Libav 0.7.1
|
00001 /* 00002 * AltiVec-enhanced yuv2yuvX 00003 * 00004 * Copyright (C) 2004 Romain Dolbeau <romain@dolbeau.org> 00005 * based on the equivalent C code in swscale.c 00006 * 00007 * This file is part of Libav. 00008 * 00009 * Libav is free software; you can redistribute it and/or 00010 * modify it under the terms of the GNU Lesser General Public 00011 * License as published by the Free Software Foundation; either 00012 * version 2.1 of the License, or (at your option) any later version. 00013 * 00014 * Libav is distributed in the hope that it will be useful, 00015 * but WITHOUT ANY WARRANTY; without even the implied warranty of 00016 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 00017 * Lesser General Public License for more details. 00018 * 00019 * You should have received a copy of the GNU Lesser General Public 00020 * License along with Libav; if not, write to the Free Software 00021 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA 00022 */ 00023 00024 #include <inttypes.h> 00025 #include "config.h" 00026 #include "libswscale/swscale.h" 00027 #include "libswscale/swscale_internal.h" 00028 #include "libavutil/cpu.h" 00029 #include "yuv2rgb_altivec.h" 00030 00031 #define vzero vec_splat_s32(0) 00032 00033 static inline void 00034 altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW) 00035 { 00036 register int i; 00037 vector unsigned int altivec_vectorShiftInt19 = 00038 vec_add(vec_splat_u32(10), vec_splat_u32(9)); 00039 if ((uintptr_t)dest % 16) { 00040 /* badly aligned store, we force store alignment */ 00041 /* and will handle load misalignment on val w/ vec_perm */ 00042 vector unsigned char perm1; 00043 vector signed int v1; 00044 for (i = 0 ; (i < dstW) && 00045 (((uintptr_t)dest + i) % 16) ; i++) { 00046 int t = val[i] >> 19; 00047 dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t); 00048 } 00049 perm1 = vec_lvsl(i << 2, val); 00050 v1 = vec_ld(i << 2, val); 00051 for ( ; i < (dstW - 15); i+=16) { 00052 int offset = i << 2; 00053 vector signed int v2 = vec_ld(offset + 16, val); 00054 vector signed int v3 = vec_ld(offset + 32, val); 00055 vector signed int v4 = vec_ld(offset + 48, val); 00056 vector signed int v5 = vec_ld(offset + 64, val); 00057 vector signed int v12 = vec_perm(v1, v2, perm1); 00058 vector signed int v23 = vec_perm(v2, v3, perm1); 00059 vector signed int v34 = vec_perm(v3, v4, perm1); 00060 vector signed int v45 = vec_perm(v4, v5, perm1); 00061 00062 vector signed int vA = vec_sra(v12, altivec_vectorShiftInt19); 00063 vector signed int vB = vec_sra(v23, altivec_vectorShiftInt19); 00064 vector signed int vC = vec_sra(v34, altivec_vectorShiftInt19); 00065 vector signed int vD = vec_sra(v45, altivec_vectorShiftInt19); 00066 vector unsigned short vs1 = vec_packsu(vA, vB); 00067 vector unsigned short vs2 = vec_packsu(vC, vD); 00068 vector unsigned char vf = vec_packsu(vs1, vs2); 00069 vec_st(vf, i, dest); 00070 v1 = v5; 00071 } 00072 } else { // dest is properly aligned, great 00073 for (i = 0; i < (dstW - 15); i+=16) { 00074 int offset = i << 2; 00075 vector signed int v1 = vec_ld(offset, val); 00076 vector signed int v2 = vec_ld(offset + 16, val); 00077 vector signed int v3 = vec_ld(offset + 32, val); 