• Main Page
  • Related Pages
  • Modules
  • Data Structures
  • Files
  • File List
  • Globals

libswscale/ppc/swscale_altivec_template.c

Go to the documentation of this file.
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 FFmpeg.
00008  *
00009  * FFmpeg 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  * FFmpeg 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 FFmpeg; if not, write to the Free Software
00021  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00022  */
00023 
00024 #define vzero vec_splat_s32(0)
00025 
00026 static inline void
00027 altivec_packIntArrayToCharArray(int *val, uint8_t* dest, int dstW)
00028 {
00029     register int i;
00030     vector unsigned int altivec_vectorShiftInt19 =
00031         vec_add(vec_splat_u32(10), vec_splat_u32(9));
00032     if ((unsigned long)dest % 16) {
00033         /* badly aligned store, we force store alignment */
00034         /* and will handle load misalignment on val w/ vec_perm */
00035         vector unsigned char perm1;
00036         vector signed int v1;
00037         for (i = 0 ; (i < dstW) &&
00038             (((unsigned long)dest + i) % 16) ; i++) {
00039                 int t = val[i] >> 19;
00040                 dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
00041         }
00042         perm1 = vec_lvsl(i << 2, val);
00043         v1 = vec_ld(i << 2, val);
00044         for ( ; i < (dstW - 15); i+=16) {
00045             int offset = i << 2;
00046             vector signed int v2 = vec_ld(offset + 16, val);
00047             vector signed int v3 = vec_ld(offset + 32, val);
00048             vector signed int v4 = vec_ld(offset + 48, val);
00049             vector signed int v5 = vec_ld(offset + 64, val);
00050             vector signed int v12 = vec_perm(v1, v2, perm1);
00051             vector signed int v23 = vec_perm(v2, v3, perm1);
00052             vector signed int v34 = vec_perm(v3, v4, perm1);
00053             vector signed int v45 = vec_perm(v4, v5, perm1);
00054 
00055             vector signed int vA = vec_sra(v12, altivec_vectorShiftInt19);
00056             vector signed int vB = vec_sra(v23, altivec_vectorShiftInt19);
00057             vector signed int vC = vec_sra(v34, altivec_vectorShiftInt19);
00058             vector signed int vD = vec_sra(v45, altivec_vectorShiftInt19);
00059             vector unsigned short vs1 = vec_packsu(vA, vB);
00060             vector unsigned short vs2 = vec_packsu(vC, vD);
00061             vector unsigned char vf = vec_packsu(vs1, vs2);
00062             vec_st(vf, i, dest);
00063             v1 = v5;
00064         }
00065     } else { // dest is properly aligned, great
00066         for (i = 0; i < (dstW - 15); i+=16) {
00067             int offset = i << 2;
00068             vector signed int v1 = vec_ld(offset, val);
00069             vector signed int v2 = vec_ld(offset + 16, val);
00070             vector signed int v3 = vec_ld(offset + 32, val);
00071             vector signed int v4 = vec_ld(offset + 48, val);
00072             vector signed int v5 = vec_sra(v1, altivec_vectorShiftInt19);
00073             vector signed int v6 = vec_sra(v2, altivec_vectorShiftInt19);
00074             vector signed int v7 = vec_sra(v3, altivec_vectorShiftInt19);
00075             vector signed int v8 = vec_sra(v4, altivec_vectorShiftInt19);
00076             vector unsigned short vs1 = vec_packsu(v5, v6);
00077             vector unsigned short vs2 = vec_packsu(v7, v8);
00078             vector unsigned char vf = vec_packsu(vs1, vs2);
00079             vec_st(vf, i, dest);
00080         }
00081     }
00082     for ( ; i < dstW ; i++) {
00083         int t = val[i] >> 19;
00084         dest[i] = (t < 0) ? 0 : ((t > 255) ? 255 : t);
00085     }
00086 }
00087 
00088 static inline void
00089 yuv2yuvX_altivec_real(const int16_t *lumFilter, int16_t **lumSrc, int lumFilterSize,
