virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-23 12:43:46 +02:00
Code Issues Releases Activity

avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for qpel functions

Browse Source 
This patch adds MSA (MIPS-SIMD-Arch) optimizations for qpel functions in new file qpeldsp_msa.c
Adds new generic macros (needed for this patch) in libavutil/mips/generic_macros_msa.h

Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Shivraj Patil 2015-06-14 23:26:23 +05:30 committed by Michael Niedermayer
parent d1050d9950
commit 98eb1ac901
7 changed files with 6972 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
libavcodec/mips/Makefile
View File

@ -25,6 +25,7 @@ OBJS-$(CONFIG_H264QPEL) += mips/h264qpel_init_mips.o
OBJS-$(CONFIG_H264CHROMA) += mips/h264chroma_init_mips.o
OBJS-$(CONFIG_H264PRED) += mips/h264pred_init_mips.o
OBJS-$(CONFIG_H263DSP) += mips/h263dsp_init_mips.o
OBJS-$(CONFIG_QPELDSP) += mips/qpeldsp_init_mips.o
MSA-OBJS-$(CONFIG_HEVC_DECODER) += mips/hevcdsp_msa.o \
mips/hevc_mc_uni_msa.o \
mips/hevc_mc_uniw_msa.o \
@ -39,5 +40,6 @@ MSA-OBJS-$(CONFIG_H264QPEL) += mips/h264qpel_msa.o
MSA-OBJS-$(CONFIG_H264CHROMA) += mips/h264chroma_msa.o
MSA-OBJS-$(CONFIG_H264PRED) += mips/h264pred_msa.o
MSA-OBJS-$(CONFIG_H263DSP) += mips/h263dsp_msa.o
MSA-OBJS-$(CONFIG_QPELDSP) += mips/qpeldsp_msa.o
LOONGSON3-OBJS-$(CONFIG_H264DSP) += mips/h264dsp_mmi.o
LOONGSON3-OBJS-$(CONFIG_H264CHROMA) += mips/h264chroma_mmi.o