00078 vector signed int v4 = vec_ld(offset + 48, val); 00079 vector signed int v5 = vec_sra(v1, altivec_vectorShiftInt19); 00080 vector signed int v6 = vec_sra(v2, altivec_vectorShiftInt19); 00081 vector signed int v7 = vec_sra(v3, altivec_vectorShiftInt19); 00082 vector signed int v8 = vec_sra(v4, altivec_vectorShiftInt19); 00083 vector unsigned short vs1 = vec_packsu(v5, v6); 00084 vector unsigned short vs2 = vec_packsu(v7, v8); 00085 vector unsigned char vf = vec_packsu(vs1, vs2); 00086 vec_st(vf, i, dest); 00087 } 00088 } 00089 for ( ; i < dstW ; i++) { 00090 int t = val[i] >> 19; 00091 dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t); 00092 } 00093 } 00094 00095 static void 00096 yuv2yuvX_altivec_real(SwsContext *c, 00097 const int16_t *lumFilter, const int16_t **lumSrc, 00098 int lumFilterSize, const int16_t *chrFilter, 00099 const int16_t **chrUSrc, const int16_t **chrVSrc, 00100 int chrFilterSize, const int16_t **alpSrc, 00101 uint8_t *dest, uint8_t *uDest, 00102 uint8_t *vDest, uint8_t *aDest, 00103 int dstW, int chrDstW) 00104 { 00105 const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)}; 00106 register int i, j; 00107 { 00108 DECLARE_ALIGNED(16, int, val)[dstW]; 00109 00110 for (i = 0; i < (dstW -7); i+=4) { 00111 vec_st(vini, i << 2, val); 00112 } 00113 for (; i < dstW; i++) { 00114 val[i] = (1 << 18); 00115 } 00116 00117 for (j = 0; j < lumFilterSize; j++) { 00118 vector signed short l1, vLumFilter = vec_ld(j << 1, lumFilter); 00119 vector unsigned char perm, perm0 = vec_lvsl(j << 1, lumFilter); 00120 vLumFilter = vec_perm(vLumFilter, vLumFilter, perm0); 00121 vLumFilter = vec_splat(vLumFilter, 0); // lumFilter[j] is loaded 8 times in vLumFilter 00122 00123 perm = vec_lvsl(0, lumSrc[j]); 00124 l1 = vec_ld(0, lumSrc[j]); 00125 00126 for (i = 0; i < (dstW - 7); i+=8) { 00127 int offset = i << 2; 00128 vector signed short l2 = vec_ld((i << 1) + 16, lumSrc[j]); 00129 00130 vector signed int v1 = vec_ld(offset, val); 00131 vector signed int v2 = vec_ld(offset + 16, val); 00132 00133 vector signed short ls = vec_perm(l1, l2, perm); // lumSrc[j][i] ... lumSrc[j][i+7] 00134 00135 vector signed int i1 = vec_mule(vLumFilter, ls); 00136 vector signed int i2 = vec_mulo(vLumFilter, ls); 00137 00138 vector signed int vf1 = vec_mergeh(i1, i2); 00139 vector signed int vf2 = vec_mergel(i1, i2); // lumSrc[j][i] * lumFilter[j] ... lumSrc[j][i+7] * lumFilter[j] 00140 00141 vector signed int vo1 = vec_add(v1, vf1); 00142 vector signed int vo2 = vec_add(v2, vf2); 00143 00144 vec_st(vo1, offset, val); 00145 vec_st(vo2, offset + 16, val); 00146 00147 l1 = l2; 00148 } 00149 for ( ; i < dstW; i++) { 00150 val[i] += lumSrc[j][i] * lumFilter[j]; 00151 } 00152 } 00153 altivec_packIntArrayToCharArray(val, dest, dstW); 00154 } 00155 if (uDest != 0) { 00156 DECLARE_ALIGNED(16, int, u)[chrDstW]; 00157 DECLARE_ALIGNED(16, int, v)[chrDstW]; 00158 00159 for (i = 0; i < (chrDstW -7); i+=4) { 00160 vec_st(vini, i << 2, u); 00161 vec_st(vini, i << 2, v); 00162 } 00163 for (; i < chrDstW; i++) { 00164 u[i] = (1 << 18); 00165 v[i] = (1 << 18); 00166 } 00167 00168 for (j = 0; j < chrFilterSize; j++) { 00169 vector signed short l1, l1_V, vChrFilter = vec_ld(j << 