00090                       const int16_t *chrFilter, int16_t **chrSrc, int chrFilterSize,
00091                       uint8_t *dest, uint8_t *uDest, uint8_t *vDest, int dstW, int chrDstW)
00092 {
00093     const vector signed int vini = {(1 << 18), (1 << 18), (1 << 18), (1 << 18)};
00094     register int i, j;
00095     {
00096         DECLARE_ALIGNED(16, int, val)[dstW];
00097 
00098         for (i = 0; i < (dstW -7); i+=4) {
00099             vec_st(vini, i << 2, val);
00100         }
00101         for (; i < dstW; i++) {
00102             val[i] = (1 << 18);
00103         }
00104 
00105         for (j = 0; j < lumFilterSize; j++) {
00106             vector signed short l1, vLumFilter = vec_ld(j << 1, lumFilter);
00107             vector unsigned char perm, perm0 = vec_lvsl(j << 1, lumFilter);
00108             vLumFilter = vec_perm(vLumFilter, vLumFilter, perm0);
00109             vLumFilter = vec_splat(vLumFilter, 0); // lumFilter[j] is loaded 8 times in vLumFilter
00110 
00111             perm = vec_lvsl(0, lumSrc[j]);
00112             l1 = vec_ld(0, lumSrc[j]);
00113 
00114             for (i = 0; i < (dstW - 7); i+=8) {
00115                 int offset = i << 2;
00116                 vector signed short l2 = vec_ld((i << 1) + 16, lumSrc[j]);
00117 
00118                 vector signed int v1 = vec_ld(offset, val);
00119                 vector signed int v2 = vec_ld(offset + 16, val);
00120 
00121                 vector signed short ls = vec_perm(l1, l2, perm); // lumSrc[j][i] ... lumSrc[j][i+7]
00122 
00123                 vector signed int i1 = vec_mule(vLumFilter, ls);
00124                 vector signed int i2 = vec_mulo(vLumFilter, ls);
00125 
00126                 vector signed int vf1 = vec_mergeh(i1, i2);
00127                 vector signed int vf2 = vec_mergel(i1, i2); // lumSrc[j][i] * lumFilter[j] ... lumSrc[j][i+7] * lumFilter[j]
00128 
00129                 vector signed int vo1 = vec_add(v1, vf1);
00130                 vector signed int vo2 = vec_add(v2, vf2);
00131 
00132                 vec_st(vo1, offset, val);
00133                 vec_st(vo2, offset + 16, val);
00134 
00135                 l1 = l2;
00136             }
00137             for ( ; i < dstW; i++) {
00138                 val[i] += lumSrc[j][i] * lumFilter[j];
00139             }
00140         }
00141         altivec_packIntArrayToCharArray(val, dest, dstW);
00142     }
00143     if (uDest != 0) {
00144         DECLARE_ALIGNED(16, int, u)[chrDstW];
00145         DECLARE_ALIGNED(16, int, v)[chrDstW];
00146 
00147         for (i = 0; i < (chrDstW -7); i+=4) {
00148             vec_st(vini, i << 2, u);
00149             vec_st(vini, i << 2, v);
00150         }
00151         for (; i < chrDstW; i++) {
00152             u[i] = (1 << 18);
00153             v[i] = (1 << 18);
00154         }
00155 
00156         for (j = 0; j < chrFilterSize; j++) {
00157             vector signed short l1, l1_V, vChrFilter = vec_ld(j << 1, chrFilter);
00158             vector unsigned char perm, perm0 = vec_lvsl(j << 1, chrFilter);
00159             vChrFilter = vec_perm(vChrFilter, vChrFilter, perm0);
00160             vChrFilter = vec_splat(vChrFilter, 0); // chrFilter[j] is loaded 8 times in vChrFilter
00161 
00162             perm = vec_lvsl(0, chrSrc[j]);
00163             l1 = vec_ld(0, chrSrc[j]);
00164             l1_V = vec_ld(2048 << 1, chrSrc[j]);
00165 
00166             for (i = 0; i < (chrDstW - 7); i+=8) {
00167                 int offset = i << 2;
00168                 vector signed short l2 = vec_ld((i << 1) + 16, chrSrc[j]);
00169                 vector signed short l2_V = vec_ld(((i + 2048) << 1) + 16, chrSrc[j]);
00170 
00171                 vector signed int v1 = vec_ld(offset, u);
00172                 vector signed int v2 = vec_ld(offset + 16, u);
00173                 vector signed int v1_V = vec_ld(offset, v);
00174                 vector signed int v2_V = vec_ld(offset + 16, v);
00175 
