616 PCKEV_B_STORE_VEC(tmp3, tmp2, dst + 3 * dst_stride); |
| 617 |
| 618 ILV_B_LRLR_UB(src5, src6, src6, src7, vec1, vec0, vec3, vec2); |
| 619 |
| 620 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 621 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 622 |
| 623 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 624 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 625 |
| 626 PCKEV_B_STORE_VEC(tmp1, tmp0, dst + 16); |
| 627 |
| 628 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 629 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 630 |
| 631 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 632 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 633 |
| 634 PCKEV_B_STORE_VEC(tmp3, tmp2, dst + 16 + dst_stride); |
| 635 |
| 636 ILV_B_LRLR_UB(src7, src8, src8, src9, vec5, vec4, vec7, vec6); |
| 637 |
| 638 tmp0 = __msa_dotp_u_h(vec4, filt0); |
| 639 tmp1 = __msa_dotp_u_h(vec5, filt0); |
| 640 |
| 641 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 642 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 643 |
| 644 PCKEV_B_STORE_VEC(tmp1, tmp0, dst + 16 + 2 * dst_stride); |
| 645 |
| 646 tmp2 = __msa_dotp_u_h(vec6, filt0); |
| 647 tmp3 = __msa_dotp_u_h(vec7, filt0); |
| 648 |
| 649 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 650 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 651 |
| 652 PCKEV_B_STORE_VEC(tmp3, tmp2, dst + 16 + 3 * dst_stride); |
| 653 dst += (4 * dst_stride); |
| 654 |
| 655 src0 = src4; |
| 656 src5 = src9; |
| 657 } |
| 658 } |
| 659 |
| 660 static void common_vt_2t_64w_msa(const uint8_t *src, int32_t src_stride, |
| 661 uint8_t *dst, int32_t dst_stride, |
| 662 int8_t *filter, int32_t height) { |
| 663 uint32_t loop_cnt; |
| 664 v16u8 src0, src1, src2, src3, src4, src5, src6, src7; |
| 665 v16u8 src8, src9, src10, src11; |
| 666 v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7, filt0; |
| 667 v8u16 tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7; |
| 668 v8u16 filt; |
| 669 |
| 670 /* rearranging filter_y */ |
| 671 filt = LOAD_UH(filter); |
| 672 filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0); |
| 673 |
| 674 LOAD_4VECS_UB(src, 16, src0, src3, src6, src9); |
| 675 src += src_stride; |
| 676 |
| 677 for (loop_cnt = (height >> 1); loop_cnt--;) { |
| 678 LOAD_2VECS_UB(src, src_stride, src1, src2); |
| 679 LOAD_2VECS_UB(src + 16, src_stride, src4, src5); |
| 680 LOAD_2VECS_UB(src + 32, src_stride, src7, src8); |
| 681 LOAD_2VECS_UB(src + 48, src_stride, src10, src11); |
| 682 src += (2 * src_stride); |
| 683 |
| 684 ILV_B_LRLR_UB(src0, src1, src1, src2, vec1, vec0, vec3, vec2); |
| 685 |
| 686 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 687 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 688 |
