* VC-1 and WMV3 decoder - DSP functions AltiVec-optimized
* Copyright (c) 2006 Konstantin Shishkov
*
- * This file is part of FFmpeg.
+ * This file is part of Libav.
*
- * FFmpeg is free software; you can redistribute it and/or
+ * Libav is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
- * FFmpeg is distributed in the hope that it will be useful,
+ * Libav is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
- * License along with FFmpeg; if not, write to the Free Software
+ * License along with Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
- *
*/
-#include "../dsputil.h"
-
-#include "gcc_fixes.h"
-
-#include "dsputil_altivec.h"
-
-// Transpose 8x8 matrix of 16-bit elements. Borrowed from mpegvideo_altivec.c
-#define TRANSPOSE8(a,b,c,d,e,f,g,h) \
-do { \
- vector signed short A1, B1, C1, D1, E1, F1, G1, H1; \
- vector signed short A2, B2, C2, D2, E2, F2, G2, H2; \
- \
- A1 = vec_mergeh (a, e); \
- B1 = vec_mergel (a, e); \
- C1 = vec_mergeh (b, f); \
- D1 = vec_mergel (b, f); \
- E1 = vec_mergeh (c, g); \
- F1 = vec_mergel (c, g); \
- G1 = vec_mergeh (d, h); \
- H1 = vec_mergel (d, h); \
- \
- A2 = vec_mergeh (A1, E1); \
- B2 = vec_mergel (A1, E1); \
- C2 = vec_mergeh (B1, F1); \
- D2 = vec_mergel (B1, F1); \
- E2 = vec_mergeh (C1, G1); \
- F2 = vec_mergel (C1, G1); \
- G2 = vec_mergeh (D1, H1); \
- H2 = vec_mergel (D1, H1); \
- \
- a = vec_mergeh (A2, E2); \
- b = vec_mergel (A2, E2); \
- c = vec_mergeh (B2, F2); \
- d = vec_mergel (B2, F2); \
- e = vec_mergeh (C2, G2); \
- f = vec_mergel (C2, G2); \
- g = vec_mergeh (D2, H2); \
- h = vec_mergel (D2, H2); \
-} while (0)
+#include "libavutil/ppc/types_altivec.h"
+#include "libavutil/ppc/util_altivec.h"
+#include "libavcodec/dsputil.h"
+#include "libavcodec/vc1dsp.h"
// main steps of 8x8 transform
#define STEP8(s0, s1, s2, s3, s4, s5, s6, s7, vec_rnd) \
vector signed int t0, t1, t2, t3, t4, t5, t6, t7;
const vector signed int vec_64 = vec_sl(vec_splat_s32(4), vec_splat_u32(4));
const vector unsigned int vec_7 = vec_splat_u32(7);
- const vector unsigned int vec_5 = vec_splat_u32(5);
const vector unsigned int vec_4 = vec_splat_u32(4);
const vector signed int vec_4s = vec_splat_s32(4);
const vector unsigned int vec_3 = vec_splat_u32(3);
const vector signed int vec_1s = vec_splat_s32(1);
const vector unsigned int vec_1 = vec_splat_u32(1);
-
src0 = vec_ld( 0, block);
src1 = vec_ld( 16, block);
src2 = vec_ld( 32, block);
src6 = vec_ld( 96, block);
src7 = vec_ld(112, block);
- TRANSPOSE8(src0, src1, src2, src3, src4, src5, src6, src7);
s0 = vec_unpackl(src0);
s1 = vec_unpackl(src1);
s2 = vec_unpackl(src2);
/** Do inverse transform on 8x4 part of block
*/
-static void vc1_inv_trans_8x4_altivec(DCTELEM block[64], int n)
+static void vc1_inv_trans_8x4_altivec(uint8_t *dest, int stride, DCTELEM *block)
{
vector signed short src0, src1, src2, src3, src4, src5, src6, src7;
vector signed int s0, s1, s2, s3, s4, s5, s6, s7;
const vector unsigned int vec_3 = vec_splat_u32(3);
const vector unsigned int vec_2 = vec_splat_u32(2);
const vector unsigned int vec_1 = vec_splat_u32(1);
+ vector unsigned char tmp;
+ vector signed short tmp2, tmp3;
+ vector unsigned char perm0, perm1, p0, p1, p;
src0 = vec_ld( 0, block);
src1 = vec_ld( 16, block);
src7 = vec_pack(sF, s7);
TRANSPOSE8(src0, src1, src2, src3, src4, src5, src6, src7);
