summaryrefslogtreecommitdiff
path: root/src/libffmpeg/libavcodec/ppc
diff options
context:
space:
mode:
Diffstat (limited to 'src/libffmpeg/libavcodec/ppc')
-rw-r--r--src/libffmpeg/libavcodec/ppc/dsputil_altivec.c612
-rw-r--r--src/libffmpeg/libavcodec/ppc/dsputil_altivec.h77
-rw-r--r--src/libffmpeg/libavcodec/ppc/dsputil_ppc.c130
-rw-r--r--src/libffmpeg/libavcodec/ppc/dsputil_ppc.h21
-rw-r--r--src/libffmpeg/libavcodec/ppc/fdct_altivec.c15
-rw-r--r--src/libffmpeg/libavcodec/ppc/fft_altivec.c95
-rw-r--r--src/libffmpeg/libavcodec/ppc/gcc_fixes.h16
-rw-r--r--src/libffmpeg/libavcodec/ppc/gmc_altivec.c38
-rw-r--r--src/libffmpeg/libavcodec/ppc/idct_altivec.c24
-rw-r--r--src/libffmpeg/libavcodec/ppc/mpegvideo_altivec.c58
-rw-r--r--src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c14
11 files changed, 346 insertions, 754 deletions
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c
index 81a32c9e3..6f48893a4 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.c
@@ -3,18 +3,20 @@
* Copyright (c) 2002 Dieter Shirley
* Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -616,61 +618,28 @@ void diff_pixels_altivec(DCTELEM *restrict block, const uint8_t *s1,
}
void add_bytes_altivec(uint8_t *dst, uint8_t *src, int w) {
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int i;
- for(i=0; i+7<w; i++){
- dst[i+0] += src[i+0];
- dst[i+1] += src[i+1];
- dst[i+2] += src[i+2];
- dst[i+3] += src[i+3];
- dst[i+4] += src[i+4];
- dst[i+5] += src[i+5];
- dst[i+6] += src[i+6];
- dst[i+7] += src[i+7];
- }
- for(; i<w; i++)
- dst[i+0] += src[i+0];
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register int i;
register vector unsigned char vdst, vsrc;
/* dst and src are 16 bytes-aligned (guaranteed) */
- for(i = 0 ; (i + 15) < w ; i++)
+ for(i = 0 ; (i + 15) < w ; i+=16)
{
- vdst = vec_ld(i << 4, (unsigned char*)dst);
- vsrc = vec_ld(i << 4, (unsigned char*)src);
+ vdst = vec_ld(i, (unsigned char*)dst);
+ vsrc = vec_ld(i, (unsigned char*)src);
vdst = vec_add(vsrc, vdst);
- vec_st(vdst, i << 4, (unsigned char*)dst);
+ vec_st(vdst, i, (unsigned char*)dst);
}
/* if w is not a multiple of 16 */
for (; (i < w) ; i++)
{
dst[i] = src[i];
}
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 16) == 0) */
void put_pixels16_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_put_pixels16_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int i;
-
-POWERPC_PERF_START_COUNT(altivec_put_pixels16_num, 1);
-
- for(i=0; i<h; i++) {
- *((uint32_t*)(block)) = LD32(pixels);
- *((uint32_t*)(block+4)) = LD32(pixels+4);
- *((uint32_t*)(block+8)) = LD32(pixels+8);
- *((uint32_t*)(block+12)) = LD32(pixels+12);
- pixels+=line_size;
- block +=line_size;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register vector unsigned char pixelsv1, pixelsv2;
register vector unsigned char pixelsv1B, pixelsv2B;
register vector unsigned char pixelsv1C, pixelsv2C;
@@ -700,13 +669,13 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels16_num, 1);
#else
for(i=0; i<h; i+=4) {
pixelsv1 = vec_ld(0, (unsigned char*)pixels);
- pixelsv2 = vec_ld(16, (unsigned char*)pixels);
+ pixelsv2 = vec_ld(15, (unsigned char*)pixels);
pixelsv1B = vec_ld(line_size, (unsigned char*)pixels);
- pixelsv2B = vec_ld(16 + line_size, (unsigned char*)pixels);
+ pixelsv2B = vec_ld(15 + line_size, (unsigned char*)pixels);
pixelsv1C = vec_ld(line_size_2, (unsigned char*)pixels);
- pixelsv2C = vec_ld(16 + line_size_2, (unsigned char*)pixels);
+ pixelsv2C = vec_ld(15 + line_size_2, (unsigned char*)pixels);
pixelsv1D = vec_ld(line_size_3, (unsigned char*)pixels);
- pixelsv2D = vec_ld(16 + line_size_3, (unsigned char*)pixels);
+ pixelsv2D = vec_ld(15 + line_size_3, (unsigned char*)pixels);
vec_st(vec_perm(pixelsv1, pixelsv2, perm),
0, (unsigned char*)block);
vec_st(vec_perm(pixelsv1B, pixelsv2B, perm),
@@ -720,8 +689,6 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels16_num, 1);
}
#endif
POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_num, 1);
-
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 16) == 0) */
@@ -729,23 +696,6 @@ POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_num, 1);
void avg_pixels16_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_avg_pixels16_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int i;
-
-POWERPC_PERF_START_COUNT(altivec_avg_pixels16_num, 1);
-
- for(i=0; i<h; i++) {
- op_avg(*((uint32_t*)(block)),LD32(pixels));
- op_avg(*((uint32_t*)(block+4)),LD32(pixels+4));
- op_avg(*((uint32_t*)(block+8)),LD32(pixels+8));
- op_avg(*((uint32_t*)(block+12)),LD32(pixels+12));
- pixels+=line_size;
- block +=line_size;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_avg_pixels16_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register vector unsigned char pixelsv1, pixelsv2, pixelsv, blockv;
register vector unsigned char perm = vec_lvsl(0, pixels);
int i;
@@ -764,37 +714,12 @@ POWERPC_PERF_START_COUNT(altivec_avg_pixels16_num, 1);
}
POWERPC_PERF_STOP_COUNT(altivec_avg_pixels16_num, 1);
-
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 8) == 0) */
void avg_pixels8_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_avg_pixels8_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int i;
-POWERPC_PERF_START_COUNT(altivec_avg_pixels8_num, 1);
- for (i = 0; i < h; i++) {
- *((uint32_t *) (block)) =
- (((*((uint32_t *) (block))) |
- ((((const struct unaligned_32 *) (pixels))->l))) -
- ((((*((uint32_t *) (block))) ^
- ((((const struct unaligned_32 *) (pixels))->
- l))) & 0xFEFEFEFEUL) >> 1));
- *((uint32_t *) (block + 4)) =
- (((*((uint32_t *) (block + 4))) |
- ((((const struct unaligned_32 *) (pixels + 4))->l))) -
- ((((*((uint32_t *) (block + 4))) ^
- ((((const struct unaligned_32 *) (pixels +
- 4))->
- l))) & 0xFEFEFEFEUL) >> 1));
- pixels += line_size;
- block += line_size;
- }
-POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register vector unsigned char pixelsv1, pixelsv2, pixelsv, blockv;
int i;