| 689 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 690 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 691 |
| 692 PCKEV_B_STORE_VEC(tmp1, tmp0, dst); |
| 693 |
| 694 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 695 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 696 |
| 697 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 698 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 699 |
| 700 PCKEV_B_STORE_VEC(tmp3, tmp2, dst + dst_stride); |
| 701 |
| 702 ILV_B_LRLR_UB(src3, src4, src4, src5, vec5, vec4, vec7, vec6); |
| 703 |
| 704 tmp4 = __msa_dotp_u_h(vec4, filt0); |
| 705 tmp5 = __msa_dotp_u_h(vec5, filt0); |
| 706 |
| 707 tmp4 = SRARI_SATURATE_UNSIGNED_H(tmp4, FILTER_BITS, 7); |
| 708 tmp5 = SRARI_SATURATE_UNSIGNED_H(tmp5, FILTER_BITS, 7); |
| 709 |
| 710 PCKEV_B_STORE_VEC(tmp5, tmp4, dst + 16); |
| 711 |
| 712 tmp6 = __msa_dotp_u_h(vec6, filt0); |
| 713 tmp7 = __msa_dotp_u_h(vec7, filt0); |
| 714 |
| 715 tmp6 = SRARI_SATURATE_UNSIGNED_H(tmp6, FILTER_BITS, 7); |
| 716 tmp7 = SRARI_SATURATE_UNSIGNED_H(tmp7, FILTER_BITS, 7); |
| 717 |
| 718 PCKEV_B_STORE_VEC(tmp7, tmp6, dst + 16 + dst_stride); |
| 719 |
| 720 ILV_B_LRLR_UB(src6, src7, src7, src8, vec1, vec0, vec3, vec2); |
| 721 |
| 722 tmp0 = __msa_dotp_u_h(vec0, filt0); |
| 723 tmp1 = __msa_dotp_u_h(vec1, filt0); |
| 724 |
| 725 tmp0 = SRARI_SATURATE_UNSIGNED_H(tmp0, FILTER_BITS, 7); |
| 726 tmp1 = SRARI_SATURATE_UNSIGNED_H(tmp1, FILTER_BITS, 7); |
| 727 |
| 728 PCKEV_B_STORE_VEC(tmp1, tmp0, dst + 32); |
| 729 |
| 730 tmp2 = __msa_dotp_u_h(vec2, filt0); |
| 731 tmp3 = __msa_dotp_u_h(vec3, filt0); |
| 732 |
| 733 tmp2 = SRARI_SATURATE_UNSIGNED_H(tmp2, FILTER_BITS, 7); |
| 734 tmp3 = SRARI_SATURATE_UNSIGNED_H(tmp3, FILTER_BITS, 7); |
| 735 |
| 736 PCKEV_B_STORE_VEC(tmp3, tmp2, dst + 32 + dst_stride); |
| 737 |
| 738 ILV_B_LRLR_UB(src9, src10, src10, src11, vec5, vec4, vec7, vec6); |
| 739 |
| 740 tmp4 = __msa_dotp_u_h(vec4, filt0); |
| 741 tmp5 = __msa_dotp_u_h(vec5, filt0); |
| 742 |
| 743 tmp4 = SRARI_SATURATE_UNSIGNED_H(tmp4, FILTER_BITS, 7); |
| 744 tmp5 = SRARI_SATURATE_UNSIGNED_H(tmp5, FILTER_BITS, 7); |
| 745 |
| 746 PCKEV_B_STORE_VEC(tmp5, tmp4, dst + 48); |
| 747 |
| 748 tmp6 = __msa_dotp_u_h(vec6, filt0); |
| 749 tmp7 = __msa_dotp_u_h(vec7, filt0); |
| 750 |
| 751 tmp6 = SRARI_SATURATE_UNSIGNED_H(tmp6, FILTER_BITS, 7); |
| 752 tmp7 = SRARI_SATURATE_UNSIGNED_H(tmp7, FILTER_BITS, 7); |
| 753 |
| 754 PCKEV_B_STORE_VEC(tmp7, tmp6, dst + 48 + dst_stride); |
| 755 dst += (2 * dst_stride); |
| 756 |
| 757 src0 = src2; |
| 758 src3 = src5; |
| 759 src6 = src8; |
| 760 src9 = src11; |
| 761 } |
| 762 } |
| 763 |
| 764 void vp9_convolve8_vert_msa(const uint8_t *src, ptrdiff_t src_stride, |
| 765 uint8_t *dst, ptrdiff_t dst_stride, |
