virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-18 03:19:31 +02:00
Code Issues Releases Activity
f2da650258
FFmpeg/libavcodec/mips/simple_idct_msa.c

558 lines
27 KiB
C
Raw Normal View History

avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
/*
* 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 "libavutil/mips/generic_macros_msa.h"
#include "idctdsp_mips.h"
static void simple_idct_msa(int16_t *block)
{
int32_t const_val;
v8i16 weights = { 0, 22725, 21407, 19266, 16383, 12873, 8867, 4520 };
v8i16 in0, in1, in2, in3, in4, in5, in6, in7;
v8i16 w1, w3, w5, w7;
v8i16 const0, const1, const2, const3, const4, const5, const6, const7;
v4i32 temp0_r, temp1_r, temp2_r, temp3_r;
v4i32 temp0_l, temp1_l, temp2_l, temp3_l;
v4i32 a0_r, a1_r, a2_r, a3_r, a0_l, a1_l, a2_l, a3_l;
v4i32 b0_r, b1_r, b2_r, b3_r, b0_l, b1_l, b2_l, b3_l;
v4i32 w2, w4, w6;
v8i16 select_vec, temp;
v8i16 zero = { 0 };
v4i32 const_val0 = __msa_ldi_w(1);
v4i32 const_val1 = __msa_ldi_w(1);
LD_SH8(block, 8, in0, in1, in2, in3, in4, in5, in6, in7);
const_val0 <<= 10;
const_val = 16383 * ((1 << 19) / 16383);
const_val1 = __msa_insert_w(const_val0, 0, const_val);
const_val1 = __msa_splati_w(const_val1, 0);
TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
in0, in1, in2, in3, in4, in5, in6, in7);
select_vec = in1 | in2 | in3 | in4 | in5 | in6 | in7;
select_vec = __msa_clti_u_h((v8u16) select_vec, 1);
UNPCK_SH_SW(in0, a0_r, a0_l);
UNPCK_SH_SW(in2, temp3_r, temp3_l);
temp = in0 << 3;
w2 = (v4i32) __msa_splati_h(weights, 2);
w2 = (v4i32) __msa_ilvr_h(zero, (v8i16) w2);
w4 = (v4i32) __msa_splati_h(weights, 4);
w4 = (v4i32) __msa_ilvr_h(zero, (v8i16) w4);
w6 = (v4i32) __msa_splati_h(weights, 6);
w6 = (v4i32) __msa_ilvr_h(zero, (v8i16) w6);
MUL2(a0_r, w4, a0_l, w4, a0_r, a0_l);
ADD2(a0_r, const_val0, a0_l, const_val0, temp0_r, temp0_l);
MUL4(w2, temp3_r, w2, temp3_l, w6, temp3_r, w6, temp3_l,
temp1_r, temp1_l, temp2_r, temp2_l);
BUTTERFLY_8(temp0_r, temp0_l, temp0_r, temp0_l,
temp2_l, temp2_r, temp1_l, temp1_r,
a0_r, a0_l, a1_r, a1_l, a2_l, a2_r, a3_l, a3_r);
UNPCK_SH_SW(in4, temp0_r, temp0_l);
UNPCK_SH_SW(in6, temp3_r, temp3_l);
MUL2(temp0_r, w4, temp0_l, w4, temp0_r, temp0_l);
MUL4(w2, temp3_r, w2, temp3_l, w6, temp3_r, w6, temp3_l,
temp2_r, temp2_l, temp1_r, temp1_l);
ADD2(a0_r, temp0_r, a0_l, temp0_l, a0_r, a0_l);
SUB4(a1_r, temp0_r, a1_l, temp0_l, a2_r, temp0_r, a2_l, temp0_l,
a1_r, a1_l, a2_r, a2_l);
ADD4(a3_r, temp0_r, a3_l, temp0_l, a0_r, temp1_r, a0_l, temp1_l,
a3_r, a3_l, a0_r, a0_l);
SUB2(a1_r, temp2_r, a1_l, temp2_l, a1_r, a1_l);
ADD2(a2_r, temp2_r, a2_l, temp2_l, a2_r, a2_l);
SUB2(a3_r, temp1_r, a3_l, temp1_l, a3_r, a3_l);
ILVRL_H2_SW(in1, in3, b3_r, b3_l);
SPLATI_H4_SH(weights, 1, 3, 5, 7, w1, w3, w5, w7);
ILVRL_H2_SW(in5, in7, temp0_r, temp0_l);
ILVR_H4_SH(w1, w3, w3, -w7, w5, -w1, w7, -w5,
const0, const1, const2, const3);
ILVR_H2_SH(w5, w7, w7, w3, const4, const6);
const5 = __msa_ilvod_h(-w1, -w5);
const7 = __msa_ilvod_h(w3, -w1);
DOTP_SH4_SW(b3_r, b3_r, b3_r, b3_r, const0, const1, const2, const3,
b0_r, b1_r, b2_r, b3_r);
DPADD_SH4_SW(temp0_r, temp0_r, temp0_r, temp0_r,
const4, const5, const6, const7, b0_r, b1_r, b2_r, b3_r);
DOTP_SH4_SW(b3_l, b3_l, b3_l, b3_l, const0, const1, const2, const3,
b0_l, b1_l, b2_l, b3_l);
DPADD_SH4_SW(temp0_l, temp0_l, temp0_l, temp0_l,
const4, const5, const6, const7, b0_l, b1_l, b2_l, b3_l);
