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
b69486ea18
FFmpeg/libavcodec/mips/sbrdsp_mips.c

913 lines
46 KiB
C
Raw Normal View History

mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
/*
* Copyright (c) 2012
* MIPS Technologies, Inc., California.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the MIPS Technologies, Inc., nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE MIPS TECHNOLOGIES, INC. ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE MIPS TECHNOLOGIES, INC. BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* Authors: Darko Laus (darko@mips.com)
* Djordje Pesut (djordje@mips.com)
* Mirjana Vulin (mvulin@mips.com)
*
* AAC Spectral Band Replication decoding functions optimized for MIPS
*
* 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
*/
/**
* @file
* Reference: libavcodec/sbrdsp.c
*/
#include "config.h"
#include "libavcodec/sbrdsp.h"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
#include "libavutil/mips/asmdefs.h"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
#if HAVE_INLINE_ASM
avcodec/aacdec: fix compilation under soft float MIPS Place HAVE_MIPSFPU further up so that functions that use floating point ASM are defined away. Otherwise compilation failures result when soft float in enabled on the toolchain. Signed-off-by: Rosen Penev <rosenp@gmail.com>
2020-04-06 05:37:10 +02:00
#if HAVE_MIPSFPU
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
static void sbr_qmf_pre_shuffle_mips(float *z)
{
int Temp1, Temp2, Temp3, Temp4, Temp5, Temp6;
float *z1 = &z[66];
float *z2 = &z[59];
float *z3 = &z[2];
float *z4 = z1 + 60;
/* loop unrolled 5 times */
__asm__ volatile (
"lui %[Temp6], 0x8000 \n\t"
"1: \n\t"
"lw %[Temp1], 0(%[z2]) \n\t"
"lw %[Temp2], 4(%[z2]) \n\t"
"lw %[Temp3], 8(%[z2]) \n\t"
"lw %[Temp4], 12(%[z2]) \n\t"
"lw %[Temp5], 16(%[z2]) \n\t"
"xor %[Temp1], %[Temp1], %[Temp6] \n\t"
"xor %[Temp2], %[Temp2], %[Temp6] \n\t"
"xor %[Temp3], %[Temp3], %[Temp6] \n\t"
"xor %[Temp4], %[Temp4], %[Temp6] \n\t"
"xor %[Temp5], %[Temp5], %[Temp6] \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[z2], %[z2], -20 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"sw %[Temp1], 32(%[z1]) \n\t"
"sw %[Temp2], 24(%[z1]) \n\t"
"sw %[Temp3], 16(%[z1]) \n\t"
"sw %[Temp4], 8(%[z1]) \n\t"
"sw %[Temp5], 0(%[z1]) \n\t"
"lw %[Temp1], 0(%[z3]) \n\t"
"lw %[Temp2], 4(%[z3]) \n\t"
"lw %[Temp3], 8(%[z3]) \n\t"
"lw %[Temp4], 12(%[z3]) \n\t"
"lw %[Temp5], 16(%[z3]) \n\t"
"sw %[Temp1], 4(%[z1]) \n\t"
"sw %[Temp2], 12(%[z1]) \n\t"
"sw %[Temp3], 20(%[z1]) \n\t"
"sw %[Temp4], 28(%[z1]) \n\t"