00176                 vector signed short ls = vec_perm(l1, l2, perm); // chrSrc[j][i] ... chrSrc[j][i+7]
00177                 vector signed short ls_V = vec_perm(l1_V, l2_V, perm); // chrSrc[j][i+2048] ... chrSrc[j][i+2055]
00178 
00179                 vector signed int i1 = vec_mule(vChrFilter, ls);
00180                 vector signed int i2 = vec_mulo(vChrFilter, ls);
00181                 vector signed int i1_V = vec_mule(vChrFilter, ls_V);
00182                 vector signed int i2_V = vec_mulo(vChrFilter, ls_V);
00183 
00184                 vector signed int vf1 = vec_mergeh(i1, i2);
00185                 vector signed int vf2 = vec_mergel(i1, i2); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
00186                 vector signed int vf1_V = vec_mergeh(i1_V, i2_V);
00187                 vector signed int vf2_V = vec_mergel(i1_V, i2_V); // chrSrc[j][i] * chrFilter[j] ... chrSrc[j][i+7] * chrFilter[j]
00188 
00189                 vector signed int vo1 = vec_add(v1, vf1);
00190                 vector signed int vo2 = vec_add(v2, vf2);
00191                 vector signed int vo1_V = vec_add(v1_V, vf1_V);
00192                 vector signed int vo2_V = vec_add(v2_V, vf2_V);
00193 
00194                 vec_st(vo1, offset, u);
00195                 vec_st(vo2, offset + 16, u);
00196                 vec_st(vo1_V, offset, v);
00197                 vec_st(vo2_V, offset + 16, v);
00198 
00199                 l1 = l2;
00200                 l1_V = l2_V;
00201             }
00202             for ( ; i < chrDstW; i++) {
00203                 u[i] += chrSrc[j][i] * chrFilter[j];
00204                 v[i] += chrSrc[j][i + 2048] * chrFilter[j];
00205             }
00206         }
00207         altivec_packIntArrayToCharArray(u, uDest, chrDstW);
00208         altivec_packIntArrayToCharArray(v, vDest, chrDstW);
00209     }
00210 }
00211 
00212 static inline void hScale_altivec_real(int16_t *dst, int dstW,
00213                                        const uint8_t *src, int srcW,
00214                                        int xInc, const int16_t *filter,
00215                                        const int16_t *filterPos, int filterSize)
00216 {
00217     register int i;
00218     DECLARE_ALIGNED(16, int, tempo)[4];
00219 
00220     if (filterSize % 4) {
00221         for (i=0; i<dstW; i++) {
00222             register int j;
00223             register int srcPos = filterPos[i];
00224             register int val = 0;
00225             for (j=0; j<filterSize; j++) {
00226                 val += ((int)src[srcPos + j])*filter[filterSize*i + j];
00227             }
00228             dst[i] = FFMIN(val>>7, (1<<15)-1);
00229         }
00230     }
00231     else
00232     switch (filterSize) {
00233     case 4:
00234     {
00235     for (i=0; i<dstW; i++) {
00236         register int srcPos = filterPos[i];
00237 
00238         vector unsigned char src_v0 = vec_ld(srcPos, src);
00239         vector unsigned char src_v1, src_vF;
00240         vector signed short src_v, filter_v;
00241         vector signed int val_vEven, val_s;
00242         if ((((int)src + srcPos)% 16) > 12) {
00243             src_v1 = vec_ld(srcPos + 16, src);
00244         }
00245         src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
00246 
00247         src_v = // vec_unpackh sign-extends...
00248             (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
00249         // now put our elements in the even slots
00250         src_v = vec_mergeh(src_v, (vector signed short)vzero);
00251 
00252         filter_v = vec_ld(i << 3, filter);
00253         // The 3 above is 2 (filterSize == 4) + 1 (sizeof(short) == 2).
00254 
00255         // The neat trick: We only care for half the elements,
00256         // high or low depending on (i<<3)%16 (it's 0 or 8 here),
00257         // and we're going to use vec_mule, so we choose
00258         // carefully how to "unpack" the elements into the even slots.