167
libavcodec/mips/qpeldsp_init_mips.c Normal file
View File

@ -0,0 +1,167 @@
/*
* Copyright (c) 2015 Parag Salasakar (Parag.Salasakar@imgtec.com)
*
* 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
*/
#include "qpeldsp_mips.h"
#if HAVE_MSA
static av_cold void qpeldsp_init_msa(QpelDSPContext *c)
{
c->put_qpel_pixels_tab[0][0] = ff_copy_16x16_msa;
c->put_qpel_pixels_tab[0][1] = ff_horiz_mc_qpel_aver_src0_16width_msa;
c->put_qpel_pixels_tab[0][2] = ff_horiz_mc_qpel_16width_msa;
c->put_qpel_pixels_tab[0][3] = ff_horiz_mc_qpel_aver_src1_16width_msa;
c->put_qpel_pixels_tab[0][4] = ff_vert_mc_qpel_aver_src0_16x16_msa;
c->put_qpel_pixels_tab[0][5] = ff_hv_mc_qpel_aver_hv_src00_16x16_msa;
c->put_qpel_pixels_tab[0][6] = ff_hv_mc_qpel_aver_v_src0_16x16_msa;
c->put_qpel_pixels_tab[0][7] = ff_hv_mc_qpel_aver_hv_src10_16x16_msa;
c->put_qpel_pixels_tab[0][8] = ff_vert_mc_qpel_16x16_msa;
c->put_qpel_pixels_tab[0][9] = ff_hv_mc_qpel_aver_h_src0_16x16_msa;
c->put_qpel_pixels_tab[0][10] = ff_hv_mc_qpel_16x16_msa;
c->put_qpel_pixels_tab[0][11] = ff_hv_mc_qpel_aver_h_src1_16x16_msa;
c->put_qpel_pixels_tab[0][12] = ff_vert_mc_qpel_aver_src1_16x16_msa;
c->put_qpel_pixels_tab[0][13] = ff_hv_mc_qpel_aver_hv_src01_16x16_msa;
c->put_qpel_pixels_tab[0][14] = ff_hv_mc_qpel_aver_v_src1_16x16_msa;
c->put_qpel_pixels_tab[0][15] = ff_hv_mc_qpel_aver_hv_src11_16x16_msa;
c->put_qpel_pixels_tab[1][0] = ff_copy_8x8_msa;
c->put_qpel_pixels_tab[1][1] = ff_horiz_mc_qpel_aver_src0_8width_msa;
c->put_qpel_pixels_tab[1][2] = ff_horiz_mc_qpel_8width_msa;
c->put_qpel_pixels_tab[1][3] = ff_horiz_mc_qpel_aver_src1_8width_msa;
c->put_qpel_pixels_tab[1][4] = ff_vert_mc_qpel_aver_src0_8x8_msa;
c->put_qpel_pixels_tab[1][5] = ff_hv_mc_qpel_aver_hv_src00_8x8_msa;
c->put_qpel_pixels_tab[1][6] = ff_hv_mc_qpel_aver_v_src0_8x8_msa;
c->put_qpel_pixels_tab[1][7] = ff_hv_mc_qpel_aver_hv_src10_8x8_msa;
c->put_qpel_pixels_tab[1][8] = ff_vert_mc_qpel_8x8_msa;
c->put_qpel_pixels_tab[1][9] = ff_hv_mc_qpel_aver_h_src0_8x8_msa;
c->put_qpel_pixels_tab[1][10] = ff_hv_mc_qpel_8x8_msa;
c->put_qpel_pixels_tab[1][11] = ff_hv_mc_qpel_aver_h_src1_8x8_msa;
c->put_qpel_pixels_tab[1][12] = ff_vert_mc_qpel_aver_src1_8x8_msa;
c->put_qpel_pixels_tab[1][13] = ff_hv_mc_qpel_aver_hv_src01_8x8_msa;
c->put_qpel_pixels_tab[1][14] = ff_hv_mc_qpel_aver_v_src1_8x8_msa;
c->put_qpel_pixels_tab[1][15] = ff_hv_mc_qpel_aver_hv_src11_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[0][0] = ff_copy_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][1] =
ff_horiz_mc_qpel_no_rnd_aver_src0_16width_msa;
c->put_no_rnd_qpel_pixels_tab[0][2] = ff_horiz_mc_qpel_no_rnd_16width_msa;
c->put_no_rnd_qpel_pixels_tab[0][3] =
ff_horiz_mc_qpel_no_rnd_aver_src1_16width_msa;
c->put_no_rnd_qpel_pixels_tab[0][4] =
ff_vert_mc_qpel_no_rnd_aver_src0_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][5] =
ff_hv_mc_qpel_no_rnd_aver_hv_src00_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][6] =
ff_hv_mc_qpel_no_rnd_aver_v_src0_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][7] =
ff_hv_mc_qpel_no_rnd_aver_hv_src10_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][8] = ff_vert_mc_qpel_no_rnd_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][9] =
ff_hv_mc_qpel_no_rnd_aver_h_src0_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][10] = ff_hv_mc_qpel_no_rnd_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][11] =
ff_hv_mc_qpel_no_rnd_aver_h_src1_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][12] =
ff_vert_mc_qpel_no_rnd_aver_src1_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][13] =
ff_hv_mc_qpel_no_rnd_aver_hv_src01_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][14] =
ff_hv_mc_qpel_no_rnd_aver_v_src1_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[0][15] =
ff_hv_mc_qpel_no_rnd_aver_hv_src11_16x16_msa;