"sw %[Temp5], 36(%[z1]) \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[z3], %[z3], 20 \n\t"
PTR_ADDIU "%[z1], %[z1], 40 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"bne %[z1], %[z4], 1b \n\t"
"lw %[Temp1], 132(%[z]) \n\t"
"lw %[Temp2], 128(%[z]) \n\t"
"lw %[Temp3], 0(%[z]) \n\t"
"lw %[Temp4], 4(%[z]) \n\t"
"xor %[Temp1], %[Temp1], %[Temp6] \n\t"
"sw %[Temp1], 504(%[z]) \n\t"
"sw %[Temp2], 508(%[z]) \n\t"
"sw %[Temp3], 256(%[z]) \n\t"
"sw %[Temp4], 260(%[z]) \n\t"
: [Temp1]"=&r"(Temp1), [Temp2]"=&r"(Temp2),
[Temp3]"=&r"(Temp3), [Temp4]"=&r"(Temp4),
[Temp5]"=&r"(Temp5), [Temp6]"=&r"(Temp6),
[z1]"+r"(z1), [z2]"+r"(z2), [z3]"+r"(z3)
: [z4]"r"(z4), [z]"r"(z)
: "memory"
);
}
static void sbr_qmf_post_shuffle_mips(float W[32][2], const float *z)
{
int Temp1, Temp2, Temp3, Temp4, Temp5;
float *W_ptr = (float *)W;
float *z1 = (float *)z;
float *z2 = (float *)&z[60];
float *z_end = z1 + 32;
/* loop unrolled 4 times */
__asm__ volatile (
"lui %[Temp5], 0x8000 \n\t"
"1: \n\t"
"lw %[Temp1], 0(%[z2]) \n\t"
"lw %[Temp2], 4(%[z2]) \n\t"
"lw %[Temp3], 8(%[z2]) \n\t"
"lw %[Temp4], 12(%[z2]) \n\t"
"xor %[Temp1], %[Temp1], %[Temp5] \n\t"
"xor %[Temp2], %[Temp2], %[Temp5] \n\t"
"xor %[Temp3], %[Temp3], %[Temp5] \n\t"
"xor %[Temp4], %[Temp4], %[Temp5] \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[z2], %[z2], -16 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"sw %[Temp1], 24(%[W_ptr]) \n\t"
"sw %[Temp2], 16(%[W_ptr]) \n\t"
"sw %[Temp3], 8(%[W_ptr]) \n\t"
"sw %[Temp4], 0(%[W_ptr]) \n\t"
"lw %[Temp1], 0(%[z1]) \n\t"
"lw %[Temp2], 4(%[z1]) \n\t"
"lw %[Temp3], 8(%[z1]) \n\t"
"lw %[Temp4], 12(%[z1]) \n\t"
"sw %[Temp1], 4(%[W_ptr]) \n\t"
"sw %[Temp2], 12(%[W_ptr]) \n\t"
"sw %[Temp3], 20(%[W_ptr]) \n\t"
"sw %[Temp4], 28(%[W_ptr]) \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[z1], %[z1], 16 \n\t"
PTR_ADDIU "%[W_ptr],%[W_ptr], 32 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"bne %[z1], %[z_end], 1b \n\t"
: [Temp1]"=&r"(Temp1), [Temp2]"=&r"(Temp2),
[Temp3]"=&r"(Temp3), [Temp4]"=&r"(Temp4),
[Temp5]"=&r"(Temp5), [z1]"+r"(z1),
[z2]"+r"(z2), [W_ptr]"+r"(W_ptr)
: [z_end]"r"(z_end)
: "memory"
);
}
mips: add support for R6 Understanding the mips32r6 and mips64r6 ISAs in the configure script is not enough. In order to have full support for MIPS R6 in FFmpeg we need to be able to build it, and for that we need to make sure we don't use incompatible assembler code which makes the build fail. Ifdefing the offending code is sufficient to fix the problem. Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2016-02-24 18:38:22 +02:00