BUTTERFLY_16(a0_r, a0_l, a1_r, a1_l, a2_r, a2_l, a3_r, a3_l,
b3_l, b3_r, b2_l, b2_r, b1_l, b1_r, b0_l, b0_r,
temp0_r, temp0_l, temp1_r, temp1_l,
temp2_r, temp2_l, temp3_r, temp3_l,
a3_l, a3_r, a2_l, a2_r, a1_l, a1_r, a0_l, a0_r);
SRA_4V(temp0_r, temp0_l, temp1_r, temp1_l, 11);
SRA_4V(temp2_r, temp2_l, temp3_r, temp3_l, 11);
PCKEV_H4_SW(temp0_l, temp0_r, temp1_l, temp1_r,
temp2_l, temp2_r, temp3_l, temp3_r,
temp0_r, temp1_r, temp2_r, temp3_r);
in0 = (v8i16) __msa_bmnz_v((v16u8) temp0_r, (v16u8) temp,
(v16u8) select_vec);
in1 = (v8i16) __msa_bmnz_v((v16u8) temp1_r, (v16u8) temp,
(v16u8) select_vec);
in2 = (v8i16) __msa_bmnz_v((v16u8) temp2_r, (v16u8) temp,
(v16u8) select_vec);
in3 = (v8i16) __msa_bmnz_v((v16u8) temp3_r, (v16u8) temp,
(v16u8) select_vec);
SRA_4V(a3_r, a3_l, a2_r, a2_l, 11);
SRA_4V(a1_r, a1_l, a0_r, a0_l, 11);
PCKEV_H4_SW(a0_l, a0_r, a1_l, a1_r, a2_l, a2_r, a3_l, a3_r,
a0_r, a1_r, a2_r, a3_r);
in4 = (v8i16) __msa_bmnz_v((v16u8) a3_r, (v16u8) temp, (v16u8) select_vec);
in5 = (v8i16) __msa_bmnz_v((v16u8) a2_r, (v16u8) temp, (v16u8) select_vec);
in6 = (v8i16) __msa_bmnz_v((v16u8) a1_r, (v16u8) temp, (v16u8) select_vec);
in7 = (v8i16) __msa_bmnz_v((v16u8) a0_r, (v16u8) temp, (v16u8) select_vec);
TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
in0, in1, in2, in3, in4, in5, in6, in7);
UNPCK_SH_SW(in0, a0_r, a0_l);
UNPCK_SH_SW(in2, temp3_r, temp3_l);
w2 = (v4i32) __msa_splati_h(weights, 2);
w2 = (v4i32) __msa_ilvr_h(zero, (v8i16) w2);
w4 = (v4i32) __msa_splati_h(weights, 4);
w4 = (v4i32) __msa_ilvr_h(zero, (v8i16) w4);
w6 = (v4i32) __msa_splati_h(weights, 6);
w6 = (v4i32) __msa_ilvr_h(zero, (v8i16) w6);
MUL2(a0_r, w4, a0_l, w4, a0_r, a0_l);
ADD2(a0_r, const_val1, a0_l, const_val1, temp0_r, temp0_l);
MUL4(w2, temp3_r, w2, temp3_l, w6, temp3_r, w6, temp3_l,
temp1_r, temp1_l, temp2_r, temp2_l);
BUTTERFLY_8(temp0_r, temp0_l, temp0_r, temp0_l,
temp2_l, temp2_r, temp1_l, temp1_r,
a0_r, a0_l, a1_r, a1_l, a2_l, a2_r, a3_l, a3_r);
UNPCK_SH_SW(in4, temp0_r, temp0_l);
UNPCK_SH_SW(in6, temp3_r, temp3_l);
MUL2(temp0_r, w4, temp0_l, w4, temp0_r, temp0_l);
MUL4(w2, temp3_r, w2, temp3_l, w6, temp3_r, w6, temp3_l,
temp2_r, temp2_l, temp1_r, temp1_l);
ADD2(a0_r, temp0_r, a0_l, temp0_l, a0_r, a0_l);
SUB4(a1_r, temp0_r, a1_l, temp0_l, a2_r, temp0_r, a2_l, temp0_l,
a1_r, a1_l, a2_r, a2_l);
ADD4(a3_r, temp0_r, a3_l, temp0_l, a0_r, temp1_r, a0_l, temp1_l,
a3_r, a3_l, a0_r, a0_l);
SUB2(a1_r, temp2_r, a1_l, temp2_l, a1_r, a1_l);
ADD2(a2_r, temp2_r, a2_l, temp2_l, a2_r, a2_l);
SUB2(a3_r, temp1_r, a3_l, temp1_l, a3_r, a3_l);
ILVRL_H2_SW(in1, in3, b3_r, b3_l);
SPLATI_H4_SH(weights, 1, 3, 5, 7, w1, w3, w5, w7);
ILVR_H4_SH(w1, w3, w3, -w7, w5, -w1, w7, -w5,
const0, const1, const2, const3);
DOTP_SH4_SW(b3_r, b3_r, b3_r, b3_r, const0, const1, const2, const3,
b0_r, b1_r, b2_r, b3_r);
DOTP_SH4_SW(b3_l, b3_l, b3_l, b3_l, const0, const1, const2, const3,
b0_l, b1_l, b2_l, b3_l);
ILVRL_H2_SW(in5, in7, temp0_r, temp0_l);
ILVR_H2_SH(w5, w7, w7, w3, const4, const6);
const5 = __msa_ilvod_h(-w1, -w5);
const7 = __msa_ilvod_h(w3, -w1);
DPADD_SH4_SW(temp0_r, temp0_r, temp0_r, temp0_r,
const4, const5, const6, const7, b0_r, b1_r, b2_r, b3_r);
DPADD_SH4_SW(temp0_l, temp0_l, temp0_l, temp0_l,
const4, const5, const6, const7, b0_l, b1_l, b2_l, b3_l);
BUTTERFLY_16(a0_r, a0_l, a1_r, a1_l, a2_r, a2_l, a3_r, a3_l,
b3_l, b3_r, b2_l, b2_r, b1_l, b1_r, b0_l, b0_r,