| 766 const int16_t *filter_x, int x_step_q4, |
| 767 const int16_t *filter_y, int y_step_q4, |
| 768 int w, int h) { |
| 769 int8_t cnt, filt_ver[8]; |
| 770 |
| 771 if (16 != y_step_q4) { |
| 772 vp9_convolve8_vert_c(src, src_stride, dst, dst_stride, |
| 773 filter_x, x_step_q4, filter_y, y_step_q4, |
| 774 w, h); |
| 775 return; |
| 776 } |
| 777 |
| 778 if (((const int32_t *)filter_y)[1] == 0x800000) { |
| 779 vp9_convolve_copy(src, src_stride, dst, dst_stride, |
| 780 filter_x, x_step_q4, filter_y, y_step_q4, |
| 781 w, h); |
| 782 return; |
| 783 } |
| 784 |
| 785 for (cnt = 8; cnt--;) { |
| 786 filt_ver[cnt] = filter_y[cnt]; |
| 787 } |
| 788 |
| 789 if (((const int32_t *)filter_y)[0] == 0) { |
| 790 switch (w) { |
| 791 case 4: |
| 792 common_vt_2t_4w_msa(src, (int32_t)src_stride, |
| 793 dst, (int32_t)dst_stride, |
| 794 &filt_ver[3], h); |
| 795 break; |
| 796 case 8: |
| 797 common_vt_2t_8w_msa(src, (int32_t)src_stride, |
| 798 dst, (int32_t)dst_stride, |
| 799 &filt_ver[3], h); |
| 800 break; |
| 801 case 16: |
| 802 common_vt_2t_16w_msa(src, (int32_t)src_stride, |
| 803 dst, (int32_t)dst_stride, |
| 804 &filt_ver[3], h); |
| 805 break; |
| 806 case 32: |
| 807 common_vt_2t_32w_msa(src, (int32_t)src_stride, |
| 808 dst, (int32_t)dst_stride, |
| 809 &filt_ver[3], h); |
| 810 break; |
| 811 case 64: |
| 812 common_vt_2t_64w_msa(src, (int32_t)src_stride, |
| 813 dst, (int32_t)dst_stride, |
| 814 &filt_ver[3], h); |
| 815 break; |
| 816 default: |
| 817 vp9_convolve8_vert_c(src, src_stride, dst, dst_stride, |
| 818 filter_x, x_step_q4, filter_y, y_step_q4, |
| 819 w, h); |
| 820 break; |
| 821 } |
| 822 } else { |
| 823 switch (w) { |
| 824 case 4: |
| 825 common_vt_8t_4w_msa(src, (int32_t)src_stride, |
| 826 dst, (int32_t)dst_stride, |
| 827 filt_ver, h); |
| 828 break; |
| 829 case 8: |
| 830 common_vt_8t_8w_msa(src, (int32_t)src_stride, |
| 831 dst, (int32_t)dst_stride, |
| 832 filt_ver, h); |
| 833 break; |
| 834 case 16: |
| 835 common_vt_8t_16w_msa(src, (int32_t)src_stride, |
| 836 dst, (int32_t)dst_stride, |
| 837 filt_ver, h); |
| 838 break; |
| 839 case 32: |
| 840 common_vt_8t_32w_msa(src, (int32_t)src_stride, |
| 841 dst, (int32_t)dst_stride, |
| 842 filt_ver, h); |
| 843 break; |
| 844 case 64: |
| 845 common_vt_8t_64w_msa(src, (int32_t)src_stride, |
| 846 dst, (int32_t)dst_stride, |
| 847 filt_ver, h); |
| 848 break; |
| 849 default: |
| 850 vp9_convolve8_vert_c(src, src_stride, dst, dst_stride, |
| 851 filter_x, x_step_q4, filter_y, y_step_q4, |
| 852 w, h); |
| 853 break; |
| 854 } |
| 855 } |
| 856 } |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 1124333011:
libvpx: Pull from upstream (Closed)
Base URL: https://chromium.googlesource.com/chromium/deps/libvpx.git@master