#if !HAVE_MIPS32R6 && !HAVE_MIPS64R6
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
static void sbr_sum64x5_mips(float *z)
{
int k;
float *z1;
float f1, f2, f3, f4, f5, f6, f7, f8;
for (k = 0; k < 64; k += 8) {
z1 = &z[k];
/* loop unrolled 8 times */
__asm__ volatile (
"lwc1 $f0, 0(%[z1]) \n\t"
"lwc1 $f1, 256(%[z1]) \n\t"
"lwc1 $f2, 4(%[z1]) \n\t"
"lwc1 $f3, 260(%[z1]) \n\t"
"lwc1 $f4, 8(%[z1]) \n\t"
"add.s %[f1], $f0, $f1 \n\t"
"lwc1 $f5, 264(%[z1]) \n\t"
"add.s %[f2], $f2, $f3 \n\t"
"lwc1 $f6, 12(%[z1]) \n\t"
"lwc1 $f7, 268(%[z1]) \n\t"
"add.s %[f3], $f4, $f5 \n\t"
"lwc1 $f8, 16(%[z1]) \n\t"
"lwc1 $f9, 272(%[z1]) \n\t"
"add.s %[f4], $f6, $f7 \n\t"
"lwc1 $f10, 20(%[z1]) \n\t"
"lwc1 $f11, 276(%[z1]) \n\t"
"add.s %[f5], $f8, $f9 \n\t"
"lwc1 $f12, 24(%[z1]) \n\t"
"lwc1 $f13, 280(%[z1]) \n\t"
"add.s %[f6], $f10, $f11 \n\t"
"lwc1 $f14, 28(%[z1]) \n\t"
"lwc1 $f15, 284(%[z1]) \n\t"
"add.s %[f7], $f12, $f13 \n\t"
"lwc1 $f0, 512(%[z1]) \n\t"
"lwc1 $f1, 516(%[z1]) \n\t"
"add.s %[f8], $f14, $f15 \n\t"
"lwc1 $f2, 520(%[z1]) \n\t"
"add.s %[f1], %[f1], $f0 \n\t"
"add.s %[f2], %[f2], $f1 \n\t"
"lwc1 $f3, 524(%[z1]) \n\t"
"add.s %[f3], %[f3], $f2 \n\t"
"lwc1 $f4, 528(%[z1]) \n\t"
"lwc1 $f5, 532(%[z1]) \n\t"
"add.s %[f4], %[f4], $f3 \n\t"
"lwc1 $f6, 536(%[z1]) \n\t"
"add.s %[f5], %[f5], $f4 \n\t"
"add.s %[f6], %[f6], $f5 \n\t"
"lwc1 $f7, 540(%[z1]) \n\t"
"add.s %[f7], %[f7], $f6 \n\t"
"lwc1 $f0, 768(%[z1]) \n\t"
"lwc1 $f1, 772(%[z1]) \n\t"
"add.s %[f8], %[f8], $f7 \n\t"
"lwc1 $f2, 776(%[z1]) \n\t"
"add.s %[f1], %[f1], $f0 \n\t"
"add.s %[f2], %[f2], $f1 \n\t"
"lwc1 $f3, 780(%[z1]) \n\t"
"add.s %[f3], %[f3], $f2 \n\t"
"lwc1 $f4, 784(%[z1]) \n\t"
"lwc1 $f5, 788(%[z1]) \n\t"
"add.s %[f4], %[f4], $f3 \n\t"
"lwc1 $f6, 792(%[z1]) \n\t"
"add.s %[f5], %[f5], $f4 \n\t"
"add.s %[f6], %[f6], $f5 \n\t"
"lwc1 $f7, 796(%[z1]) \n\t"
"add.s %[f7], %[f7], $f6 \n\t"
"lwc1 $f0, 1024(%[z1]) \n\t"
"lwc1 $f1, 1028(%[z1]) \n\t"
"add.s %[f8], %[f8], $f7 \n\t"
"lwc1 $f2, 1032(%[z1]) \n\t"
"add.s %[f1], %[f1], $f0 \n\t"
"add.s %[f2], %[f2], $f1 \n\t"
"lwc1 $f3, 1036(%[z1]) \n\t"
"add.s %[f3], %[f3], $f2 \n\t"
"lwc1 $f4, 1040(%[z1]) \n\t"
"lwc1 $f5, 1044(%[z1]) \n\t"
"add.s %[f4], %[f4], $f3 \n\t"
"lwc1 $f6, 1048(%[z1]) \n\t"
"add.s %[f5], %[f5], $f4 \n\t"
"add.s %[f6], %[f6], $f5 \n\t"
"lwc1 $f7, 1052(%[z1]) \n\t"
"add.s %[f7], %[f7], $f6 \n\t"
"swc1 %[f1], 0(%[z1]) \n\t"
"swc1 %[f2], 4(%[z1]) \n\t"
"add.s %[f8], %[f8], $f7 \n\t"
"swc1 %[f3], 8(%[z1]) \n\t"
"swc1 %[f4], 12(%[z1]) \n\t"
"swc1 %[f5], 16(%[z1]) \n\t"
"swc1 %[f6], 20(%[z1]) \n\t"
"swc1 %[f7], 24(%[z1]) \n\t"