00259         if ((i << 3) % 16)
00260             filter_v = vec_mergel(filter_v, (vector signed short)vzero);
00261         else
00262             filter_v = vec_mergeh(filter_v, (vector signed short)vzero);
00263 
00264         val_vEven = vec_mule(src_v, filter_v);
00265         val_s = vec_sums(val_vEven, vzero);
00266         vec_st(val_s, 0, tempo);
00267         dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
00268     }
00269     }
00270     break;
00271 
00272     case 8:
00273     {
00274     for (i=0; i<dstW; i++) {
00275         register int srcPos = filterPos[i];
00276 
00277         vector unsigned char src_v0 = vec_ld(srcPos, src);
00278         vector unsigned char src_v1, src_vF;
00279         vector signed short src_v, filter_v;
00280         vector signed int val_v, val_s;
00281         if ((((int)src + srcPos)% 16) > 8) {
00282             src_v1 = vec_ld(srcPos + 16, src);
00283         }
00284         src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
00285 
00286         src_v = // vec_unpackh sign-extends...
00287             (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
00288         filter_v = vec_ld(i << 4, filter);
00289         // the 4 above is 3 (filterSize == 8) + 1 (sizeof(short) == 2)
00290 
00291         val_v = vec_msums(src_v, filter_v, (vector signed int)vzero);
00292         val_s = vec_sums(val_v, vzero);
00293         vec_st(val_s, 0, tempo);
00294         dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
00295     }
00296     }
00297     break;
00298 
00299     case 16:
00300     {
00301         for (i=0; i<dstW; i++) {
00302             register int srcPos = filterPos[i];
00303 
00304             vector unsigned char src_v0 = vec_ld(srcPos, src);
00305             vector unsigned char src_v1 = vec_ld(srcPos + 16, src);
00306             vector unsigned char src_vF = vec_perm(src_v0, src_v1, vec_lvsl(srcPos, src));
00307 
00308             vector signed short src_vA = // vec_unpackh sign-extends...
00309                 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
00310             vector signed short src_vB = // vec_unpackh sign-extends...
00311                 (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
00312 
00313             vector signed short filter_v0 = vec_ld(i << 5, filter);
00314             vector signed short filter_v1 = vec_ld((i << 5) + 16, filter);
00315             // the 5 above are 4 (filterSize == 16) + 1 (sizeof(short) == 2)
00316 
00317             vector signed int val_acc = vec_msums(src_vA, filter_v0, (vector signed int)vzero);
00318             vector signed int val_v = vec_msums(src_vB, filter_v1, val_acc);
00319 
00320             vector signed int val_s = vec_sums(val_v, vzero);
00321 
00322             vec_st(val_s, 0, tempo);
00323             dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
00324         }
00325     }
00326     break;
00327 
00328     default:
00329     {
00330     for (i=0; i<dstW; i++) {
00331         register int j;
00332         register int srcPos = filterPos[i];
00333 
00334         vector signed int val_s, val_v = (vector signed int)vzero;
00335         vector signed short filter_v0R = vec_ld(i * 2 * filterSize, filter);
00336         vector unsigned char permF = vec_lvsl((i * 2 * filterSize), filter);
00337 
00338         vector unsigned char src_v0 = vec_ld(srcPos, src);
00339         vector unsigned char permS = vec_lvsl(srcPos, src);
00340 
00341         for (j = 0 ; j < filterSize - 15; j += 16) {
00342             vector unsigned char src_v1 = vec_ld(srcPos + j + 16, src);
00343             vector unsigned char src_vF = vec_perm(src_v0, src_v1, permS);
00344 
00345             vector signed short src_vA = // vec_unpackh sign-extends...
00346                 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
00347             vector signed short src_vB = // vec_unpackh sign-extends...
00348                 (vector signed short)(vec_mergel((vector unsigned char)vzero, src_vF));
00349 
00350             vector signed short filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
00351             vector signed short filter_v2R = vec_ld((i * 2 * filterSize) + (j * 2) + 32, filter);
00352             vector signed short filter_v0  = vec_perm(filter_v0R, filter_v1R, permF);
00353             vector signed short filter_v1  = vec_perm(filter_v1R, filter_v2R, permF);
00354 
00355             vector signed int val_acc = vec_msums(src_vA, filter_v0, val_v);
00356             val_v = vec_msums(src_vB, filter_v1, val_acc);
00357 
00358             filter_v0R = filter_v2R;
00359             src_v0 = src_v1;
00360         }
00361 
00362         if (j < filterSize-7) {
00363             // loading src_v0 is useless, it's already done above
00364             //vector unsigned char src_v0 = vec_ld(srcPos + j, src);
00365             vector unsigned char src_v1, src_vF;
00366             vector signed short src_v, filter_v1R, filter_v;
00367             if ((((int)src + srcPos)% 16) > 8) {
00368                 src_v1 = vec_ld(srcPos + j + 16, src);
00369             }
00370             src_vF = vec_perm(src_v0, src_v1, permS);
00371 
00372             src_v = // vec_unpackh sign-extends...