temp0_r, temp0_l, temp1_r, temp1_l,
temp2_r, temp2_l, temp3_r, temp3_l,
a3_l, a3_r, a2_l, a2_r, a1_l, a1_r, a0_l, a0_r);
SRA_4V(temp0_r, temp0_l, temp1_r, temp1_l, 20);
SRA_4V(temp2_r, temp2_l, temp3_r, temp3_l, 20);
PCKEV_H4_SW(temp0_l, temp0_r, temp1_l, temp1_r, temp2_l, temp2_r,
temp3_l, temp3_r, temp0_r, temp1_r, temp2_r, temp3_r);
SRA_4V(a3_r, a3_l, a2_r, a2_l, 20);
SRA_4V(a1_r, a1_l, a0_r, a0_l, 20);
PCKEV_H4_SW(a0_l, a0_r, a1_l, a1_r, a2_l, a2_r, a3_l, a3_r,
a0_r, a1_r, a2_r, a3_r);
ST_SW8(temp0_r, temp1_r, temp2_r, temp3_r, a3_r, a2_r, a1_r, a0_r,
block, 8);
}
static void simple_idct_put_msa(uint8_t *dst, int32_t dst_stride,
int16_t *block)
{
int32_t const_val;
uint64_t tmp0, tmp1, tmp2, tmp3;
v8i16 weights = { 0, 22725, 21407, 19266, 16383, 12873, 8867, 4520 };
v8i16 in0, in1, in2, in3, in4, in5, in6, in7;
v8i16 w1, w3, w5, w7;
v8i16 const0, const1, const2, const3, const4, const5, const6, const7;
v4i32 temp0_r, temp1_r, temp2_r, temp3_r;
v4i32 temp0_l, temp1_l, temp2_l, temp3_l;
v4i32 a0_r, a1_r, a2_r, a3_r, a0_l, a1_l, a2_l, a3_l;
v4i32 b0_r, b1_r, b2_r, b3_r, b0_l, b1_l, b2_l, b3_l;
v4i32 w2, w4, w6;
v8i16 select_vec, temp;
v8i16 zero = { 0 };
v4i32 const_val0 = __msa_ldi_w(1);
v4i32 const_val1 = __msa_ldi_w(1);
LD_SH8(block, 8, in0, in1, in2, in3, in4, in5, in6, in7);
const_val0 <<= 10;
const_val = 16383 * ((1 << 19) / 16383);
const_val1 = __msa_insert_w(const_val0, 0, const_val);
const_val1 = __msa_splati_w(const_val1, 0);
TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
in0, in1, in2, in3, in4, in5, in6, in7);
select_vec = in1 | in2 | in3 | in4 | in5 | in6 | in7;
select_vec = __msa_clti_u_h((v8u16) select_vec, 1);
UNPCK_SH_SW(in0, a0_r, a0_l);
UNPCK_SH_SW(in2, temp3_r, temp3_l);
temp = in0 << 3;
w2 = (v4i32) __msa_splati_h(weights, 2);
w2 = (v4i32) __msa_ilvr_h(zero, (v8i16) w2);
w4 = (v4i32) __msa_splati_h(weights, 4);
w4 = (v4i32) __msa_ilvr_h(zero, (v8i16) w4);
w6 = (v4i32) __msa_splati_h(weights, 6);
w6 = (v4i32) __msa_ilvr_h(zero, (v8i16) w6);
MUL2(a0_r, w4, a0_l, w4, a0_r, a0_l);
ADD2(a0_r, const_val0, a0_l, const_val0, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp1_r, temp1_l);
MUL2(w6, temp3_r, w6, temp3_l, temp2_r, temp2_l);
BUTTERFLY_8(temp0_r, temp0_l, temp0_r, temp0_l,
temp2_l, temp2_r, temp1_l, temp1_r,
a0_r, a0_l, a1_r, a1_l, a2_l, a2_r, a3_l, a3_r);
UNPCK_SH_SW(in4, temp0_r, temp0_l);
UNPCK_SH_SW(in6, temp3_r, temp3_l);
MUL2(temp0_r, w4, temp0_l, w4, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp2_r, temp2_l);
MUL2(w6, temp3_r, w6, temp3_l, temp1_r, temp1_l);
ADD2(a0_r, temp0_r, a0_l, temp0_l, a0_r, a0_l);
SUB2(a1_r, temp0_r, a1_l, temp0_l, a1_r, a1_l);
SUB2(a2_r, temp0_r, a2_l, temp0_l, a2_r, a2_l);
ADD2(a3_r, temp0_r, a3_l, temp0_l, a3_r, a3_l);
ADD2(a0_r, temp1_r, a0_l, temp1_l, a0_r, a0_l);
SUB2(a1_r, temp2_r, a1_l, temp2_l, a1_r, a1_l);
ADD2(a2_r, temp2_r, a2_l, temp2_l, a2_r, a2_l);
SUB2(a3_r, temp1_r, a3_l, temp1_l, a3_r, a3_l);
ILVRL_H2_SW(in1, in3, b3_r, b3_l);
SPLATI_H4_SH(weights, 1, 3, 5, 7, w1, w3, w5, w7);
ILVRL_H2_SW(in5, in7, temp0_r, temp0_l);
ILVR_H4_SH(w1, w3, w3, -w7, w5, -w1, w7, -w5,
const0, const1, const2, const3);
ILVR_H2_SH(w5, w7, w7, w3, const4, const6);
const5 = __msa_ilvod_h(-w1, -w5);
const7 = __msa_ilvod_h(w3, -w1);
DOTP_SH4_SW(b3_r, b3_r, b3_r, b3_r, const0, const1, const2, const3,
b0_r, b1_r, b2_r, b3_r);