1, chrFilter); 00170 vector unsigned char perm, perm0 = vec_lvsl(j << 1, chrFilter); 00171 vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0); 00172 vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter 00173 00174 perm = vec_lvsl(0, chrUSrc[j]); 00175 l1 = vec_ld(0, chrUSrc[j]); 00176 l1_V = vec_ld(0, chrVSrc[j]); 00177 00178 for (i = 0; i < (chrDstW - 7); i+=8) { 00179 int offset = i << 2; 00180 vector signed short l2 = vec_ld((i << 1) + 16, chrUSrc[j]); 00181 vector signed short l2_V = vec_ld((i << 1) + 16, chrVSrc[j]); 00182 00183 vector signed int v1 = vec_ld(offset, u); 00184 vector signed int v2 = vec_ld(offset + 16, u); 00185 vector signed int v1_V = vec_ld(offset, v); 00186 vector signed int v2_V = vec_ld(offset + 16, v); 00187 00188 vector signed short ls = vec_perm(l1, l2, perm); // chrUSrc[j][i] ... chrUSrc[j][i+7] 00189 vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrVSrc[j][i] ... chrVSrc[j][i] 00190 00191 vector signed int i1 = vec_mule(vChrFilter, ls); 00192 vector signed int i2 = vec_mulo(vChrFilter, ls); 00193 vector signed int i1_V = vec_mule(vChrFilter, ls_V); 00194 vector signed int i2_V = vec_mulo(vChrFilter, ls_V); 00195 00196 vector signed int vf1 = vec_mergeh(i1, i2); 00197 vector signed int vf2 = vec_mergel(i1, i2); // chrUSrc[j][i] * chrFilter[j] ... chrUSrc[j][i+7] * chrFilter[j] 00198 vector signed int vf1_V = vec_mergeh(i1_V, i2_V); 00199 vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrVSrc[j][i] * chrFilter[j] ... chrVSrc[j][i+7] * chrFilter[j] 00200 00201 vector signed int vo1 = vec_add(v1, vf1); 00202 vector signed int vo2 = vec_add(v2, vf2); 00203 vector signed int vo1_V = vec_add(v1_V, vf1_V); 00204 vector signed int vo2_V = vec_add(v2_V, vf2_V); 00205 00206 vec_st(vo1, offset, u); 00207 vec_st(vo2, offset + 16, u); 00208 vec_st(vo1_V, offset, v); 00209 vec_st(vo2_V, offset + 16, v); 00210 00211 l1 = l2; 00212 l1_V = l2_V; 00213 } 00214 for ( ; i < chrDstW; i++) { 00215 u[i] += chrUSrc[j][i] * chrFilter[j]; 00216 v[i] += chrVSrc[j][i] * chrFilter[j]; 00217 } 00218 } 00219 altivec_packIntArrayToCharArray(u, uDest, chrDstW); 00220 altivec_packIntArrayToCharArray(v, vDest, chrDstW); 00221 } 00222 } 00223 00224 static void hScale_altivec_real(int16_t *dst, int dstW, 00225 const uint8_t *src, int srcW, 00226 int xInc, const int16_t *filter, 00227 const int16_t *filterPos, int filterSize) 00228 { 00229 register int i; 00230 DECLARE_ALIGNED(16, int, tempo)[4]; 00231 00232 if (filterSize % 4) { 00233 for (i=0; i<dstW; i++) { 00234 register int j; 00235 register int srcPos = filterPos[i]; 00236 register int val = 0; 00237 for (j=0; j<filterSize; j++) { 00238 val += ((int)src[srcPos + j])*filter[filterSize*i + j]; 00239 } 00240 dst[i] = FFMIN(val>>7, (1<<15)-1); 00241 } 00242 } 00243 else 00244 switch (filterSize) { 00245 case 4: 00246 { 00247 for (i=0; i<dstW; i++) { 00248 register int srcPos = filterPos[i]; 00249 00250 vector unsigned char src_v0 = vec_ld(srcPos, src); 00251 vector unsigned char src_v1, src_vF; 00252 vector signed short src_v, filter_v; 00253 vector signed int val_vEven, val_s; 00254 if ((((uintptr_t)src + srcPos) % 16) > 12) { 00255 src_v1 = vec_ld(srcPos + 16, src); 00256 } 00257 src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src)); 00258 00259 src_v = // vec_unpackh sign-extends... 