@@ -830,52 +755,12 @@ POWERPC_PERF_START_COUNT(altivec_avg_pixels8_num, 1);
}
POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_num, 1);
-
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 8) == 0) */
void put_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_put_pixels8_xy2_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int j;
-POWERPC_PERF_START_COUNT(altivec_put_pixels8_xy2_num, 1);
- for (j = 0; j < 2; j++) {
- int i;
- const uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- const uint32_t b =
- (((const struct unaligned_32 *) (pixels + 1))->l);
- uint32_t l0 =
- (a & 0x03030303UL) + (b & 0x03030303UL) + 0x02020202UL;
- uint32_t h0 =
- ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- uint32_t l1, h1;
- pixels += line_size;
- for (i = 0; i < h; i += 2) {
- uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- uint32_t b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l1 = (a & 0x03030303UL) + (b & 0x03030303UL);
- h1 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- a = (((const struct unaligned_32 *) (pixels))->l);
- b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l0 = (a & 0x03030303UL) + (b & 0x03030303UL) + 0x02020202UL;
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- } pixels += 4 - line_size * (h + 1);
- block += 4 - line_size * h;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_put_pixels8_xy2_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register int i;
register vector unsigned char
pixelsv1, pixelsv2,
@@ -946,51 +831,12 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels8_xy2_num, 1);
}
POWERPC_PERF_STOP_COUNT(altivec_put_pixels8_xy2_num, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 8) == 0) */
void put_no_rnd_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels8_xy2_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int j;
-POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
- for (j = 0; j < 2; j++) {
- int i;
- const uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- const uint32_t b =
- (((const struct unaligned_32 *) (pixels + 1))->l);
- uint32_t l0 =
- (a & 0x03030303UL) + (b & 0x03030303UL) + 0x01010101UL;
- uint32_t h0 =
- ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- uint32_t l1, h1;
- pixels += line_size;
- for (i = 0; i < h; i += 2) {
- uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- uint32_t b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l1 = (a & 0x03030303UL) + (b & 0x03030303UL);
- h1 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- a = (((const struct unaligned_32 *) (pixels))->l);
- b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l0 = (a & 0x03030303UL) + (b & 0x03030303UL) + 0x01010101UL;
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- } pixels += 4 - line_size * (h + 1);
- block += 4 - line_size * h;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register int i;
register vector unsigned char
pixelsv1, pixelsv2,
@@ -1062,51 +908,12 @@ POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
}
POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels8_xy2_num, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 16) == 0) */
void put_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_put_pixels16_xy2_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int j;
-POWERPC_PERF_START_COUNT(altivec_put_pixels16_xy2_num, 1);
- for (j = 0; j < 4; j++) {
- int i;
- const uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- const uint32_t b =
- (((const struct unaligned_32 *) (pixels + 1))->l);
- uint32_t l0 =
- (a & 0x03030303UL) + (b & 0x03030303UL) + 0x02020202UL;
- uint32_t h0 =
- ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- uint32_t l1, h1;
- pixels += line_size;
- for (i = 0; i < h; i += 2) {
- uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- uint32_t b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l1 = (a & 0x03030303UL) + (b & 0x03030303UL);
- h1 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- a = (((const struct unaligned_32 *) (pixels))->l);
- b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l0 = (a & 0x03030303UL) + (b & 0x03030303UL) + 0x02020202UL;
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- } pixels += 4 - line_size * (h + 1);
- block += 4 - line_size * h;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_xy2_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register int i;
register vector unsigned char
pixelsv1, pixelsv2, pixelsv3, pixelsv4;
@@ -1183,51 +990,12 @@ POWERPC_PERF_START_COUNT(altivec_put_pixels16_xy2_num, 1);
}
POWERPC_PERF_STOP_COUNT(altivec_put_pixels16_xy2_num, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* next one assumes that ((line_size % 16) == 0) */
void put_no_rnd_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_put_no_rnd_pixels16_xy2_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int j;
-POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
- for (j = 0; j < 4; j++) {
- int i;
- const uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- const uint32_t b =
- (((const struct unaligned_32 *) (pixels + 1))->l);
- uint32_t l0 =
- (a & 0x03030303UL) + (b & 0x03030303UL) + 0x01010101UL;
- uint32_t h0 =
- ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- uint32_t l1, h1;
- pixels += line_size;
- for (i = 0; i < h; i += 2) {
- uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- uint32_t b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l1 = (a & 0x03030303UL) + (b & 0x03030303UL);
- h1 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- a = (((const struct unaligned_32 *) (pixels))->l);
- b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l0 = (a & 0x03030303UL) + (b & 0x03030303UL) + 0x01010101UL;
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) =
- h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL);
- pixels += line_size;
- block += line_size;
- } pixels += 4 - line_size * (h + 1);