00373                 (vector signed short)(vec_mergeh((vector unsigned char)vzero, src_vF));
00374             // loading filter_v0R is useless, it's already done above
00375             //vector signed short filter_v0R = vec_ld((i * 2 * filterSize) + j, filter);
00376             filter_v1R = vec_ld((i * 2 * filterSize) + (j * 2) + 16, filter);
00377             filter_v = vec_perm(filter_v0R, filter_v1R, permF);
00378 
00379             val_v = vec_msums(src_v, filter_v, val_v);
00380         }
00381 
00382         val_s = vec_sums(val_v, vzero);
00383 
00384         vec_st(val_s, 0, tempo);
00385         dst[i] = FFMIN(tempo[3]>>7, (1<<15)-1);
00386     }
00387 
00388     }
00389     }
00390 }
00391 
00392 static inline int yv12toyuy2_unscaled_altivec(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
00393                                               int srcSliceH, uint8_t* dstParam[], int dstStride_a[])
00394 {
00395     uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
00396     // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
00397     uint8_t *ysrc = src[0];
00398     uint8_t *usrc = src[1];
00399     uint8_t *vsrc = src[2];
00400     const int width = c->srcW;
00401     const int height = srcSliceH;
00402     const int lumStride = srcStride[0];
00403     const int chromStride = srcStride[1];
00404     const int dstStride = dstStride_a[0];
00405     const vector unsigned char yperm = vec_lvsl(0, ysrc);
00406     const int vertLumPerChroma = 2;
00407     register unsigned int y;
00408 
00409     if (width&15) {
00410         yv12toyuy2(ysrc, usrc, vsrc, dst, c->srcW, srcSliceH, lumStride, chromStride, dstStride);
00411         return srcSliceH;
00412     }
00413 
00414     /* This code assumes:
00415 
00416     1) dst is 16 bytes-aligned
00417     2) dstStride is a multiple of 16
00418     3) width is a multiple of 16
00419     4) lum & chrom stride are multiples of 8
00420     */
00421 
00422     for (y=0; y<height; y++) {
00423         int i;
00424         for (i = 0; i < width - 31; i+= 32) {
00425             const unsigned int j = i >> 1;
00426             vector unsigned char v_yA = vec_ld(i, ysrc);
00427             vector unsigned char v_yB = vec_ld(i + 16, ysrc);
00428             vector unsigned char v_yC = vec_ld(i + 32, ysrc);
00429             vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
00430             vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
00431             vector unsigned char v_uA = vec_ld(j, usrc);
00432             vector unsigned char v_uB = vec_ld(j + 16, usrc);
00433             vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
00434             vector unsigned char v_vA = vec_ld(j, vsrc);
00435             vector unsigned char v_vB = vec_ld(j + 16, vsrc);
00436             vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
00437             vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
00438             vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
00439             vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
00440             vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
00441             vector unsigned char v_yuy2_2 = vec_mergeh(v_y2, v_uv_b);
00442             vector unsigned char v_yuy2_3 = vec_mergel(v_y2, v_uv_b);
00443             vec_st(v_yuy2_0, (i << 1), dst);
00444             vec_st(v_yuy2_1, (i << 1) + 16, dst);
00445             vec_st(v_yuy2_2, (i << 1) + 32, dst);
00446             vec_st(v_yuy2_3, (i << 1) + 48, dst);
00447         }
00448         if (i < width) {
00449             const unsigned int j = i >> 1;
00450             vector unsigned char v_y1 = vec_ld(i, ysrc);
00451             vector unsigned char v_u = vec_ld(j, usrc);
00452             vector unsigned char v_v = vec_ld(j, vsrc);
00453             vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
00454             vector unsigned char v_yuy2_0 = vec_mergeh(v_y1, v_uv_a);
00455             vector unsigned char v_yuy2_1 = vec_mergel(v_y1, v_uv_a);
00456             vec_st(v_yuy2_0, (i << 1), dst);
00457             vec_st(v_yuy2_1, (i << 1) + 16, dst);
00458         }