"swc1 %[f8], 28(%[z1]) \n\t"
: [f1]"=&f"(f1), [f2]"=&f"(f2), [f3]"=&f"(f3),
[f4]"=&f"(f4), [f5]"=&f"(f5), [f6]"=&f"(f6),
[f7]"=&f"(f7), [f8]"=&f"(f8)
: [z1]"r"(z1)
: "$f0", "$f1", "$f2", "$f3", "$f4", "$f5",
"$f6", "$f7", "$f8", "$f9", "$f10", "$f11",
"$f12", "$f13", "$f14", "$f15",
"memory"
);
}
}
static float sbr_sum_square_mips(float (*x)[2], int n)
{
float sum0 = 0.0f, sum1 = 0.0f;
float *p_x;
float temp0, temp1, temp2, temp3;
float *loop_end;
p_x = &x[0][0];
loop_end = p_x + (n >> 1)*4 - 4;
__asm__ volatile (
".set push \n\t"
".set noreorder \n\t"
"lwc1 %[temp0], 0(%[p_x]) \n\t"
"lwc1 %[temp1], 4(%[p_x]) \n\t"
"lwc1 %[temp2], 8(%[p_x]) \n\t"
"lwc1 %[temp3], 12(%[p_x]) \n\t"
"1: \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[p_x], %[p_x], 16 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"madd.s %[sum0], %[sum0], %[temp0], %[temp0] \n\t"
"lwc1 %[temp0], 0(%[p_x]) \n\t"
"madd.s %[sum1], %[sum1], %[temp1], %[temp1] \n\t"
"lwc1 %[temp1], 4(%[p_x]) \n\t"
"madd.s %[sum0], %[sum0], %[temp2], %[temp2] \n\t"
"lwc1 %[temp2], 8(%[p_x]) \n\t"
"madd.s %[sum1], %[sum1], %[temp3], %[temp3] \n\t"
"bne %[p_x], %[loop_end], 1b \n\t"
" lwc1 %[temp3], 12(%[p_x]) \n\t"
"madd.s %[sum0], %[sum0], %[temp0], %[temp0] \n\t"
"madd.s %[sum1], %[sum1], %[temp1], %[temp1] \n\t"
"madd.s %[sum0], %[sum0], %[temp2], %[temp2] \n\t"
"madd.s %[sum1], %[sum1], %[temp3], %[temp3] \n\t"
".set pop \n\t"
: [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&f"(temp3), [sum0]"+f"(sum0), [sum1]"+f"(sum1),
[p_x]"+r"(p_x)
: [loop_end]"r"(loop_end)
: "memory"
);
return sum0 + sum1;
}
static void sbr_qmf_deint_bfly_mips(float *v, const float *src0, const float *src1)
{
int i;
float temp0, temp1, temp2, temp3, temp4, temp5;
float temp6, temp7, temp8, temp9, temp10, temp11;
float *v0 = v;
float *v1 = &v[127];
float *psrc0 = (float*)src0;
float *psrc1 = (float*)&src1[63];
for (i = 0; i < 4; i++) {
/* loop unrolled 16 times */
__asm__ volatile(
"lwc1 %[temp0], 0(%[src0]) \n\t"
"lwc1 %[temp1], 0(%[src1]) \n\t"
"lwc1 %[temp3], 4(%[src0]) \n\t"
"lwc1 %[temp4], -4(%[src1]) \n\t"
"lwc1 %[temp6], 8(%[src0]) \n\t"
"lwc1 %[temp7], -8(%[src1]) \n\t"
"lwc1 %[temp9], 12(%[src0]) \n\t"
"lwc1 %[temp10], -12(%[src1]) \n\t"
"add.s %[temp2], %[temp0], %[temp1] \n\t"
"add.s %[temp5], %[temp3], %[temp4] \n\t"
"add.s %[temp8], %[temp6], %[temp7] \n\t"
"add.s %[temp11], %[temp9], %[temp10] \n\t"
"sub.s %[temp0], %[temp0], %[temp1] \n\t"
"sub.s %[temp3], %[temp3], %[temp4] \n\t"
"sub.s %[temp6], %[temp6], %[temp7] \n\t"
"sub.s %[temp9], %[temp9], %[temp10] \n\t"