- block += 4 - line_size * h;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
register int i;
register vector unsigned char
pixelsv1, pixelsv2, pixelsv3, pixelsv4;
@@ -1305,34 +1073,32 @@ POWERPC_PERF_START_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
}
POWERPC_PERF_STOP_COUNT(altivec_put_no_rnd_pixels16_xy2_num, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
int hadamard8_diff8x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
POWERPC_PERF_DECLARE(altivec_hadamard8_diff8x8_num, 1);
- int sum;
- register const_vector unsigned char vzero = (const_vector unsigned char)vec_splat_u8(0);
- register vector signed short temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
+ int sum;
+ register const_vector unsigned char vzero =
+ (const_vector unsigned char)vec_splat_u8(0);
+ register vector signed short temp0, temp1, temp2, temp3, temp4,
+ temp5, temp6, temp7;
POWERPC_PERF_START_COUNT(altivec_hadamard8_diff8x8_num, 1);
{
- register const_vector signed short vprod1 = (const_vector signed short)AVV( 1,-1, 1,-1, 1,-1, 1,-1);
- register const_vector signed short vprod2 = (const_vector signed short)AVV( 1, 1,-1,-1, 1, 1,-1,-1);
- register const_vector signed short vprod3 = (const_vector signed short)AVV( 1, 1, 1, 1,-1,-1,-1,-1);
+ register const_vector signed short vprod1 =(const_vector signed short)
+ AVV( 1,-1, 1,-1, 1,-1, 1,-1);
+ register const_vector signed short vprod2 =(const_vector signed short)
+ AVV( 1, 1,-1,-1, 1, 1,-1,-1);
+ register const_vector signed short vprod3 =(const_vector signed short)
+ AVV( 1, 1, 1, 1,-1,-1,-1,-1);
register const_vector unsigned char perm1 = (const_vector unsigned char)
- AVV(0x02, 0x03, 0x00, 0x01,
- 0x06, 0x07, 0x04, 0x05,
- 0x0A, 0x0B, 0x08, 0x09,
- 0x0E, 0x0F, 0x0C, 0x0D);
+ AVV(0x02, 0x03, 0x00, 0x01, 0x06, 0x07, 0x04, 0x05,
+ 0x0A, 0x0B, 0x08, 0x09, 0x0E, 0x0F, 0x0C, 0x0D);
register const_vector unsigned char perm2 = (const_vector unsigned char)
- AVV(0x04, 0x05, 0x06, 0x07,
- 0x00, 0x01, 0x02, 0x03,
- 0x0C, 0x0D, 0x0E, 0x0F,
- 0x08, 0x09, 0x0A, 0x0B);
+ AVV(0x04, 0x05, 0x06, 0x07, 0x00, 0x01, 0x02, 0x03,
+ 0x0C, 0x0D, 0x0E, 0x0F, 0x08, 0x09, 0x0A, 0x0B);
register const_vector unsigned char perm3 = (const_vector unsigned char)
- AVV(0x08, 0x09, 0x0A, 0x0B,
- 0x0C, 0x0D, 0x0E, 0x0F,
- 0x00, 0x01, 0x02, 0x03,
- 0x04, 0x05, 0x06, 0x07);
+ AVV(0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
+ 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07);
#define ONEITERBUTTERFLY(i, res) \
{ \
@@ -1443,45 +1209,46 @@ POWERPC_PERF_STOP_COUNT(altivec_hadamard8_diff8x8_num, 1);
*/
static int hadamard8_diff16x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h) {
- int sum;
- register vector signed short
- temp0 REG_v(v0),
- temp1 REG_v(v1),
- temp2 REG_v(v2),
- temp3 REG_v(v3),
- temp4 REG_v(v4),
- temp5 REG_v(v5),
- temp6 REG_v(v6),
- temp7 REG_v(v7);
- register vector signed short
- temp0S REG_v(v8),
- temp1S REG_v(v9),
- temp2S REG_v(v10),
- temp3S REG_v(v11),
- temp4S REG_v(v12),
- temp5S REG_v(v13),
- temp6S REG_v(v14),
- temp7S REG_v(v15);
- register const_vector unsigned char vzero REG_v(v31)= (const_vector unsigned char)vec_splat_u8(0);
+ int sum;
+ register vector signed short
+ temp0 REG_v(v0),
+ temp1 REG_v(v1),
+ temp2 REG_v(v2),
+ temp3 REG_v(v3),
+ temp4 REG_v(v4),
+ temp5 REG_v(v5),
+ temp6 REG_v(v6),
+ temp7 REG_v(v7);
+ register vector signed short
+ temp0S REG_v(v8),
+ temp1S REG_v(v9),
+ temp2S REG_v(v10),
+ temp3S REG_v(v11),
+ temp4S REG_v(v12),
+ temp5S REG_v(v13),
+ temp6S REG_v(v14),
+ temp7S REG_v(v15);
+ register const_vector unsigned char vzero REG_v(v31)=
+ (const_vector unsigned char)vec_splat_u8(0);
{
- register const_vector signed short vprod1 REG_v(v16)= (const_vector signed short)AVV( 1,-1, 1,-1, 1,-1, 1,-1);
- register const_vector signed short vprod2 REG_v(v17)= (const_vector signed short)AVV( 1, 1,-1,-1, 1, 1,-1,-1);
- register const_vector signed short vprod3 REG_v(v18)= (const_vector signed short)AVV( 1, 1, 1, 1,-1,-1,-1,-1);
- register const_vector unsigned char perm1 REG_v(v19)= (const_vector unsigned char)
- AVV(0x02, 0x03, 0x00, 0x01,
- 0x06, 0x07, 0x04, 0x05,
- 0x0A, 0x0B, 0x08, 0x09,
- 0x0E, 0x0F, 0x0C, 0x0D);
- register const_vector unsigned char perm2 REG_v(v20)= (const_vector unsigned char)
- AVV(0x04, 0x05, 0x06, 0x07,
- 0x00, 0x01, 0x02, 0x03,
- 0x0C, 0x0D, 0x0E, 0x0F,
- 0x08, 0x09, 0x0A, 0x0B);
- register const_vector unsigned char perm3 REG_v(v21)= (const_vector unsigned char)
- AVV(0x08, 0x09, 0x0A, 0x0B,
- 0x0C, 0x0D, 0x0E, 0x0F,
- 0x00, 0x01, 0x02, 0x03,
- 0x04, 0x05, 0x06, 0x07);
+ register const_vector signed short vprod1 REG_v(v16)=
+ (const_vector signed short)AVV( 1,-1, 1,-1, 1,-1, 1,-1);
+ register const_vector signed short vprod2 REG_v(v17)=
+ (const_vector signed short)AVV( 1, 1,-1,-1, 1, 1,-1,-1);
+ register const_vector signed short vprod3 REG_v(v18)=
+ (const_vector signed short)AVV( 1, 1, 1, 1,-1,-1,-1,-1);
+ register const_vector unsigned char perm1 REG_v(v19)=
+ (const_vector unsigned char)
+ AVV(0x02, 0x03, 0x00, 0x01, 0x06, 0x07, 0x04, 0x05,
+ 0x0A, 0x0B, 0x08, 0x09, 0x0E, 0x0F, 0x0C, 0x0D);
+ register const_vector unsigned char perm2 REG_v(v20)=
+ (const_vector unsigned char)
+ AVV(0x04, 0x05, 0x06, 0x07, 0x00, 0x01, 0x02, 0x03,
+ 0x0C, 0x0D, 0x0E, 0x0F, 0x08, 0x09, 0x0A, 0x0B);
+ register const_vector unsigned char perm3 REG_v(v21)=
+ (const_vector unsigned char)
+ AVV(0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
+ 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07);
#define ONEITERBUTTERFLY(i, res1, res2) \
{ \
@@ -1642,27 +1409,27 @@ static int hadamard8_diff16x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst,
int hadamard8_diff16_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
POWERPC_PERF_DECLARE(altivec_hadamard8_diff16_num, 1);
- int score;
+ int score;
POWERPC_PERF_START_COUNT(altivec_hadamard8_diff16_num, 1);