DPADD_SH4_SW(temp0_r, temp0_r, temp0_r, temp0_r,
const4, const5, const6, const7, b0_r, b1_r, b2_r, b3_r);
DOTP_SH4_SW(b3_l, b3_l, b3_l, b3_l, const0, const1, const2, const3,
b0_l, b1_l, b2_l, b3_l);
DPADD_SH4_SW(temp0_l, temp0_l, temp0_l, temp0_l,
const4, const5, const6, const7, b0_l, b1_l, b2_l, b3_l);
BUTTERFLY_16(a0_r, a0_l, a1_r, a1_l, a2_r, a2_l, a3_r, a3_l,
b3_l, b3_r, b2_l, b2_r, b1_l, b1_r, b0_l, b0_r,
temp0_r, temp0_l, temp1_r, temp1_l,
temp2_r, temp2_l, temp3_r, temp3_l,
a3_l, a3_r, a2_l, a2_r, a1_l, a1_r, a0_l, a0_r);
SRA_4V(temp0_r, temp0_l, temp1_r, temp1_l, 11);
SRA_4V(temp2_r, temp2_l, temp3_r, temp3_l, 11);
PCKEV_H4_SW(temp0_l, temp0_r, temp1_l, temp1_r,
temp2_l, temp2_r, temp3_l, temp3_r,
temp0_r, temp1_r, temp2_r, temp3_r);
in0 = (v8i16) __msa_bmnz_v((v16u8) temp0_r, (v16u8) temp,
(v16u8) select_vec);
in1 = (v8i16) __msa_bmnz_v((v16u8) temp1_r, (v16u8) temp,
(v16u8) select_vec);
in2 = (v8i16) __msa_bmnz_v((v16u8) temp2_r, (v16u8) temp,
(v16u8) select_vec);
in3 = (v8i16) __msa_bmnz_v((v16u8) temp3_r, (v16u8) temp,
(v16u8) select_vec);
SRA_4V(a3_r, a3_l, a2_r, a2_l, 11);
SRA_4V(a1_r, a1_l, a0_r, a0_l, 11);
PCKEV_H4_SW(a0_l, a0_r, a1_l, a1_r, a2_l, a2_r, a3_l, a3_r,
a0_r, a1_r, a2_r, a3_r);
in4 = (v8i16) __msa_bmnz_v((v16u8) a3_r, (v16u8) temp, (v16u8) select_vec);
in5 = (v8i16) __msa_bmnz_v((v16u8) a2_r, (v16u8) temp, (v16u8) select_vec);
in6 = (v8i16) __msa_bmnz_v((v16u8) a1_r, (v16u8) temp, (v16u8) select_vec);
in7 = (v8i16) __msa_bmnz_v((v16u8) a0_r, (v16u8) temp, (v16u8) select_vec);
TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
in0, in1, in2, in3, in4, in5, in6, in7);
UNPCK_SH_SW(in0, a0_r, a0_l);
UNPCK_SH_SW(in2, temp3_r, temp3_l);
w2 = (v4i32) __msa_splati_h(weights, 2);
w2 = (v4i32) __msa_ilvr_h(zero, (v8i16) w2);
w4 = (v4i32) __msa_splati_h(weights, 4);
w4 = (v4i32) __msa_ilvr_h(zero, (v8i16) w4);
w6 = (v4i32) __msa_splati_h(weights, 6);
w6 = (v4i32) __msa_ilvr_h(zero, (v8i16) w6);
MUL2(a0_r, w4, a0_l, w4, a0_r, a0_l);
ADD2(a0_r, const_val1, a0_l, const_val1, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp1_r, temp1_l);
MUL2(w6, temp3_r, w6, temp3_l, temp2_r, temp2_l);
BUTTERFLY_8(temp0_r, temp0_l, temp0_r, temp0_l,
temp2_l, temp2_r, temp1_l, temp1_r,
a0_r, a0_l, a1_r, a1_l, a2_l, a2_r, a3_l, a3_r);
UNPCK_SH_SW(in4, temp0_r, temp0_l);
UNPCK_SH_SW(in6, temp3_r, temp3_l);
MUL2(temp0_r, w4, temp0_l, w4, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp2_r, temp2_l);
MUL2(w6, temp3_r, w6, temp3_l, temp1_r, temp1_l);
ADD2(a0_r, temp0_r, a0_l, temp0_l, a0_r, a0_l);
SUB2(a1_r, temp0_r, a1_l, temp0_l, a1_r, a1_l);
SUB2(a2_r, temp0_r, a2_l, temp0_l, a2_r, a2_l);
ADD2(a3_r, temp0_r, a3_l, temp0_l, a3_r, a3_l);
ADD2(a0_r, temp1_r, a0_l, temp1_l, a0_r, a0_l);
SUB2(a1_r, temp2_r, a1_l, temp2_l, a1_r, a1_l);
ADD2(a2_r, temp2_r, a2_l, temp2_l, a2_r, a2_l);
SUB2(a3_r, temp1_r, a3_l, temp1_l, a3_r, a3_l);
ILVRL_H2_SW(in1, in3, b3_r, b3_l);
SPLATI_H4_SH(weights, 1, 3, 5, 7, w1, w3, w5, w7);
ILVR_H4_SH(w1, w3, w3, -w7, w5, -w1, w7, -w5,
const0, const1, const2, const3);
DOTP_SH4_SW(b3_r, b3_r, b3_r, b3_r, const0, const1, const2, const3,
b0_r, b1_r, b2_r, b3_r);
DOTP_SH4_SW(b3_l, b3_l, b3_l, b3_l, const0, const1, const2, const3,
b0_l, b1_l, b2_l, b3_l);
ILVRL_H2_SW(in5, in7, temp0_r, temp0_l);
ILVR_H2_SH(w5, w7, w7, w3, const4, const6);
const5 = __msa_ilvod_h(-w1, -w5);
const7 = __msa_ilvod_h(w3, -w1);