"swc1 %[temp2], 0(%[v1]) \n\t"
"swc1 %[temp0], 0(%[v0]) \n\t"
"swc1 %[temp5], -4(%[v1]) \n\t"
"swc1 %[temp3], 4(%[v0]) \n\t"
"swc1 %[temp8], -8(%[v1]) \n\t"
"swc1 %[temp6], 8(%[v0]) \n\t"
"swc1 %[temp11], -12(%[v1]) \n\t"
"swc1 %[temp9], 12(%[v0]) \n\t"
"lwc1 %[temp0], 16(%[src0]) \n\t"
"lwc1 %[temp1], -16(%[src1]) \n\t"
"lwc1 %[temp3], 20(%[src0]) \n\t"
"lwc1 %[temp4], -20(%[src1]) \n\t"
"lwc1 %[temp6], 24(%[src0]) \n\t"
"lwc1 %[temp7], -24(%[src1]) \n\t"
"lwc1 %[temp9], 28(%[src0]) \n\t"
"lwc1 %[temp10], -28(%[src1]) \n\t"
"add.s %[temp2], %[temp0], %[temp1] \n\t"
"add.s %[temp5], %[temp3], %[temp4] \n\t"
"add.s %[temp8], %[temp6], %[temp7] \n\t"
"add.s %[temp11], %[temp9], %[temp10] \n\t"
"sub.s %[temp0], %[temp0], %[temp1] \n\t"
"sub.s %[temp3], %[temp3], %[temp4] \n\t"
"sub.s %[temp6], %[temp6], %[temp7] \n\t"
"sub.s %[temp9], %[temp9], %[temp10] \n\t"
"swc1 %[temp2], -16(%[v1]) \n\t"
"swc1 %[temp0], 16(%[v0]) \n\t"
"swc1 %[temp5], -20(%[v1]) \n\t"
"swc1 %[temp3], 20(%[v0]) \n\t"
"swc1 %[temp8], -24(%[v1]) \n\t"
"swc1 %[temp6], 24(%[v0]) \n\t"
"swc1 %[temp11], -28(%[v1]) \n\t"
"swc1 %[temp9], 28(%[v0]) \n\t"
"lwc1 %[temp0], 32(%[src0]) \n\t"
"lwc1 %[temp1], -32(%[src1]) \n\t"
"lwc1 %[temp3], 36(%[src0]) \n\t"
"lwc1 %[temp4], -36(%[src1]) \n\t"
"lwc1 %[temp6], 40(%[src0]) \n\t"
"lwc1 %[temp7], -40(%[src1]) \n\t"
"lwc1 %[temp9], 44(%[src0]) \n\t"
"lwc1 %[temp10], -44(%[src1]) \n\t"
"add.s %[temp2], %[temp0], %[temp1] \n\t"
"add.s %[temp5], %[temp3], %[temp4] \n\t"
"add.s %[temp8], %[temp6], %[temp7] \n\t"
"add.s %[temp11], %[temp9], %[temp10] \n\t"
"sub.s %[temp0], %[temp0], %[temp1] \n\t"
"sub.s %[temp3], %[temp3], %[temp4] \n\t"
"sub.s %[temp6], %[temp6], %[temp7] \n\t"
"sub.s %[temp9], %[temp9], %[temp10] \n\t"
"swc1 %[temp2], -32(%[v1]) \n\t"
"swc1 %[temp0], 32(%[v0]) \n\t"
"swc1 %[temp5], -36(%[v1]) \n\t"
"swc1 %[temp3], 36(%[v0]) \n\t"
"swc1 %[temp8], -40(%[v1]) \n\t"
"swc1 %[temp6], 40(%[v0]) \n\t"
"swc1 %[temp11], -44(%[v1]) \n\t"
"swc1 %[temp9], 44(%[v0]) \n\t"
"lwc1 %[temp0], 48(%[src0]) \n\t"
"lwc1 %[temp1], -48(%[src1]) \n\t"
"lwc1 %[temp3], 52(%[src0]) \n\t"
"lwc1 %[temp4], -52(%[src1]) \n\t"
"lwc1 %[temp6], 56(%[src0]) \n\t"
"lwc1 %[temp7], -56(%[src1]) \n\t"
"lwc1 %[temp9], 60(%[src0]) \n\t"
"lwc1 %[temp10], -60(%[src1]) \n\t"
"add.s %[temp2], %[temp0], %[temp1] \n\t"
"add.s %[temp5], %[temp3], %[temp4] \n\t"
"add.s %[temp8], %[temp6], %[temp7] \n\t"
"add.s %[temp11], %[temp9], %[temp10] \n\t"
"sub.s %[temp0], %[temp0], %[temp1] \n\t"
"sub.s %[temp3], %[temp3], %[temp4] \n\t"
"sub.s %[temp6], %[temp6], %[temp7] \n\t"