- if(!n){ // upper half of block
- s0 = vec_unpackh(src0);
- s1 = vec_unpackh(src1);
- s2 = vec_unpackh(src2);
- s3 = vec_unpackh(src3);
- s8 = vec_unpackl(src0);
- s9 = vec_unpackl(src1);
- sA = vec_unpackl(src2);
- sB = vec_unpackl(src3);
- STEP4(s0, s1, s2, s3, vec_64);
- SHIFT_VERT4(s0, s1, s2, s3);
- STEP4(s8, s9, sA, sB, vec_64);
- SHIFT_VERT4(s8, s9, sA, sB);
- src0 = vec_pack(s0, s8);
- src1 = vec_pack(s1, s9);
- src2 = vec_pack(s2, sA);
- src3 = vec_pack(s3, sB);
+ s0 = vec_unpackh(src0);
+ s1 = vec_unpackh(src1);
+ s2 = vec_unpackh(src2);
+ s3 = vec_unpackh(src3);
+ s8 = vec_unpackl(src0);
+ s9 = vec_unpackl(src1);
+ sA = vec_unpackl(src2);
+ sB = vec_unpackl(src3);
+ STEP4(s0, s1, s2, s3, vec_64);
+ SHIFT_VERT4(s0, s1, s2, s3);
+ STEP4(s8, s9, sA, sB, vec_64);
+ SHIFT_VERT4(s8, s9, sA, sB);
+ src0 = vec_pack(s0, s8);
+ src1 = vec_pack(s1, s9);
+ src2 = vec_pack(s2, sA);
+ src3 = vec_pack(s3, sB);
+
+ p0 = vec_lvsl (0, dest);
+ p1 = vec_lvsl (stride, dest);
+ p = vec_splat_u8 (-1);
+ perm0 = vec_mergeh (p, p0);
+ perm1 = vec_mergeh (p, p1);
- vec_st(src0, 0, block);
- vec_st(src1, 16, block);
- vec_st(src2, 32, block);
- vec_st(src3, 48, block);
- } else { //lower half of block
- s0 = vec_unpackh(src4);
- s1 = vec_unpackh(src5);
- s2 = vec_unpackh(src6);
- s3 = vec_unpackh(src7);
- s8 = vec_unpackl(src4);
- s9 = vec_unpackl(src5);
- sA = vec_unpackl(src6);
- sB = vec_unpackl(src7);
- STEP4(s0, s1, s2, s3, vec_64);
- SHIFT_VERT4(s0, s1, s2, s3);
- STEP4(s8, s9, sA, sB, vec_64);
- SHIFT_VERT4(s8, s9, sA, sB);
- src4 = vec_pack(s0, s8);
- src5 = vec_pack(s1, s9);
- src6 = vec_pack(s2, sA);
- src7 = vec_pack(s3, sB);
+#define ADD(dest,src,perm) \
+ /* *(uint64_t *)&tmp = *(uint64_t *)dest; */ \
+ tmp = vec_ld (0, dest); \
+ tmp2 = (vector signed short)vec_perm (tmp, vec_splat_u8(0), perm); \
+ tmp3 = vec_adds (tmp2, src); \
+ tmp = vec_packsu (tmp3, tmp3); \
+ vec_ste ((vector unsigned int)tmp, 0, (unsigned int *)dest); \
+ vec_ste ((vector unsigned int)tmp, 4, (unsigned int *)dest);
- vec_st(src4, 64, block);
- vec_st(src5, 80, block);
- vec_st(src6, 96, block);
- vec_st(src7,112, block);
- }
+ ADD (dest, src0, perm0) dest += stride;
+ ADD (dest, src1, perm1) dest += stride;
+ ADD (dest, src2, perm0) dest += stride;
+ ADD (dest, src3, perm1)
}
+#define PUT_OP_U8_ALTIVEC(d, s, dst) d = s
+#define AVG_OP_U8_ALTIVEC(d, s, dst) d = vec_avg(dst, s)
+
+#define OP_U8_ALTIVEC PUT_OP_U8_ALTIVEC
+#define PREFIX_no_rnd_vc1_chroma_mc8_altivec put_no_rnd_vc1_chroma_mc8_altivec
+#include "h264_altivec_template.c"
+#undef OP_U8_ALTIVEC
+#undef PREFIX_no_rnd_vc1_chroma_mc8_altivec
+
+#define OP_U8_ALTIVEC AVG_OP_U8_ALTIVEC
+#define PREFIX_no_rnd_vc1_chroma_mc8_altivec avg_no_rnd_vc1_chroma_mc8_altivec
+#include "h264_altivec_template.c"
+#undef OP_U8_ALTIVEC
+#undef PREFIX_no_rnd_vc1_chroma_mc8_altivec
+
+void ff_vc1dsp_init_altivec(VC1DSPContext* dsp)
+{
+ if (!(av_get_cpu_flags() & AV_CPU_FLAG_ALTIVEC))
+ return;
-void vc1dsp_init_altivec(DSPContext* dsp, AVCodecContext *avctx) {
dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_altivec;
dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_altivec;
+ dsp->put_no_rnd_vc1_chroma_pixels_tab[0] = put_no_rnd_vc1_chroma_mc8_altivec;
+ dsp->avg_no_rnd_vc1_chroma_pixels_tab[0] = avg_no_rnd_vc1_chroma_mc8_altivec;
}