- score = hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
- if (h==16) {
- dst += 8*stride;
- src += 8*stride;
- score += hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
- }
+ score = hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
+ if (h==16) {
+ dst += 8*stride;
+ src += 8*stride;
+ score += hadamard8_diff16x8_altivec(s, dst, src, stride, 8);
+ }
POWERPC_PERF_STOP_COUNT(altivec_hadamard8_diff16_num, 1);
- return score;
+ return score;
}
int has_altivec(void)
{
#ifdef __AMIGAOS4__
- ULONG result = 0;
- extern struct ExecIFace *IExec;
+ ULONG result = 0;
+ extern struct ExecIFace *IExec;
- IExec->GetCPUInfoTags(GCIT_VectorUnit, &result, TAG_DONE);
- if (result == VECTORTYPE_ALTIVEC) return 1;
- return 0;
+ IExec->GetCPUInfoTags(GCIT_VectorUnit, &result, TAG_DONE);
+ if (result == VECTORTYPE_ALTIVEC) return 1;
+ return 0;
#else /* __AMIGAOS4__ */
#ifdef CONFIG_DARWIN
@@ -1698,112 +1465,127 @@ int has_altivec(void)
#endif /* __AMIGAOS4__ */
}
+static void vorbis_inverse_coupling_altivec(float *mag, float *ang,
+ int blocksize)
+{
+ int i;
+ vector float m, a;
+ vector bool int t0, t1;
+ const vector unsigned int v_31 = //XXX
+ vec_add(vec_add(vec_splat_u32(15),vec_splat_u32(15)),vec_splat_u32(1));
+ for(i=0; i<blocksize; i+=4) {
+ m = vec_ld(0, mag+i);
+ a = vec_ld(0, ang+i);
+ t0 = vec_cmple(m, (vector float)vec_splat_u32(0));
+ t1 = vec_cmple(a, (vector float)vec_splat_u32(0));
+ a = vec_xor(a, (vector float) vec_sl((vector unsigned int)t0, v_31));
+ t0 = (vector bool int)vec_and(a, t1);
+ t1 = (vector bool int)vec_andc(a, t1);
+ a = vec_sub(m, (vector float)t1);
+ m = vec_add(m, (vector float)t0);
+ vec_stl(a, 0, ang+i);
+ vec_stl(m, 0, mag+i);
+ }
+}
+
/* next one assumes that ((line_size % 8) == 0) */
void avg_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h)
{
POWERPC_PERF_DECLARE(altivec_avg_pixels8_xy2_num, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
-
- int j;
-POWERPC_PERF_START_COUNT(altivec_avg_pixels8_xy2_num, 1);
- for (j = 0; j < 2; j++) {
- int i;
- const uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- const uint32_t b = (((const struct unaligned_32 *) (pixels + 1))->l);
- uint32_t l0 = (a & 0x03030303UL) + (b & 0x03030303UL) + 0x02020202UL;
- uint32_t h0 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- uint32_t l1, h1;
- pixels += line_size;
- for (i = 0; i < h; i += 2) {
- uint32_t a = (((const struct unaligned_32 *) (pixels))->l);
- uint32_t b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l1 = (a & 0x03030303UL) + (b & 0x03030303UL);
- h1 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) = rnd_avg32(*((uint32_t *) block), h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL));
- pixels += line_size;
- block += line_size;
- a = (((const struct unaligned_32 *) (pixels))->l);
- b = (((const struct unaligned_32 *) (pixels + 1))->l);
- l0 = (a & 0x03030303UL) + (b & 0x03030303UL) + 0x02020202UL;
- h0 = ((a & 0xFCFCFCFCUL) >> 2) + ((b & 0xFCFCFCFCUL) >> 2);
- *((uint32_t *) block) = rnd_avg32(*((uint32_t *) block), h0 + h1 + (((l0 + l1) >> 2) & 0x0F0F0F0FUL));
- pixels += line_size;
- block += line_size;
- } pixels += 4 - line_size * (h + 1);
- block += 4 - line_size * h;
- }
-POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_xy2_num, 1);
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
- register int i;
- register vector unsigned char
- pixelsv1, pixelsv2,
- pixelsavg;
- register vector unsigned char
- blockv, temp1, temp2, blocktemp;
- register vector unsigned short
- pixelssum1, pixelssum2, temp3;
- register const_vector unsigned char vczero = (const_vector unsigned char)vec_splat_u8(0);
- register const_vector unsigned short vctwo = (const_vector unsigned short)vec_splat_u16(2);
-
- temp1 = vec_ld(0, pixels);
- temp2 = vec_ld(16, pixels);
- pixelsv1 = vec_perm(temp1, temp2, vec_lvsl(0, pixels));
- if ((((unsigned long)pixels) & 0x0000000F) == 0x0000000F)
- {
- pixelsv2 = temp2;
- }
- else
- {
- pixelsv2 = vec_perm(temp1, temp2, vec_lvsl(1, pixels));
- }
- pixelsv1 = vec_mergeh(vczero, pixelsv1);
- pixelsv2 = vec_mergeh(vczero, pixelsv2);
- pixelssum1 = vec_add((vector unsigned short)pixelsv1,
- (vector unsigned short)pixelsv2);
- pixelssum1 = vec_add(pixelssum1, vctwo);
+ register int i;
+ register vector unsigned char pixelsv1, pixelsv2, pixelsavg;
+ register vector unsigned char blockv, temp1, temp2, blocktemp;
+ register vector unsigned short pixelssum1, pixelssum2, temp3;
+
+ register const_vector unsigned char vczero = (const_vector unsigned char)
+ vec_splat_u8(0);
+ register const_vector unsigned short vctwo = (const_vector unsigned short)
+ vec_splat_u16(2);
+
+ temp1 = vec_ld(0, pixels);
+ temp2 = vec_ld(16, pixels);
+ pixelsv1 = vec_perm(temp1, temp2, vec_lvsl(0, pixels));
+ if ((((unsigned long)pixels) & 0x0000000F) == 0x0000000F) {
+ pixelsv2 = temp2;
+ } else {
+ pixelsv2 = vec_perm(temp1, temp2, vec_lvsl(1, pixels));
+ }
+ pixelsv1 = vec_mergeh(vczero, pixelsv1);
+ pixelsv2 = vec_mergeh(vczero, pixelsv2);
+ pixelssum1 = vec_add((vector unsigned short)pixelsv1,
+ (vector unsigned short)pixelsv2);
+ pixelssum1 = vec_add(pixelssum1, vctwo);
POWERPC_PERF_START_COUNT(altivec_avg_pixels8_xy2_num, 1);
- for (i = 0; i < h ; i++) {
- int rightside = ((unsigned long)block & 0x0000000F);
- blockv = vec_ld(0, block);
+ for (i = 0; i < h ; i++) {
+ int rightside = ((unsigned long)block & 0x0000000F);
+ blockv = vec_ld(0, block);
+
+ temp1 = vec_ld(line_size, pixels);
+ temp2 = vec_ld(line_size + 16, pixels);
+ pixelsv1 = vec_perm(temp1, temp2, vec_lvsl(line_size, pixels));
+ if (((((unsigned long)pixels) + line_size) & 0x0000000F) == 0x0000000F)
+ {
+ pixelsv2 = temp2;
+ } else {
+ pixelsv2 = vec_perm(temp1, temp2, vec_lvsl(line_size + 1, pixels));
+ }
+
+ pixelsv1 = vec_mergeh(vczero, pixelsv1);
+ pixelsv2 = vec_mergeh(vczero, pixelsv2);
+ pixelssum2 = vec_add((vector unsigned short)pixelsv1,
+ (vector unsigned short)pixelsv2);