DPADD_SH4_SW(temp0_r, temp0_r, temp0_r, temp0_r,
const4, const5, const6, const7, b0_r, b1_r, b2_r, b3_r);
DPADD_SH4_SW(temp0_l, temp0_l, temp0_l, temp0_l,
const4, const5, const6, const7, b0_l, b1_l, b2_l, b3_l);
BUTTERFLY_16(a0_r, a0_l, a1_r, a1_l, a2_r, a2_l, a3_r, a3_l,
b3_l, b3_r, b2_l, b2_r, b1_l, b1_r, b0_l, b0_r,
temp0_r, temp0_l, temp1_r, temp1_l,
temp2_r, temp2_l, temp3_r, temp3_l,
a3_l, a3_r, a2_l, a2_r, a1_l, a1_r, a0_l, a0_r);
SRA_4V(temp0_r, temp0_l, temp1_r, temp1_l, 20);
SRA_4V(temp2_r, temp2_l, temp3_r, temp3_l, 20);
SRA_4V(a3_r, a3_l, a2_r, a2_l, 20);
SRA_4V(a1_r, a1_l, a0_r, a0_l, 20);
avutil/mips: refine msa macros CLIP_*. Changing details as following: 1. Remove the local variable 'out_m' in 'CLIP_SH' and store the result in source vector. 2. Refine the implementation of macro 'CLIP_SH_0_255' and 'CLIP_SW_0_255'. Performance of VP8 decoding has speed up about 1.1%(from 7.03x to 7.11x). Performance of H264 decoding has speed up about 0.5%(from 4.35x to 4.37x). Performance of Theora decoding has speed up about 0.7%(from 5.79x to 5.83x). 3. Remove redundant macro 'CLIP_SH/Wn_0_255_MAX_SATU' and use 'CLIP_SH/Wn_0_255' instead, because there are no difference in the effect of this two macros. Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2019-08-07 11:52:00 +02:00
PCKEV_H4_SH(temp0_l, temp0_r, temp1_l, temp1_r, temp2_l, temp2_r,
temp3_l, temp3_r, in0, in1, in2, in3);
PCKEV_H4_SH(a0_l, a0_r, a1_l, a1_r, a2_l, a2_r, a3_l, a3_r,
in4, in5, in6, in7);
CLIP_SH4_0_255(in0, in1, in2, in3);
PCKEV_B4_SH(in0, in0, in1, in1, in2, in2, in3, in3,
in0, in1, in2, in3);
tmp0 = __msa_copy_u_d((v2i64) in0, 1);
tmp1 = __msa_copy_u_d((v2i64) in1, 1);
tmp2 = __msa_copy_u_d((v2i64) in2, 1);
tmp3 = __msa_copy_u_d((v2i64) in3, 1);
avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
SD4(tmp0, tmp1, tmp2, tmp3, dst, dst_stride);
avutil/mips: refine msa macros CLIP_*. Changing details as following: 1. Remove the local variable 'out_m' in 'CLIP_SH' and store the result in source vector. 2. Refine the implementation of macro 'CLIP_SH_0_255' and 'CLIP_SW_0_255'. Performance of VP8 decoding has speed up about 1.1%(from 7.03x to 7.11x). Performance of H264 decoding has speed up about 0.5%(from 4.35x to 4.37x). Performance of Theora decoding has speed up about 0.7%(from 5.79x to 5.83x). 3. Remove redundant macro 'CLIP_SH/Wn_0_255_MAX_SATU' and use 'CLIP_SH/Wn_0_255' instead, because there are no difference in the effect of this two macros. Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2019-08-07 11:52:00 +02:00
CLIP_SH4_0_255(in4, in5, in6, in7);
PCKEV_B4_SH(in4, in4, in5, in5, in6, in6, in7, in7,
in4, in5, in6, in7);
tmp3 = __msa_copy_u_d((v2i64) in4, 1);
tmp2 = __msa_copy_u_d((v2i64) in5, 1);
tmp1 = __msa_copy_u_d((v2i64) in6, 1);
tmp0 = __msa_copy_u_d((v2i64) in7, 1);
SD4(tmp0, tmp1, tmp2, tmp3, dst + 4 * dst_stride, dst_stride);
avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
}
static void simple_idct_add_msa(uint8_t *dst, int32_t dst_stride,
int16_t *block)
{
int32_t const_val;
uint64_t tmp0, tmp1, tmp2, tmp3;
v8i16 weights = { 0, 22725, 21407, 19266, 16383, 12873, 8867, 4520 };
v8i16 in0, in1, in2, in3, in4, in5, in6, in7;
v8i16 w1, w3, w5, w7;
v8i16 const0, const1, const2, const3, const4, const5, const6, const7;
v4i32 temp0_r, temp1_r, temp2_r, temp3_r;
v4i32 temp4_r, temp5_r, temp6_r, temp7_r, temp8_r;