c->put_no_rnd_qpel_pixels_tab[1][0] = ff_copy_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][1] =
ff_horiz_mc_qpel_no_rnd_aver_src0_8width_msa;
c->put_no_rnd_qpel_pixels_tab[1][2] = ff_horiz_mc_qpel_no_rnd_8width_msa;
c->put_no_rnd_qpel_pixels_tab[1][3] =
ff_horiz_mc_qpel_no_rnd_aver_src1_8width_msa;
c->put_no_rnd_qpel_pixels_tab[1][4] =
ff_vert_mc_qpel_no_rnd_aver_src0_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][5] =
ff_hv_mc_qpel_no_rnd_aver_hv_src00_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][6] =
ff_hv_mc_qpel_no_rnd_aver_v_src0_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][7] =
ff_hv_mc_qpel_no_rnd_aver_hv_src10_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][8] = ff_vert_mc_qpel_no_rnd_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][9] =
ff_hv_mc_qpel_no_rnd_aver_h_src0_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][10] = ff_hv_mc_qpel_no_rnd_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][11] =
ff_hv_mc_qpel_no_rnd_aver_h_src1_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][12] =
ff_vert_mc_qpel_no_rnd_aver_src1_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][13] =
ff_hv_mc_qpel_no_rnd_aver_hv_src01_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][14] =
ff_hv_mc_qpel_no_rnd_aver_v_src1_8x8_msa;
c->put_no_rnd_qpel_pixels_tab[1][15] =
ff_hv_mc_qpel_no_rnd_aver_hv_src11_8x8_msa;
c->avg_qpel_pixels_tab[0][0] = ff_avg_width16_msa;
c->avg_qpel_pixels_tab[0][1] =
ff_horiz_mc_qpel_avg_dst_aver_src0_16width_msa;
c->avg_qpel_pixels_tab[0][2] = ff_horiz_mc_qpel_avg_dst_16width_msa;
c->avg_qpel_pixels_tab[0][3] =
ff_horiz_mc_qpel_avg_dst_aver_src1_16width_msa;
c->avg_qpel_pixels_tab[0][4] = ff_vert_mc_qpel_avg_dst_aver_src0_16x16_msa;
c->avg_qpel_pixels_tab[0][5] =
ff_hv_mc_qpel_avg_dst_aver_hv_src00_16x16_msa;
c->avg_qpel_pixels_tab[0][6] = ff_hv_mc_qpel_avg_dst_aver_v_src0_16x16_msa;
c->avg_qpel_pixels_tab[0][7] =
ff_hv_mc_qpel_avg_dst_aver_hv_src10_16x16_msa;
c->avg_qpel_pixels_tab[0][8] = ff_vert_mc_qpel_avg_dst_16x16_msa;
c->avg_qpel_pixels_tab[0][9] = ff_hv_mc_qpel_avg_dst_aver_h_src0_16x16_msa;
c->avg_qpel_pixels_tab[0][10] = ff_hv_mc_qpel_avg_dst_16x16_msa;
c->avg_qpel_pixels_tab[0][11] = ff_hv_mc_qpel_avg_dst_aver_h_src1_16x16_msa;
c->avg_qpel_pixels_tab[0][12] = ff_vert_mc_qpel_avg_dst_aver_src1_16x16_msa;
c->avg_qpel_pixels_tab[0][13] =
ff_hv_mc_qpel_avg_dst_aver_hv_src01_16x16_msa;
c->avg_qpel_pixels_tab[0][14] = ff_hv_mc_qpel_avg_dst_aver_v_src1_16x16_msa;
c->avg_qpel_pixels_tab[0][15] =
ff_hv_mc_qpel_avg_dst_aver_hv_src11_16x16_msa;
c->avg_qpel_pixels_tab[1][0] = ff_avg_width8_msa;
c->avg_qpel_pixels_tab[1][1] =
ff_horiz_mc_qpel_avg_dst_aver_src0_8width_msa;
c->avg_qpel_pixels_tab[1][2] = ff_horiz_mc_qpel_avg_dst_8width_msa;
c->avg_qpel_pixels_tab[1][3] =
ff_horiz_mc_qpel_avg_dst_aver_src1_8width_msa;
c->avg_qpel_pixels_tab[1][4] = ff_vert_mc_qpel_avg_dst_aver_src0_8x8_msa;
c->avg_qpel_pixels_tab[1][5] = ff_hv_mc_qpel_avg_dst_aver_hv_src00_8x8_msa;
c->avg_qpel_pixels_tab[1][6] = ff_hv_mc_qpel_avg_dst_aver_v_src0_8x8_msa;
c->avg_qpel_pixels_tab[1][7] = ff_hv_mc_qpel_avg_dst_aver_hv_src10_8x8_msa;
c->avg_qpel_pixels_tab[1][8] = ff_vert_mc_qpel_avg_dst_8x8_msa;
c->avg_qpel_pixels_tab[1][9] = ff_hv_mc_qpel_avg_dst_aver_h_src0_8x8_msa;
c->avg_qpel_pixels_tab[1][10] = ff_hv_mc_qpel_avg_dst_8x8_msa;
c->avg_qpel_pixels_tab[1][11] = ff_hv_mc_qpel_avg_dst_aver_h_src1_8x8_msa;
c->avg_qpel_pixels_tab[1][12] = ff_vert_mc_qpel_avg_dst_aver_src1_8x8_msa;
c->avg_qpel_pixels_tab[1][13] = ff_hv_mc_qpel_avg_dst_aver_hv_src01_8x8_msa;
c->avg_qpel_pixels_tab[1][14] = ff_hv_mc_qpel_avg_dst_aver_v_src1_8x8_msa;
c->avg_qpel_pixels_tab[1][15] = ff_hv_mc_qpel_avg_dst_aver_hv_src11_8x8_msa;
}
#endif // #if HAVE_MSA
void ff_qpeldsp_init_mips(QpelDSPContext *c)
{
#if HAVE_MSA
qpeldsp_init_msa(c);
#endif // #if HAVE_MSA
}