+ temp3 = vec_add(pixelssum1, pixelssum2);
+ temp3 = vec_sra(temp3, vctwo);
+ pixelssum1 = vec_add(pixelssum2, vctwo);
+ pixelsavg = vec_packsu(temp3, (vector unsigned short) vczero);
+
+ if (rightside) {
+ blocktemp = vec_perm(blockv, pixelsavg, vcprm(0, 1, s0, s1));
+ } else {
+ blocktemp = vec_perm(blockv, pixelsavg, vcprm(s0, s1, 2, 3));
+ }
+
+ blockv = vec_avg(blocktemp, blockv);
+ vec_st(blockv, 0, block);
- temp1 = vec_ld(line_size, pixels);
- temp2 = vec_ld(line_size + 16, pixels);
- pixelsv1 = vec_perm(temp1, temp2, vec_lvsl(line_size, pixels));
- if (((((unsigned long)pixels) + line_size) & 0x0000000F) == 0x0000000F)
- {
- pixelsv2 = temp2;
- }
- else
- {
- pixelsv2 = vec_perm(temp1, temp2, vec_lvsl(line_size + 1, pixels));
- }
-
- pixelsv1 = vec_mergeh(vczero, pixelsv1);
- pixelsv2 = vec_mergeh(vczero, pixelsv2);
- pixelssum2 = vec_add((vector unsigned short)pixelsv1,
- (vector unsigned short)pixelsv2);
- temp3 = vec_add(pixelssum1, pixelssum2);
- temp3 = vec_sra(temp3, vctwo);
- pixelssum1 = vec_add(pixelssum2, vctwo);
- pixelsavg = vec_packsu(temp3, (vector unsigned short) vczero);
-
- if (rightside)
- {
- blocktemp = vec_perm(blockv, pixelsavg, vcprm(0, 1, s0, s1));
- }
- else
- {
- blocktemp = vec_perm(blockv, pixelsavg, vcprm(s0, s1, 2, 3));
- }
-
- blockv = vec_avg(blocktemp, blockv);
- vec_st(blockv, 0, block);
-
- block += line_size;
- pixels += line_size;
- }
+ block += line_size;
+ pixels += line_size;
+ }
POWERPC_PERF_STOP_COUNT(altivec_avg_pixels8_xy2_num, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
+}
+
+void dsputil_init_altivec(DSPContext* c, AVCodecContext *avctx)
+{
+ c->pix_abs[0][1] = sad16_x2_altivec;
+ c->pix_abs[0][2] = sad16_y2_altivec;
+ c->pix_abs[0][3] = sad16_xy2_altivec;
+ c->pix_abs[0][0] = sad16_altivec;
+ c->pix_abs[1][0] = sad8_altivec;
+ c->sad[0]= sad16_altivec;
+ c->sad[1]= sad8_altivec;
+ c->pix_norm1 = pix_norm1_altivec;
+ c->sse[1]= sse8_altivec;
+ c->sse[0]= sse16_altivec;
+ c->pix_sum = pix_sum_altivec;
+ c->diff_pixels = diff_pixels_altivec;
+ c->get_pixels = get_pixels_altivec;
+ c->add_bytes= add_bytes_altivec;
+ c->put_pixels_tab[0][0] = put_pixels16_altivec;
+ /* the two functions do the same thing, so use the same code */
+ c->put_no_rnd_pixels_tab[0][0] = put_pixels16_altivec;
+ c->avg_pixels_tab[0][0] = avg_pixels16_altivec;
+ c->avg_pixels_tab[1][0] = avg_pixels8_altivec;
+ c->avg_pixels_tab[1][3] = avg_pixels8_xy2_altivec;
+ c->put_pixels_tab[1][3] = put_pixels8_xy2_altivec;
+ c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_altivec;
+ c->put_pixels_tab[0][3] = put_pixels16_xy2_altivec;
+ c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_altivec;
+
+ c->hadamard8_diff[0] = hadamard8_diff16_altivec;
+ c->hadamard8_diff[1] = hadamard8_diff8x8_altivec;
+#ifdef CONFIG_VORBIS_DECODER
+ c->vorbis_inverse_coupling = vorbis_inverse_coupling_altivec;
+#endif
}
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h
index ac54817d0..560d778bb 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_altivec.h
@@ -3,18 +3,20 @@
* Copyright (c) 2002 Dieter Shirley
* Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -25,34 +27,11 @@
#ifdef HAVE_ALTIVEC
-extern int sad16_x2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int sad16_y2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int sad16_xy2_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int sad16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int sad8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int pix_norm1_altivec(uint8_t *pix, int line_size);
-extern int sse8_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int sse16_altivec(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h);
-extern int pix_sum_altivec(uint8_t * pix, int line_size);
-extern void diff_pixels_altivec(DCTELEM* block, const uint8_t* s1, const uint8_t* s2, int stride);
-extern void get_pixels_altivec(DCTELEM* block, const uint8_t * pixels, int line_size);
+extern int has_altivec(void);
-extern void add_bytes_altivec(uint8_t *dst, uint8_t *src, int w);
-extern void put_pixels_clamped_altivec(const DCTELEM *block, uint8_t *restrict pixels, int line_size);
-extern void put_pixels16_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
-extern void avg_pixels16_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
-extern void avg_pixels8_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h);
-extern void put_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
-extern void put_no_rnd_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
-extern void put_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h);
-extern void put_no_rnd_pixels16_xy2_altivec(uint8_t * block, const uint8_t * pixels, int line_size, int h);
-extern int hadamard8_diff8x8_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h);
-extern int hadamard8_diff16_altivec(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h);
-extern void avg_pixels8_xy2_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
+void put_pixels16_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
-extern void gmc1_altivec(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y16, int rounder);
-
-extern int has_altivec(void);
+void avg_pixels16_altivec(uint8_t *block, const uint8_t *pixels, int line_size, int h);
// used to build registers permutation vectors (vcprm)
// the 's' are for words in the _s_econd vector
@@ -88,10 +67,40 @@ extern int has_altivec(void);
#define vcii(a,b,c,d) (const vector float){FLOAT_ ## a, FLOAT_ ## b, FLOAT_ ## c, FLOAT_ ## d}
#endif
-#else /* HAVE_ALTIVEC */
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
-#error "I can't use ALTIVEC_USE_REFERENCE_C_CODE if I don't use HAVE_ALTIVEC"
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
+// Transpose 8x8 matrix of 16-bit elements (in-place)
+#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)
+
#endif /* HAVE_ALTIVEC */
#endif /* _DSPUTIL_ALTIVEC_ */