v4i32 temp0_l, temp1_l, temp2_l, temp3_l;
v4i32 temp4_l, temp5_l, temp6_l, temp7_l, temp8_l;
v4i32 a0_r, a1_r, a2_r, a3_r, a0_l, a1_l, a2_l, a3_l;
v4i32 b0_r, b1_r, b2_r, b3_r, b0_l, b1_l, b2_l, b3_l;
v4i32 w2, w4, w6;
v8i16 select_vec, temp;
v8i16 zero = { 0 };
v4i32 const_val0 = __msa_ldi_w(1);
v4i32 const_val1 = __msa_ldi_w(1);
const_val0 <<= 10;
const_val = 16383 * ((1 << 19) / 16383);
const_val1 = __msa_insert_w(const_val0, 0, const_val);
const_val1 = __msa_splati_w(const_val1, 0);
LD_SH8(block, 8, in0, in1, in2, in3, in4, in5, in6, in7);
TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
in0, in1, in2, in3, in4, in5, in6, in7);
select_vec = in1 | in2 | in3 | in4 | in5 | in6 | in7;
select_vec = __msa_clti_u_h((v8u16) select_vec, 1);
UNPCK_SH_SW(in0, a0_r, a0_l);
UNPCK_SH_SW(in2, temp3_r, temp3_l);
ILVRL_H2_SW(in1, in3, b3_r, b3_l);
UNPCK_SH_SW(in4, temp4_r, temp4_l);
UNPCK_SH_SW(in6, temp7_r, temp7_l);
ILVRL_H2_SW(in5, in7, temp8_r, temp8_l);
temp = in0 << 3;
SPLATI_H4_SH(weights, 1, 3, 5, 7, w1, w3, w5, w7);
ILVR_H4_SH(w1, w3, w3, -w7, w5, -w1, w7, -w5,
const0, const1, const2, const3);
ILVR_H2_SH(w5, w7, w7, w3, const4, const6);
const5 = __msa_ilvod_h(-w1, -w5);
const7 = __msa_ilvod_h(w3, -w1);
DOTP_SH4_SW(b3_r, b3_r, b3_r, b3_r, const0, const1, const2, const3,
b0_r, b1_r, b2_r, b3_r);
DPADD_SH4_SW(temp8_r, temp8_r, temp8_r, temp8_r,
const4, const5, const6, const7, b0_r, b1_r, b2_r, b3_r);
DOTP_SH4_SW(b3_l, b3_l, b3_l, b3_l, const0, const1, const2, const3,
b0_l, b1_l, b2_l, b3_l);
DPADD_SH4_SW(temp8_l, temp8_l, temp8_l, temp8_l,
const4, const5, const6, const7, b0_l, b1_l, b2_l, b3_l);
w2 = (v4i32) __msa_splati_h(weights, 2);
w2 = (v4i32) __msa_ilvr_h(zero, (v8i16) w2);
w4 = (v4i32) __msa_splati_h(weights, 4);
w4 = (v4i32) __msa_ilvr_h(zero, (v8i16) w4);
w6 = (v4i32) __msa_splati_h(weights, 6);
w6 = (v4i32) __msa_ilvr_h(zero, (v8i16) w6);
MUL2(a0_r, w4, a0_l, w4, a0_r, a0_l);
ADD2(a0_r, const_val0, a0_l, const_val0, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp1_r, temp1_l);
MUL2(w6, temp3_r, w6, temp3_l, temp2_r, temp2_l);
BUTTERFLY_8(temp0_r, temp0_l, temp0_r, temp0_l,
temp2_l, temp2_r, temp1_l, temp1_r,
a0_r, a0_l, a1_r, a1_l, a2_l, a2_r, a3_l, a3_r);
MUL2(temp4_r, w4, temp4_l, w4, temp4_r, temp4_l);
MUL2(temp7_r, w2, temp7_l, w2, temp6_r, temp6_l);
MUL2(temp7_r, w6, temp7_l, w6, temp5_r, temp5_l);
ADD2(a0_r, temp4_r, a0_l, temp4_l, a0_r, a0_l);
SUB2(a1_r, temp4_r, a1_l, temp4_l, a1_r, a1_l);
SUB2(a2_r, temp4_r, a2_l, temp4_l, a2_r, a2_l);
ADD2(a3_r, temp4_r, a3_l, temp4_l, a3_r, a3_l);
ADD2(a0_r, temp5_r, a0_l, temp5_l, a0_r, a0_l);
SUB2(a1_r, temp6_r, a1_l, temp6_l, a1_r, a1_l);
ADD2(a2_r, temp6_r, a2_l, temp6_l, a2_r, a2_l);
SUB2(a3_r, temp5_r, a3_l, temp5_l, a3_r, a3_l);
BUTTERFLY_16(a0_r, a0_l, a1_r, a1_l, a2_r, a2_l, a3_r, a3_l,
b3_l, b3_r, b2_l, b2_r, b1_l, b1_r, b0_l, b0_r,
temp0_r, temp0_l, temp1_r, temp1_l,
temp2_r, temp2_l, temp3_r, temp3_l,
a3_l, a3_r, a2_l, a2_r, a1_l, a1_r, a0_l, a0_r);
SRA_4V(temp0_r, temp0_l, temp1_r, temp1_l, 11);
SRA_4V(temp2_r, temp2_l, temp3_r, temp3_l, 11);
PCKEV_H4_SW(temp0_l, temp0_r, temp1_l, temp1_r,
temp2_l, temp2_r, temp3_l, temp3_r,
temp0_r, temp1_r, temp2_r, temp3_r);
in0 = (v8i16) __msa_bmnz_v((v16u8) temp0_r, (v16u8) temp,
(v16u8) select_vec);
in1 = (v8i16) __msa_bmnz_v((v16u8) temp1_r, (v16u8) temp,
(v16u8) select_vec);