00260 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF)); 00261 // now put our elements in the even slots 00262 src_v = vec_mergeh(src_v, (vector signed short)vzero); 00263 00264 filter_v = vec_ld(i << 3, filter); 00265 // The 3 above is 2 (filterSize == 4) + 1 (sizeof(short) == 2). 00266 00267 // The neat trick: We only care for half the elements, 00268 // high or low depending on (i<<3)%16 (it's 0 or 8 here), 00269 // and we're going to use vec_mule, so we choose 00270 // carefully how to "unpack" the elements into the even slots. 00271 if ((i << 3) % 16) 00272 filter_v = vec_mergel(filter_v, (vector signed short)vzero); 00273 else 00274 filter_v = vec_mergeh(filter_v, (vector signed short)vzero); 00275 00276 val_vEven = vec_mule(src_v, filter_v); 00277 val_s = vec_sums(val_vEven, vzero); 00278 vec_st(val_s, 0, tempo); 00279 dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1); 00280 } 00281 } 00282 break; 00283 00284 case 8: 00285 { 00286 for (i=0; i<dstW; i++) { 00287 register int srcPos = filterPos[i]; 00288 00289 vector unsigned char src_v0 = vec_ld(srcPos, src); 00290 vector unsigned char src_v1, src_vF; 00291 vector signed short src_v, filter_v; 00292 vector signed int val_v, val_s; 00293 if ((((uintptr_t)src + srcPos) % 16) > 8) { 00294 src_v1 = vec_ld(srcPos + 16, src); 00295 } 00296 src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src)); 00297 00298 src_v = // vec_unpackh sign-extends... 00299 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF)); 00300 filter_v = vec_ld(i << 4, filter); 00301 // the 4 above is 3 (filterSize == 8) + 1 (sizeof(short) == 2) 00302 00303 val_v = vec_msums(src_v, filter_v, (vector signed int)vzero); 00304 val_s = vec_sums(val_v, vzero); 00305 vec_st(val_s, 0, tempo); 00306 dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1); 00307 } 00308 } 00309 break; 00310 00311 case 16: 00312 { 00313 for (i=0; i<dstW; i++) { 00314 register int srcPos = filterPos[i]; 00315 00316 vector unsigned char src_v0 = vec_ld(srcPos, src); 00317 vector unsigned char src_v1 = vec_ld(srcPos + 16, src); 00318 vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src)); 00319 00320 vector signed short src_vA = // vec_unpackh sign-extends... 00321 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF)); 00322 vector signed short src_vB = // vec_unpackh sign-extends... 00323 (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF)); 00324 00325 vector signed short filter_v0 = vec_ld(i << 5, filter); 00326 vector signed short filter_v1 = vec_ld((i << 5) + 16, filter); 00327 // the 5 above are 4 (filterSize == 16) + 1 (sizeof(short) == 2) 00328 00329 vector signed int val_acc = vec_msums(src_vA, filter_v0, (vector signed int)vzero); 00330 vector signed int val_v = vec_msums(src_vB, filter_v1, val_acc); 00331 00332 vector signed int val_s = vec_sums(val_v, vzero); 00333 00334 vec_st(val_s, 0, tempo); 00335 dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1); 00336 } 00337 } 00338 break; 00339 00340 default: 00341 { 00342 for (i=0; i<dstW; i++) { 00343 register int j; 00344 register int srcPos = filterPos[i]; 00345 00346 