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c
index b63c8dd84..9169eaef0 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.c
@@ -3,18 +3,20 @@
* Copyright (c) 2002 Dieter Shirley
* Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -24,22 +26,21 @@
#ifdef HAVE_ALTIVEC
#include "dsputil_altivec.h"
-#endif
extern void fdct_altivec(int16_t *block);
+extern void gmc1_altivec(uint8_t *dst, uint8_t *src, int stride, int h,
+ int x16, int y16, int rounder);
extern void idct_put_altivec(uint8_t *dest, int line_size, int16_t *block);
extern void idct_add_altivec(uint8_t *dest, int line_size, int16_t *block);
-extern void ff_snow_horizontal_compose97i_altivec(DWTELEM *b, int width);
-extern void ff_snow_vertical_compose97i_altivec(DWTELEM *b0, DWTELEM *b1,
- DWTELEM *b2, DWTELEM *b3,
- DWTELEM *b4, DWTELEM *b5,
- int width);
-extern void ff_snow_inner_add_yblock_altivec(uint8_t *obmc, const int obmc_stride,
- uint8_t * * block, int b_w, int b_h,
- int src_x, int src_y, int src_stride,
- slice_buffer * sb, int add,
- uint8_t * dst8);
+void dsputil_h264_init_ppc(DSPContext* c, AVCodecContext *avctx);
+
+void dsputil_init_altivec(DSPContext* c, AVCodecContext *avctx);
+void vc1dsp_init_altivec(DSPContext* c, AVCodecContext *avctx);
+void snow_init_altivec(DSPContext* c, AVCodecContext *avctx);
+void float_init_altivec(DSPContext* c, AVCodecContext *avctx);
+
+#endif
int mm_flags = 0;
@@ -100,7 +101,7 @@ void powerpc_display_perf_report(void)
{
if (perfdata[j][i][powerpc_data_num] != (unsigned long long)0)
av_log(NULL, AV_LOG_INFO,
- " Function \"%s\" (pmc%d):\n\tmin: %llu\n\tmax: %llu\n\tavg: %1.2lf (%llu)\n",
+ " Function \"%s\" (pmc%d):\n\tmin: %"PRIu64"\n\tmax: %"PRIu64"\n\tavg: %1.2lf (%"PRIu64")\n",
perfname[i],
j+1,
perfdata[j][i][powerpc_data_min],
@@ -174,7 +175,7 @@ POWERPC_PERF_STOP_COUNT(powerpc_clear_blocks_dcbz32, 1);
/* same as above, when dcbzl clear a whole 128B cache line
i.e. the PPC970 aka G5 */
-#ifndef NO_DCBZL
+#ifdef HAVE_DCBZL
void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
{
POWERPC_PERF_DECLARE(powerpc_clear_blocks_dcbz128, 1);
@@ -204,7 +205,7 @@ void clear_blocks_dcbz128_ppc(DCTELEM *blocks)
}
#endif
-#ifndef NO_DCBZL
+#ifdef HAVE_DCBZL
/* check dcbz report how many bytes are set to 0 by dcbz */
/* update 24/06/2003 : replace dcbz by dcbzl to get
the intended effect (Apple "fixed" dcbz)
@@ -248,69 +249,43 @@ long check_dcbzl_effect(void)
}
#endif
-
-void dsputil_h264_init_ppc(DSPContext* c, AVCodecContext *avctx);
+static void prefetch_ppc(void *mem, int stride, int h)
+{
+ register const uint8_t *p = mem;
+ do {
+ asm volatile ("dcbt 0,%0" : : "r" (p));
+ p+= stride;
+ } while(--h);
+}
void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
{
// Common optimizations whether Altivec is available or not
-
- switch (check_dcbzl_effect()) {
- case 32:
- c->clear_blocks = clear_blocks_dcbz32_ppc;
- break;
- case 128:
- c->clear_blocks = clear_blocks_dcbz128_ppc;
- break;
- default:
- break;
- }
+ c->prefetch = prefetch_ppc;
+ switch (check_dcbzl_effect()) {
+ case 32:
+ c->clear_blocks = clear_blocks_dcbz32_ppc;
+ break;
+ case 128:
+ c->clear_blocks = clear_blocks_dcbz128_ppc;
+ break;
+ default:
+ break;
+ }
#ifdef HAVE_ALTIVEC
- dsputil_h264_init_ppc(c, avctx);
+ if(ENABLE_H264_DECODER) dsputil_h264_init_ppc(c, avctx);
if (has_altivec()) {
mm_flags |= MM_ALTIVEC;
- // Altivec specific optimisations
- c->pix_abs[0][1] = sad16_x2_altivec;
- c->pix_abs[0][2] = sad16_y2_altivec;
- c->pix_abs[0][3] = sad16_xy2_altivec;
- c->pix_abs[0][0] = sad16_altivec;
- c->pix_abs[1][0] = sad8_altivec;
- c->sad[0]= sad16_altivec;
- c->sad[1]= sad8_altivec;
- c->pix_norm1 = pix_norm1_altivec;
- c->sse[1]= sse8_altivec;
- c->sse[0]= sse16_altivec;
- c->pix_sum = pix_sum_altivec;
- c->diff_pixels = diff_pixels_altivec;
- c->get_pixels = get_pixels_altivec;
-// next one disabled as it's untested.
-#if 0
- c->add_bytes= add_bytes_altivec;
-#endif /* 0 */
- c->put_pixels_tab[0][0] = put_pixels16_altivec;
- /* the two functions do the same thing, so use the same code */
- c->put_no_rnd_pixels_tab[0][0] = put_pixels16_altivec;
- c->avg_pixels_tab[0][0] = avg_pixels16_altivec;
- c->avg_pixels_tab[1][0] = avg_pixels8_altivec;
- c->avg_pixels_tab[1][3] = avg_pixels8_xy2_altivec;
- c->put_pixels_tab[1][3] = put_pixels8_xy2_altivec;
- c->put_no_rnd_pixels_tab[1][3] = put_no_rnd_pixels8_xy2_altivec;
- c->put_pixels_tab[0][3] = put_pixels16_xy2_altivec;
- c->put_no_rnd_pixels_tab[0][3] = put_no_rnd_pixels16_xy2_altivec;
-
+ dsputil_init_altivec(c, avctx);
+ if(ENABLE_SNOW_DECODER) snow_init_altivec(c, avctx);
+ if(ENABLE_VC1_DECODER || ENABLE_WMV3_DECODER)
+ vc1dsp_init_altivec(c, avctx);
+ float_init_altivec(c, avctx);
c->gmc1 = gmc1_altivec;
- c->hadamard8_diff[0] = hadamard8_diff16_altivec;
- c->hadamard8_diff[1] = hadamard8_diff8x8_altivec;
-
-
- c->horizontal_compose97i = ff_snow_horizontal_compose97i_altivec;
- c->vertical_compose97i = ff_snow_vertical_compose97i_altivec;
- c->inner_add_yblock = ff_snow_inner_add_yblock_altivec;
-
#ifdef CONFIG_ENCODERS
if (avctx->dct_algo == FF_DCT_AUTO ||
avctx->dct_algo == FF_DCT_ALTIVEC)
@@ -319,20 +294,16 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
}
#endif //CONFIG_ENCODERS
- if (avctx->lowres==0)
- {
+ if (avctx->lowres==0)
+ {
if ((avctx->idct_algo == FF_IDCT_AUTO) ||
(avctx->idct_algo == FF_IDCT_ALTIVEC))
{
c->idct_put = idct_put_altivec;
c->idct_add = idct_add_altivec;
-#ifndef ALTIVEC_USE_REFERENCE_C_CODE
c->idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
- c->idct_permutation_type = FF_NO_IDCT_PERM;
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
- }
+ }
#ifdef POWERPC_PERFORMANCE_REPORT
{
@@ -349,11 +320,6 @@ void dsputil_init_ppc(DSPContext* c, AVCodecContext *avctx)
}
}
#endif /* POWERPC_PERFORMANCE_REPORT */
- } else
-#endif /* HAVE_ALTIVEC */
- {
- // Non-AltiVec PPC optimisations
-
- // ... pending ...