in2 = (v8i16) __msa_bmnz_v((v16u8) temp2_r, (v16u8) temp,
(v16u8) select_vec);
in3 = (v8i16) __msa_bmnz_v((v16u8) temp3_r, (v16u8) temp,
(v16u8) select_vec);
SRA_4V(a3_r, a3_l, a2_r, a2_l, 11);
SRA_4V(a1_r, a1_l, a0_r, a0_l, 11);
PCKEV_H4_SW(a0_l, a0_r, a1_l, a1_r, a2_l, a2_r, a3_l, a3_r,
a0_r, a1_r, a2_r, a3_r);
in4 = (v8i16) __msa_bmnz_v((v16u8) a3_r, (v16u8) temp, (v16u8) select_vec);
in5 = (v8i16) __msa_bmnz_v((v16u8) a2_r, (v16u8) temp, (v16u8) select_vec);
in6 = (v8i16) __msa_bmnz_v((v16u8) a1_r, (v16u8) temp, (v16u8) select_vec);
in7 = (v8i16) __msa_bmnz_v((v16u8) a0_r, (v16u8) temp, (v16u8) select_vec);
TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7,
in0, in1, in2, in3, in4, in5, in6, in7);
UNPCK_SH_SW(in0, a0_r, a0_l);
UNPCK_SH_SW(in2, temp3_r, temp3_l);
MUL2(a0_r, w4, a0_l, w4, a0_r, a0_l);
ADD2(a0_r, const_val1, a0_l, const_val1, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp1_r, temp1_l);
MUL2(w6, temp3_r, w6, temp3_l, temp2_r, temp2_l);
BUTTERFLY_8(temp0_r, temp0_l, temp0_r, temp0_l,
temp2_l, temp2_r, temp1_l, temp1_r,
a0_r, a0_l, a1_r, a1_l, a2_l, a2_r, a3_l, a3_r);
UNPCK_SH_SW(in4, temp0_r, temp0_l);
UNPCK_SH_SW(in6, temp3_r, temp3_l);
MUL2(temp0_r, w4, temp0_l, w4, temp0_r, temp0_l);
MUL2(w2, temp3_r, w2, temp3_l, temp2_r, temp2_l);
MUL2(w6, temp3_r, w6, temp3_l, temp1_r, temp1_l);
ADD2(a0_r, temp0_r, a0_l, temp0_l, a0_r, a0_l);
SUB2(a1_r, temp0_r, a1_l, temp0_l, a1_r, a1_l);
SUB2(a2_r, temp0_r, a2_l, temp0_l, a2_r, a2_l);
ADD2(a3_r, temp0_r, a3_l, temp0_l, a3_r, a3_l);
ADD2(a0_r, temp1_r, a0_l, temp1_l, a0_r, a0_l);
SUB2(a1_r, temp2_r, a1_l, temp2_l, a1_r, a1_l);
ADD2(a2_r, temp2_r, a2_l, temp2_l, a2_r, a2_l);
SUB2(a3_r, temp1_r, a3_l, temp1_l, a3_r, a3_l);
ILVRL_H2_SW(in1, in3, b3_r, b3_l);
ILVRL_H2_SW(in5, in7, temp0_r, temp0_l);
DOTP_SH4_SW(b3_r, b3_r, b3_r, b3_r, const0, const1, const2, const3,
b0_r, b1_r, b2_r, b3_r);
DOTP_SH4_SW(b3_l, b3_l, b3_l, b3_l, const0, const1, const2, const3,
b0_l, b1_l, b2_l, b3_l);
DPADD_SH4_SW(temp0_r, temp0_r, temp0_r, temp0_r,
const4, const5, const6, const7, b0_r, b1_r, b2_r, b3_r);
DPADD_SH4_SW(temp0_l, temp0_l, temp0_l, temp0_l,
const4, const5, const6, const7, b0_l, b1_l, b2_l, b3_l);
BUTTERFLY_16(a0_r, a0_l, a1_r, a1_l, a2_r, a2_l, a3_r, a3_l,
b3_l, b3_r, b2_l, b2_r, b1_l, b1_r, b0_l, b0_r,
temp0_r, temp0_l, temp1_r, temp1_l,
temp2_r, temp2_l, temp3_r, temp3_l,
a3_l, a3_r, a2_l, a2_r, a1_l, a1_r, a0_l, a0_r);
SRA_4V(temp0_r, temp0_l, temp1_r, temp1_l, 20);
SRA_4V(temp2_r, temp2_l, temp3_r, temp3_l, 20);
LD_SH4(dst, dst_stride, in0, in1, in2, in3);
PCKEV_H4_SW(temp0_l, temp0_r, temp1_l, temp1_r, temp2_l, temp2_r,
temp3_l, temp3_r, temp0_r, temp1_r, temp2_r, temp3_r);
ILVR_B4_SW(zero, in0, zero, in1, zero, in2, zero, in3,
temp0_l, temp1_l, temp2_l, temp3_l);
avutil/mips: refine msa macros CLIP_*. Changing details as following: 1. Remove the local variable 'out_m' in 'CLIP_SH' and store the result in source vector. 2. Refine the implementation of macro 'CLIP_SH_0_255' and 'CLIP_SW_0_255'. Performance of VP8 decoding has speed up about 1.1%(from 7.03x to 7.11x). Performance of H264 decoding has speed up about 0.5%(from 4.35x to 4.37x). Performance of Theora decoding has speed up about 0.7%(from 5.79x to 5.83x). 3. Remove redundant macro 'CLIP_SH/Wn_0_255_MAX_SATU' and use 'CLIP_SH/Wn_0_255' instead, because there are no difference in the effect of this two macros. Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2019-08-07 11:52:00 +02:00
