X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fppc%2Fidct_altivec.c;h=d65ba24f3a8ee298a2023f34002256ad262bb55d;hb=6d6b0c96fdc576f4643e3d4066e5ebebdb693ade;hp=88843fb5d63ace34bd18fc310da6ec4da1d030ef;hpb=b78e7197a81e193827cf2408fe25bc1f14843a72;p=ffmpeg diff --git a/libavcodec/ppc/idct_altivec.c b/libavcodec/ppc/idct_altivec.c index 88843fb5d63..d65ba24f3a8 100644 --- a/libavcodec/ppc/idct_altivec.c +++ b/libavcodec/ppc/idct_altivec.c @@ -16,43 +16,35 @@ * You should have received a copy of the GNU Lesser General Public * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - * */ /* * NOTE: This code is based on GPL code from the libmpeg2 project. The * author, Michel Lespinasses, has given explicit permission to release - * under LGPL as part of ffmpeg. - * + * under LGPL as part of FFmpeg. */ /* - * FFMpeg integration by Dieter Shirley + * FFmpeg integration by Dieter Shirley * - * This file is a direct copy of the altivec idct module from the libmpeg2 - * project. I've deleted all of the libmpeg2 specific code, renamed the functions and - * re-ordered the function parameters. The only change to the IDCT function - * itself was to factor out the partial transposition, and to perform a full - * transpose at the end of the function. + * This file is a direct copy of the AltiVec IDCT module from the libmpeg2 + * project. I've deleted all of the libmpeg2-specific code, renamed the + * functions and reordered the function parameters. The only change to the + * IDCT function itself was to factor out the partial transposition, and to + * perform a full transpose at the end of the function. */ #include /* malloc(), free() */ #include -#include "../dsputil.h" - -#include "gcc_fixes.h" - +#include "config.h" +#if HAVE_ALTIVEC_H +#include +#endif +#include "libavcodec/dsputil.h" +#include "types_altivec.h" #include "dsputil_altivec.h" -#define vector_s16_t vector signed short -#define const_vector_s16_t const_vector signed short -#define vector_u16_t vector unsigned short -#define vector_s8_t vector signed char -#define vector_u8_t vector unsigned char -#define vector_s32_t vector signed int -#define vector_u32_t vector unsigned int - #define IDCT_HALF \ /* 1st stage */ \ t1 = vec_mradds (a1, vx7, vx1 ); \ @@ -90,11 +82,11 @@ #define IDCT \ - vector_s16_t vx0, vx1, vx2, vx3, vx4, vx5, vx6, vx7; \ - vector_s16_t vy0, vy1, vy2, vy3, vy4, vy5, vy6, vy7; \ - vector_s16_t a0, a1, a2, ma2, c4, mc4, zero, bias; \ - vector_s16_t t0, t1, t2, t3, t4, t5, t6, t7, t8; \ - vector_u16_t shift; \ + vec_s16 vx0, vx1, vx2, vx3, vx4, vx5, vx6, vx7; \ + vec_s16 vy0, vy1, vy2, vy3, vy4, vy5, vy6, vy7; \ + vec_s16 a0, a1, a2, ma2, c4, mc4, zero, bias; \ + vec_s16 t0, t1, t2, t3, t4, t5, t6, t7, t8; \ + vec_u16 shift; \ \ c4 = vec_splat (constants[0], 0); \ a0 = vec_splat (constants[0], 1); \ @@ -102,7 +94,7 @@ a2 = vec_splat (constants[0], 3); \ mc4 = vec_splat (constants[0], 4); \ ma2 = vec_splat (constants[0], 5); \ - bias = (vector_s16_t)vec_splat ((vector_s32_t)constants[0], 3); \ + bias = (vec_s16)vec_splat ((vec_s32)constants[0], 3); \ \ zero = vec_splat_s16 (0); \ shift = vec_splat_u16 (4); \ @@ -158,34 +150,25 @@ vx7 = vec_sra (vy7, shift); -static const_vector_s16_t constants[5] = { - (vector_s16_t) AVV(23170, 13573, 6518, 21895, -23170, -21895, 32, 31), - (vector_s16_t) AVV(16384, 22725, 21407, 19266, 16384, 19266, 21407, 22725), - (vector_s16_t) AVV(22725, 31521, 29692, 26722, 22725, 26722, 29692, 31521), - (vector_s16_t) AVV(21407, 29692, 27969, 25172, 