}
+#endif /* HAVE_ALTIVEC */
}
diff --git a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h
index 966ffa71a..ab2b05780 100644
--- a/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h
+++ b/src/libffmpeg/libavcodec/ppc/dsputil_ppc.h
@@ -1,35 +1,26 @@
/*
* Copyright (c) 2003-2004 Romain Dolbeau <romain@dolbeau.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef _DSPUTIL_PPC_
#define _DSPUTIL_PPC_
-#ifdef CONFIG_DARWIN
-/* The Apple assembler shipped w/ gcc-3.3 knows about DCBZL, previous assemblers don't
- We assume here that the Darwin GCC is from Apple.... */
-#if (__GNUC__ * 100 + __GNUC_MINOR__ < 303)
-#define NO_DCBZL
-#endif
-#else /* CONFIG_DARWIN */
-/* I don't think any non-Apple assembler knows about DCBZL */
-#define NO_DCBZL
-#endif /* CONFIG_DARWIN */
-
#ifdef POWERPC_PERFORMANCE_REPORT
void powerpc_display_perf_report(void);
/* the 604* have 2, the G3* have 4, the G4s have 6,
diff --git a/src/libffmpeg/libavcodec/ppc/fdct_altivec.c b/src/libffmpeg/libavcodec/ppc/fdct_altivec.c
index f5778c24e..2418c32bb 100644
--- a/src/libffmpeg/libavcodec/ppc/fdct_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/fdct_altivec.c
@@ -2,18 +2,20 @@
* AltiVec optimized library for the FFMPEG Multimedia System
* Copyright (C) 2003 James Klicman <james@klicman.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -196,12 +198,6 @@ static vector float fdctconsts[3] = {
void fdct_altivec(int16_t *block)
{
POWERPC_PERF_DECLARE(altivec_fdct, 1);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
-POWERPC_PERF_START_COUNT(altivec_fdct, 1);
- void ff_jpeg_fdct_islow(int16_t *block);
- ff_jpeg_fdct_islow(block);
-POWERPC_PERF_STOP_COUNT(altivec_fdct, 1);
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
vector signed short *bp;
vector float *cp;
vector float b00, b10, b20, b30, b40, b50, b60, b70;
@@ -492,7 +488,6 @@ POWERPC_PERF_STOP_COUNT(altivec_fdct, 1);
/* }}} */
POWERPC_PERF_STOP_COUNT(altivec_fdct, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
/* vim:set foldmethod=marker foldlevel=0: */
diff --git a/src/libffmpeg/libavcodec/ppc/fft_altivec.c b/src/libffmpeg/libavcodec/ppc/fft_altivec.c
index f4ea78359..384a774ff 100644
--- a/src/libffmpeg/libavcodec/ppc/fft_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/fft_altivec.c
@@ -4,18 +4,20 @@
* Copyright (c) 2003 Romain Dolbeau <romain@dolbeau.org>
* Based on code Copyright (c) 2002 Fabrice Bellard.
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "../dsputil.h"
@@ -63,88 +65,7 @@
void ff_fft_calc_altivec(FFTContext *s, FFTComplex *z)
{
POWERPC_PERF_DECLARE(altivec_fft_num, s->nbits >= 6);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- int ln = s->nbits;
- int j, np, np2;
- int nblocks, nloops;
- register FFTComplex *p, *q;
- FFTComplex *exptab = s->exptab;
- int l;
- FFTSample tmp_re, tmp_im;
-
-POWERPC_PERF_START_COUNT(altivec_fft_num, s->nbits >= 6);
-
- np = 1 << ln;
-
- /* pass 0 */
-
- p=&z[0];
- j=(np >> 1);
- do {
- BF(p[0].re, p[0].im, p[1].re, p[1].im,
- p[0].re, p[0].im, p[1].re, p[1].im);
- p+=2;
- } while (--j != 0);
-
- /* pass 1 */
-
-
- p=&z[0];
- j=np >> 2;
- if (s->inverse) {
- do {
- BF(p[0].re, p[0].im, p[2].re, p[2].im,
- p[0].re, p[0].im, p[2].re, p[2].im);
- BF(p[1].re, p[1].im, p[3].re, p[3].im,
- p[1].re, p[1].im, -p[3].im, p[3].re);
- p+=4;
- } while (--j != 0);
- } else {
- do {
- BF(p[0].re, p[0].im, p[2].re, p[2].im,
- p[0].re, p[0].im, p[2].re, p[2].im);
- BF(p[1].re, p[1].im, p[3].re, p[3].im,
- p[1].re, p[1].im, p[3].im, -p[3].re);
- p+=4;
- } while (--j != 0);
- }
- /* pass 2 .. ln-1 */
-
- nblocks = np >> 3;
- nloops = 1 << 2;
- np2 = np >> 1;
- do {
- p = z;
- q = z + nloops;
- for (j = 0; j < nblocks; ++j) {
- BF(p->re, p->im, q->re, q->im,
- p->re, p->im, q->re, q->im);
-
- p++;
- q++;
- for(l = nblocks; l < np2; l += nblocks) {
- CMUL(tmp_re, tmp_im, exptab[l].re, exptab[l].im, q->re, q->im);
- BF(p->re, p->im, q->re, q->im,
- p->re, p->im, tmp_re, tmp_im);
- p++;
- q++;
- }
-
- p += nloops;
- q += nloops;
- }
- nblocks = nblocks >> 1;
- nloops = nloops << 1;
- } while (nblocks != 0);
-
-POWERPC_PERF_STOP_COUNT(altivec_fft_num, s->nbits >= 6);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
-#ifdef CONFIG_DARWIN
- register const vector float vczero = (const vector float)(0.);
-#else
- register const vector float vczero = (const vector float){0.,0.,0.,0.};
-#endif
+ register const vector float vczero = (const vector float)vec_splat_u32(0.);
int ln = s->nbits;
int j, np, np2;
@@ -242,6 +163,4 @@ POWERPC_PERF_START_COUNT(altivec_fft_num, s->nbits >= 6);
} while (nblocks != 0);
POWERPC_PERF_STOP_COUNT(altivec_fft_num, s->nbits >= 6);
-
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
diff --git a/src/libffmpeg/libavcodec/ppc/gcc_fixes.h b/src/libffmpeg/libavcodec/ppc/gcc_fixes.h
index 943905bc5..5a4a55188 100644
--- a/src/libffmpeg/libavcodec/ppc/gcc_fixes.h
+++ b/src/libffmpeg/libavcodec/ppc/gcc_fixes.h
@@ -2,6 +2,22 @@
* gcc fixes for altivec.