00459         if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
00460             usrc += chromStride;
00461             vsrc += chromStride;
00462         }
00463         ysrc += lumStride;
00464         dst += dstStride;
00465     }
00466 
00467     return srcSliceH;
00468 }
00469 
00470 static inline int yv12touyvy_unscaled_altivec(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY,
00471                                               int srcSliceH, uint8_t* dstParam[], int dstStride_a[])
00472 {
00473     uint8_t *dst=dstParam[0] + dstStride_a[0]*srcSliceY;
00474     // yv12toyuy2(src[0], src[1], src[2], dst, c->srcW, srcSliceH, srcStride[0], srcStride[1], dstStride[0]);
00475     uint8_t *ysrc = src[0];
00476     uint8_t *usrc = src[1];
00477     uint8_t *vsrc = src[2];
00478     const int width = c->srcW;
00479     const int height = srcSliceH;
00480     const int lumStride = srcStride[0];
00481     const int chromStride = srcStride[1];
00482     const int dstStride = dstStride_a[0];
00483     const int vertLumPerChroma = 2;
00484     const vector unsigned char yperm = vec_lvsl(0, ysrc);
00485     register unsigned int y;
00486 
00487     if (width&15) {
00488         yv12touyvy(ysrc, usrc, vsrc, dst, c->srcW, srcSliceH, lumStride, chromStride, dstStride);
00489         return srcSliceH;
00490     }
00491 
00492     /* This code assumes:
00493 
00494     1) dst is 16 bytes-aligned
00495     2) dstStride is a multiple of 16
00496     3) width is a multiple of 16
00497     4) lum & chrom stride are multiples of 8
00498     */
00499 
00500     for (y=0; y<height; y++) {
00501         int i;
00502         for (i = 0; i < width - 31; i+= 32) {
00503             const unsigned int j = i >> 1;
00504             vector unsigned char v_yA = vec_ld(i, ysrc);
00505             vector unsigned char v_yB = vec_ld(i + 16, ysrc);
00506             vector unsigned char v_yC = vec_ld(i + 32, ysrc);
00507             vector unsigned char v_y1 = vec_perm(v_yA, v_yB, yperm);
00508             vector unsigned char v_y2 = vec_perm(v_yB, v_yC, yperm);
00509             vector unsigned char v_uA = vec_ld(j, usrc);
00510             vector unsigned char v_uB = vec_ld(j + 16, usrc);
00511             vector unsigned char v_u = vec_perm(v_uA, v_uB, vec_lvsl(j, usrc));
00512             vector unsigned char v_vA = vec_ld(j, vsrc);
00513             vector unsigned char v_vB = vec_ld(j + 16, vsrc);
00514             vector unsigned char v_v = vec_perm(v_vA, v_vB, vec_lvsl(j, vsrc));
00515             vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
00516             vector unsigned char v_uv_b = vec_mergel(v_u, v_v);
00517             vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
00518             vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
00519             vector unsigned char v_uyvy_2 = vec_mergeh(v_uv_b, v_y2);
00520             vector unsigned char v_uyvy_3 = vec_mergel(v_uv_b, v_y2);
00521             vec_st(v_uyvy_0, (i << 1), dst);
00522             vec_st(v_uyvy_1, (i << 1) + 16, dst);
00523             vec_st(v_uyvy_2, (i << 1) + 32, dst);
00524             vec_st(v_uyvy_3, (i << 1) + 48, dst);
00525         }
00526         if (i < width) {
00527             const unsigned int j = i >> 1;
00528             vector unsigned char v_y1 = vec_ld(i, ysrc);
00529             vector unsigned char v_u = vec_ld(j, usrc);
00530             vector unsigned char v_v = vec_ld(j, vsrc);
00531             vector unsigned char v_uv_a = vec_mergeh(v_u, v_v);
00532             vector unsigned char v_uyvy_0 = vec_mergeh(v_uv_a, v_y1);
00533             vector unsigned char v_uyvy_1 = vec_mergel(v_uv_a, v_y1);
00534             vec_st(v_uyvy_0, (i << 1), dst);
00535             vec_st(v_uyvy_1, (i << 1) + 16, dst);
00536         }
00537         if ((y&(vertLumPerChroma-1)) == vertLumPerChroma-1) {
00538             usrc += chromStride;
00539             vsrc += chromStride;
00540         }
00541         ysrc += lumStride;
00542         dst += dstStride;
00543     }
00544     return srcSliceH;
00545 }

Generated on Fri Sep 16 2011 17:17:52 for FFmpeg by  doxygen 1.7.1