21407, 25172, 27969, 29692), - (vector_s16_t) AVV(19266, 26722, 25172, 22654, 19266, 22654, 25172, 26722) +static const vec_s16 constants[5] = { + {23170, 13573, 6518, 21895, -23170, -21895, 32, 31}, + {16384, 22725, 21407, 19266, 16384, 19266, 21407, 22725}, + {22725, 31521, 29692, 26722, 22725, 26722, 29692, 31521}, + {21407, 29692, 27969, 25172, 21407, 25172, 27969, 29692}, + {19266, 26722, 25172, 22654, 19266, 22654, 25172, 26722} }; -void idct_put_altivec(uint8_t* dest, int stride, vector_s16_t* block) +void idct_put_altivec(uint8_t* dest, int stride, int16_t *blk) { -POWERPC_PERF_DECLARE(altivec_idct_put_num, 1); -#ifdef ALTIVEC_USE_REFERENCE_C_CODE -POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1); - void simple_idct_put(uint8_t *dest, int line_size, int16_t *block); - simple_idct_put(dest, stride, (int16_t*)block); -POWERPC_PERF_STOP_COUNT(altivec_idct_put_num, 1); -#else /* ALTIVEC_USE_REFERENCE_C_CODE */ - vector_u8_t tmp; - -#ifdef POWERPC_PERFORMANCE_REPORT -POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1); -#endif + vec_s16 *block = (vec_s16*)blk; + vec_u8 tmp; + IDCT #define COPY(dest,src) \ tmp = vec_packsu (src, src); \ - vec_ste ((vector_u32_t)tmp, 0, (unsigned int *)dest); \ - vec_ste ((vector_u32_t)tmp, 4, (unsigned int *)dest); + vec_ste ((vec_u32)tmp, 0, (unsigned int *)dest); \ + vec_ste ((vec_u32)tmp, 4, (unsigned int *)dest); COPY (dest, vx0) dest += stride; COPY (dest, vx1) dest += stride; @@ -195,29 +178,16 @@ POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1); COPY (dest, vx5) dest += stride; COPY (dest, vx6) dest += stride; COPY (dest, vx7) - -POWERPC_PERF_STOP_COUNT(altivec_idct_put_num, 1); -#endif /* ALTIVEC_USE_REFERENCE_C_CODE */ } -void idct_add_altivec(uint8_t* dest, int stride, vector_s16_t* block) +void idct_add_altivec(uint8_t* dest, int stride, int16_t *blk) { -POWERPC_PERF_DECLARE(altivec_idct_add_num, 1); -#ifdef ALTIVEC_USE_REFERENCE_C_CODE -POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1); - void simple_idct_add(uint8_t *dest, int line_size, int16_t *block); - simple_idct_add(dest, stride, (int16_t*)block); -POWERPC_PERF_STOP_COUNT(altivec_idct_add_num, 1); -#else /* ALTIVEC_USE_REFERENCE_C_CODE */ - vector_u8_t tmp; - vector_s16_t tmp2, tmp3; - vector_u8_t perm0; - vector_u8_t perm1; - vector_u8_t p0, p1, p; - -#ifdef POWERPC_PERFORMANCE_REPORT -POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1); -#endif + vec_s16 *block = (vec_s16*)blk; + vec_u8 tmp; + vec_s16 tmp2, tmp3; + vec_u8 perm0; + vec_u8 perm1; + vec_u8 p0, p1, p; IDCT @@ -230,11 +200,11 @@ POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1); #define ADD(dest,src,perm) \ /* *(uint64_t *)&tmp = *(uint64_t *)dest; */ \ tmp = vec_ld (0, dest); \ - tmp2 = (vector_s16_t)vec_perm (tmp, (vector_u8_t)zero, perm); \ + tmp2 = (vec_s16)vec_perm (tmp, (vec_u8)zero, perm); \ tmp3 = vec_adds (tmp2, src); \ tmp = vec_packsu (tmp3, tmp3); \ - vec_ste ((vector_u32_t)tmp, 0, (unsigned int *)dest); \ - vec_ste ((vector_u32_t)tmp, 4, (unsigned int *)dest); + vec_ste ((vec_u32)tmp, 0, (unsigned int *)dest); \ + vec_ste ((vec_u32)tmp, 4, (unsigned int *)dest); ADD (dest, vx0, perm0) dest += stride; ADD (dest, vx1, perm1) dest += stride; @@ -244,8 +214,5 @@ POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1); ADD (dest, vx5, perm1) dest += stride; ADD (dest, vx6, perm0) dest += stride; ADD (dest, vx7, perm1) - -POWERPC_PERF_STOP_COUNT(altivec_idct_add_num, 1); -#endif /* ALTIVEC_USE_REFERENCE_C_CODE */ }