in0 = (v8i16) (temp0_r) + (v8i16) (temp0_l);
in1 = (v8i16) (temp1_r) + (v8i16) (temp1_l);
in2 = (v8i16) (temp2_r) + (v8i16) (temp2_l);
in3 = (v8i16) (temp3_r) + (v8i16) (temp3_l);
CLIP_SH4_0_255(in0, in1, in2, in3);
PCKEV_B4_SH(in0, in0, in1, in1, in2, in2, in3, in3,
in0, in1, in2, in3);
tmp0 = __msa_copy_u_d((v2i64) in0, 1);
tmp1 = __msa_copy_u_d((v2i64) in1, 1);
tmp2 = __msa_copy_u_d((v2i64) in2, 1);
tmp3 = __msa_copy_u_d((v2i64) in3, 1);
avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
SD4(tmp0, tmp1, tmp2, tmp3, dst, dst_stride);
SRA_4V(a3_r, a3_l, a2_r, a2_l, 20);
SRA_4V(a1_r, a1_l, a0_r, a0_l, 20);
LD_SH4(dst + 4 * dst_stride, dst_stride, in4, in5, in6, in7);
PCKEV_H4_SW(a0_l, a0_r, a1_l, a1_r, a2_l, a2_r, a3_l, a3_r,
a0_r, a1_r, a2_r, a3_r);
ILVR_B4_SW(zero, in4, zero, in5, zero, in6, zero, in7,
a3_l, a2_l, a1_l, a0_l);
avutil/mips: refine msa macros CLIP_*. Changing details as following: 1. Remove the local variable 'out_m' in 'CLIP_SH' and store the result in source vector. 2. Refine the implementation of macro 'CLIP_SH_0_255' and 'CLIP_SW_0_255'. Performance of VP8 decoding has speed up about 1.1%(from 7.03x to 7.11x). Performance of H264 decoding has speed up about 0.5%(from 4.35x to 4.37x). Performance of Theora decoding has speed up about 0.7%(from 5.79x to 5.83x). 3. Remove redundant macro 'CLIP_SH/Wn_0_255_MAX_SATU' and use 'CLIP_SH/Wn_0_255' instead, because there are no difference in the effect of this two macros. Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2019-08-07 11:52:00 +02:00
in4 = (v8i16) (a3_r) + (v8i16) (a3_l);
in5 = (v8i16) (a2_r) + (v8i16) (a2_l);
in6 = (v8i16) (a1_r) + (v8i16) (a1_l);
in7 = (v8i16) (a0_r) + (v8i16) (a0_l);
CLIP_SH4_0_255(in4, in5, in6, in7);
PCKEV_B4_SH(in4, in4, in5, in5, in6, in6, in7, in7,
in4, in5, in6, in7);
tmp0 = __msa_copy_u_d((v2i64) in4, 1);
tmp1 = __msa_copy_u_d((v2i64) in5, 1);
tmp2 = __msa_copy_u_d((v2i64) in6, 1);
tmp3 = __msa_copy_u_d((v2i64) in7, 1);
avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
SD4(tmp0, tmp1, tmp2, tmp3, dst + 4 * dst_stride, dst_stride);
}
void ff_simple_idct_msa(int16_t *block)
{
simple_idct_msa(block);
}
Merge commit '2ec9fa5ec60dcd10e1cb10d8b4e4437e634ea428' * commit '2ec9fa5ec60dcd10e1cb10d8b4e4437e634ea428': idct: Change type of array stride parameters to ptrdiff_t Merged-by: James Almer <jamrial@gmail.com>
2017-03-21 19:29:52 +02:00
void ff_simple_idct_put_msa(uint8_t *dst, ptrdiff_t dst_stride, int16_t *block)
avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
{
simple_idct_put_msa(dst, dst_stride, block);
}
Merge commit '2ec9fa5ec60dcd10e1cb10d8b4e4437e634ea428' * commit '2ec9fa5ec60dcd10e1cb10d8b4e4437e634ea428': idct: Change type of array stride parameters to ptrdiff_t Merged-by: James Almer <jamrial@gmail.com>
2017-03-21 19:29:52 +02:00
void ff_simple_idct_add_msa(uint8_t *dst, ptrdiff_t dst_stride, int16_t *block)
avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions This patch adds MSA (MIPS-SIMD-Arch) optimizations for idctdsp functions in new file idctdsp_msa.c and simple_idct_msa.c Signed-off-by: Shivraj Patil <shivraj.patil@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-06-29 17:27:15 +02:00
{
simple_idct_add_msa(dst, dst_stride, block);
}
Reference in New Issue Copy Permalink