* Used to workaround broken gcc (FSF gcc-3 pre gcc-3.3)
* and to stay somewhat compatible with Darwin.
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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,
+ * 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
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#ifndef _GCC_FIXES_
diff --git a/src/libffmpeg/libavcodec/ppc/gmc_altivec.c b/src/libffmpeg/libavcodec/ppc/gmc_altivec.c
index 04978d825..42c936bb3 100644
--- a/src/libffmpeg/libavcodec/ppc/gmc_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/gmc_altivec.c
@@ -3,18 +3,20 @@
* AltiVec-enabled
* Copyright (c) 2003 Romain Dolbeau <romain@dolbeau.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -32,32 +34,6 @@
void gmc1_altivec(uint8_t *dst /* align 8 */, uint8_t *src /* align1 */, int stride, int h, int x16, int y16, int rounder)
{
POWERPC_PERF_DECLARE(altivec_gmc1_num, GMC1_PERF_COND);
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- const int A=(16-x16)*(16-y16);
- const int B=( x16)*(16-y16);
- const int C=(16-x16)*( y16);
- const int D=( x16)*( y16);
- int i;
-
-POWERPC_PERF_START_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
-
- for(i=0; i<h; i++)
- {
- dst[0]= (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + rounder)>>8;
- dst[1]= (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + rounder)>>8;
- dst[2]= (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + rounder)>>8;
- dst[3]= (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + rounder)>>8;
- dst[4]= (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + rounder)>>8;
- dst[5]= (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + rounder)>>8;
- dst[6]= (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + rounder)>>8;
- dst[7]= (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + rounder)>>8;
- dst+= stride;
- src+= stride;
- }
-
-POWERPC_PERF_STOP_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
-
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
const unsigned short __attribute__ ((aligned(16))) rounder_a[8] =
{rounder, rounder, rounder, rounder,
rounder, rounder, rounder, rounder};
@@ -167,6 +143,4 @@ POWERPC_PERF_START_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
}
POWERPC_PERF_STOP_COUNT(altivec_gmc1_num, GMC1_PERF_COND);
-
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
diff --git a/src/libffmpeg/libavcodec/ppc/idct_altivec.c b/src/libffmpeg/libavcodec/ppc/idct_altivec.c
index 93d63cfd3..cee46fc25 100644
--- a/src/libffmpeg/libavcodec/ppc/idct_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/idct_altivec.c
@@ -1,18 +1,20 @@
/*
* Copyright (c) 2001 Michel Lespinasse
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*
*/
@@ -167,12 +169,6 @@ static const_vector_s16_t constants[5] = {
void idct_put_altivec(uint8_t* dest, int stride, vector_s16_t* block)
{
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
@@ -195,18 +191,11 @@ POWERPC_PERF_START_COUNT(altivec_idct_put_num, 1);
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)
{
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;
@@ -244,6 +233,5 @@ POWERPC_PERF_START_COUNT(altivec_idct_add_num, 1);
ADD (dest, vx7, perm1)
POWERPC_PERF_STOP_COUNT(altivec_idct_add_num, 1);
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
diff --git a/src/libffmpeg/libavcodec/ppc/mpegvideo_altivec.c b/src/libffmpeg/libavcodec/ppc/mpegvideo_altivec.c
index 7a771a8ec..3822cb20e 100644
--- a/src/libffmpeg/libavcodec/ppc/mpegvideo_altivec.c
+++ b/src/libffmpeg/libavcodec/ppc/mpegvideo_altivec.c
@@ -4,18 +4,20 @@
* dct_unquantize_h263_altivec:
* Copyright (c) 2003 Romain Dolbeau <romain@dolbeau.org>
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -50,39 +52,6 @@ do { \
d = vec_mergel(_trans_acl, _trans_bdl); \
} while (0)
-#define TRANSPOSE8(a,b,c,d,e,f,g,h) \
-do { \
- __typeof__(a) _A1, _B1, _C1, _D1, _E1, _F1, _G1, _H1; \
- __typeof__(a) _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)
-
// Loads a four-byte value (int or float) from the target address
// into every element in the target vector. Only works if the
@@ -552,19 +521,6 @@ POWERPC_PERF_START_COUNT(altivec_dct_unquantize_h263_num, 1);
nCoeffs= s->intra_scantable.raster_end[ s->block_last_index[n] ];
}
-#ifdef ALTIVEC_USE_REFERENCE_C_CODE
- for(;i<=nCoeffs;i++) {
- level = block[i];
- if (level) {
- if (level < 0) {
- level = level * qmul - qadd;
- } else {
- level = level * qmul + qadd;
- }
- block[i] = level;
- }
- }
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
{
register const_vector signed short vczero = (const_vector signed short)vec_splat_s16(0);
short __attribute__ ((aligned(16))) qmul8[] =
@@ -643,7 +599,5 @@ POWERPC_PERF_START_COUNT(altivec_dct_unquantize_h263_num, 1);
block[0] = backup_0;
}
}
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
-
POWERPC_PERF_STOP_COUNT(altivec_dct_unquantize_h263_num, nCoeffs == 63);
}
diff --git a/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c b/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c
index b391b4294..c5e822f77 100644
--- a/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c
+++ b/src/libffmpeg/libavcodec/ppc/mpegvideo_ppc.c
@@ -1,18 +1,20 @@
/*
* Copyright (c) 2002 Dieter Shirley
*
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg 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 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
*
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg 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 this library; if not, write to the Free Software
+ * License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
@@ -46,11 +48,7 @@ void MPV_common_init_ppc(MpegEncContext *s)
{
s->dsp.idct_put = idct_put_altivec;
s->dsp.idct_add = idct_add_altivec;
-#ifndef ALTIVEC_USE_REFERENCE_C_CODE
s->dsp.idct_permutation_type = FF_TRANSPOSE_IDCT_PERM;
-#else /* ALTIVEC_USE_REFERENCE_C_CODE */
- s->dsp.idct_permutation_type = FF_NO_IDCT_PERM;
-#endif /* ALTIVEC_USE_REFERENCE_C_CODE */
}
}
generated by cgit v1.2.3 (git 2.25.1) at 2025-04-25 22:59:09 +0000