261
libavcodec/mips/qpeldsp_mips.h Normal file
View File

@ -0,0 +1,261 @@
/*
* Copyright (c) 2015 Parag Salasakar (Parag.Salasakar@imgtec.com)
*
* 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 AVCODEC_MIPS_QPELDSP_MIPS_H
#define AVCODEC_MIPS_QPELDSP_MIPS_H
#include "../mpegvideo.h"
void ff_copy_8x8_msa(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
void ff_copy_16x16_msa(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
void ff_avg_width8_msa(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
void ff_avg_width16_msa(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
void ff_horiz_mc_qpel_aver_src0_8width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_aver_src0_16width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_8width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_16width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_aver_src1_8width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_aver_src1_16width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_no_rnd_aver_src0_8width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_no_rnd_aver_src0_16width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_no_rnd_8width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_no_rnd_16width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_no_rnd_aver_src1_8width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_no_rnd_aver_src1_16width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_avg_dst_aver_src0_8width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_avg_dst_aver_src0_16width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_avg_dst_8width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_avg_dst_16width_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_avg_dst_aver_src1_8width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_horiz_mc_qpel_avg_dst_aver_src1_16width_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_aver_src0_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_aver_src0_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_aver_src1_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_aver_src1_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_no_rnd_aver_src0_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_no_rnd_aver_src0_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_no_rnd_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_no_rnd_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_no_rnd_aver_src1_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_no_rnd_aver_src1_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_avg_dst_aver_src0_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_avg_dst_aver_src0_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_avg_dst_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_avg_dst_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_avg_dst_aver_src1_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_vert_mc_qpel_avg_dst_aver_src1_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src00_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src00_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_v_src0_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_v_src0_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src10_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src10_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_h_src0_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_h_src0_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_8x8_msa(uint8_t *dst, const uint8_t *src, ptrdiff_t stride);
void ff_hv_mc_qpel_aver_h_src1_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_h_src1_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src01_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src01_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_v_src1_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_v_src1_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src11_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_aver_hv_src11_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src00_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src00_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_v_src0_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_v_src0_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src10_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src10_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_h_src0_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_h_src0_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_h_src1_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_h_src1_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src01_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src01_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_v_src1_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_v_src1_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src11_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_avg_dst_aver_hv_src11_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src00_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src00_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_v_src0_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_v_src0_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src10_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src10_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_h_src0_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_h_src0_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_16x16_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_8x8_msa(uint8_t *dst, const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_h_src1_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_h_src1_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src01_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src01_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_v_src1_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_v_src1_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src11_16x16_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
void ff_hv_mc_qpel_no_rnd_aver_hv_src11_8x8_msa(uint8_t *dst,
const uint8_t *src,
ptrdiff_t stride);
#endif // #ifndef AVCODEC_MIPS_QPELDSP_MIPS_H

6518
libavcodec/mips/qpeldsp_msa.c Normal file
View File

File diff suppressed because it is too large Load Diff

2
libavcodec/qpeldsp.c
View File

@ -811,4 +811,6 @@ av_cold void ff_qpeldsp_init(QpelDSPContext *c)
if (ARCH_X86)
ff_qpeldsp_init_x86(c);
if (ARCH_MIPS)
ff_qpeldsp_init_mips(c);
}

1
libavcodec/qpeldsp.h
View File

@ -78,5 +78,6 @@ typedef struct QpelDSPContext {
void ff_qpeldsp_init(QpelDSPContext *c);
void ff_qpeldsp_init_x86(QpelDSPContext *c);
void ff_qpeldsp_init_mips(QpelDSPContext *c);
#endif /* AVCODEC_QPELDSP_H */

21
libavutil/mips/generic_macros_msa.h
View File

@ -1027,6 +1027,27 @@
}
#define DPADD_SB4_SH(...) DPADD_SB4(v8i16, __VA_ARGS__)
/* Description : Dot product & addition of byte vector elements
Arguments : Inputs - mult0, mult1
cnst0, cnst1
Outputs - out0, out1
Return Type - unsigned halfword
Details : Unsigned byte elements from mult0 are multiplied with
unsigned byte elements from cnst0 producing a result
twice the size of input i.e. unsigned halfword.
Then this multiplication results of adjacent odd-even elements
are added to the out vector
(2 unsigned halfword results)
*/
#define DPADD_UB2(RTYPE, mult0, mult1, cnst0, cnst1, out0, out1) \
{ \
out0 = (RTYPE) __msa_dpadd_u_h((v8u16) out0, \
(v16u8) mult0, (v16u8) cnst0); \
out1 = (RTYPE) __msa_dpadd_u_h((v8u16) out1, \
(v16u8) mult1, (v16u8) cnst1); \
}
#define DPADD_UB2_UH(...) DPADD_UB2(v8u16, __VA_ARGS__)
/* Description : Dot product & addition of halfword vector elements
Arguments : Inputs - mult0, mult1
cnst0, cnst1