vector signed int val_s, val_v = (vector signed int)vzero; 00347 vector signed short filter_v0R = vec_ld(i * 2 * filterSize, filter); 00348 vector unsigned char permF = vec_lvsl((i * 2 * filterSize), filter); 00349 00350 vector unsigned char src_v0 = vec_ld(srcPos, src); 00351 vector unsigned char permS = vec_lvsl(srcPos, src); 00352 00353 for (j = 0 ; j < filterSize - 15; j += 16) { 00354 vector unsigned char src_v1 = vec_ld(srcPos + j + 16, src); 00355 vector unsigned char src_vF = vec_perm(src_v0, src_v1, permS); 00356 00357 vector signed short src_vA = // vec_unpackh sign-extends... 00358 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF)); 00359 vector signed short src_vB = // vec_unpackh sign-extends... 00360 (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF)); 00361 00362 vector signed short filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter); 00363 vector signed short filter_v2R = vec_ld((i * 2 * filterSize) + (j * 2) + 32, filter); 00364 vector signed short filter_v0 = vec_perm(filter_v0R, filter_v1R, permF); 00365 vector signed short filter_v1 = vec_perm(filter_v1R, filter_v2R, permF); 00366 00367 vector signed int val_acc = vec_msums(src_vA, filter_v0, val_v); 00368 val_v = vec_msums(src_vB, filter_v1, val_acc); 00369 00370 filter_v0R = filter_v2R; 00371 src_v0 = src_v1; 00372 } 00373 00374 if (j < filterSize-7) { 00375 // loading src_v0 is useless, it's already done above 00376 //vector unsigned char src_v0 = vec_ld(srcPos + j, src); 00377 vector unsigned char src_v1, src_vF; 00378 vector signed short src_v, filter_v1R, filter_v; 00379 if ((((uintptr_t)src + srcPos) % 16) > 8) { 00380 src_v1 = vec_ld(srcPos + j + 16, src); 00381 } 00382 src_vF = vec_perm(src_v0, src_v1, permS); 00383 00384 src_v = // vec_unpackh sign-extends... 00385 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF)); 00386 // loading filter_v0R is useless, it's already done above 00387 //vector signed short filter_v0R = vec_ld((i * 2 * filterSize) + j, filter); 00388 filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter); 00389 filter_v = vec_perm(filter_v0R, filter_v1R, permF); 00390 00391 val_v = vec_msums(src_v, filter_v, val_v); 00392 } 00393 00394 val_s = vec_sums(val_v, vzero); 00395 00396 vec_st(val_s, 0, tempo); 00397 dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1); 00398 } 00399 00400 } 00401 } 00402 } 00403 00404 void ff_sws_init_swScale_altivec(SwsContext *c) 00405 { 00406 enum PixelFormat dstFormat = c->dstFormat; 00407 00408 if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC)) 00409 return; 00410 00411 c->hScale = hScale_altivec_real; 00412 if (!is16BPS(dstFormat) && !is9_OR_10BPS(dstFormat)) { 00413 c->yuv2yuvX = yuv2yuvX_altivec_real; 00414 } 00415 00416 /* The following list of supported dstFormat values should 00417 * match what's found in the body of ff_yuv2packedX_altivec() */ 00418 if (!(c->flags & (SWS_BITEXACT | SWS_FULL_CHR_H_INT)) && !c->alpPixBuf && 00419 (c->dstFormat==PIX_FMT_ABGR || c->dstFormat==PIX_FMT_BGRA || 00420 c->dstFormat==PIX_FMT_BGR24 || c->dstFormat==PIX_FMT_RGB24 || 00421 c->dstFormat==PIX_FMT_RGBA || c->dstFormat==PIX_FMT_ARGB)) { 00422 c->yuv2packedX = ff_yuv2packedX_altivec; 00423 } 00424 }