"sub.s %[temp9], %[temp9], %[temp10] \n\t"
"swc1 %[temp2], -48(%[v1]) \n\t"
"swc1 %[temp0], 48(%[v0]) \n\t"
"swc1 %[temp5], -52(%[v1]) \n\t"
"swc1 %[temp3], 52(%[v0]) \n\t"
"swc1 %[temp8], -56(%[v1]) \n\t"
"swc1 %[temp6], 56(%[v0]) \n\t"
"swc1 %[temp11], -60(%[v1]) \n\t"
"swc1 %[temp9], 60(%[v0]) \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU " %[src0], %[src0], 64 \n\t"
PTR_ADDIU " %[src1], %[src1], -64 \n\t"
PTR_ADDIU " %[v0], %[v0], 64 \n\t"
PTR_ADDIU " %[v1], %[v1], -64 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
: [v0]"+r"(v0), [v1]"+r"(v1), [src0]"+r"(psrc0), [src1]"+r"(psrc1),
[temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
[temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp8]"=&f"(temp8),
[temp9]"=&f"(temp9), [temp10]"=&f"(temp10), [temp11]"=&f"(temp11)
:
:"memory"
);
}
}
static void sbr_autocorrelate_mips(const float x[40][2], float phi[3][2][2])
{
int i;
float real_sum_0 = 0.0f;
float real_sum_1 = 0.0f;
float real_sum_2 = 0.0f;
float imag_sum_1 = 0.0f;
float imag_sum_2 = 0.0f;
float *p_x, *p_phi;
float temp0, temp1, temp2, temp3, temp4, temp5, temp6;
float temp7, temp_r, temp_r1, temp_r2, temp_r3, temp_r4;
p_x = (float*)&x[0][0];
p_phi = &phi[0][0][0];
__asm__ volatile (
"lwc1 %[temp0], 8(%[p_x]) \n\t"
"lwc1 %[temp1], 12(%[p_x]) \n\t"
"lwc1 %[temp2], 16(%[p_x]) \n\t"
"lwc1 %[temp3], 20(%[p_x]) \n\t"
"lwc1 %[temp4], 24(%[p_x]) \n\t"
"lwc1 %[temp5], 28(%[p_x]) \n\t"
"mul.s %[temp_r], %[temp1], %[temp1] \n\t"
"mul.s %[temp_r1], %[temp1], %[temp3] \n\t"
"mul.s %[temp_r2], %[temp1], %[temp2] \n\t"
"mul.s %[temp_r3], %[temp1], %[temp5] \n\t"
"mul.s %[temp_r4], %[temp1], %[temp4] \n\t"
"madd.s %[temp_r], %[temp_r], %[temp0], %[temp0] \n\t"
"madd.s %[temp_r1], %[temp_r1], %[temp0], %[temp2] \n\t"
"msub.s %[temp_r2], %[temp_r2], %[temp0], %[temp3] \n\t"
"madd.s %[temp_r3], %[temp_r3], %[temp0], %[temp4] \n\t"
"msub.s %[temp_r4], %[temp_r4], %[temp0], %[temp5] \n\t"
"add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
"add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
"add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
"add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
"add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[p_x], %[p_x], 8 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
: [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
[real_sum_0]"+f"(real_sum_0), [real_sum_1]"+f"(real_sum_1),
[imag_sum_1]"+f"(imag_sum_1), [real_sum_2]"+f"(real_sum_2),
[temp_r]"=&f"(temp_r), [temp_r1]"=&f"(temp_r1), [temp_r2]"=&f"(temp_r2),
[temp_r3]"=&f"(temp_r3), [temp_r4]"=&f"(temp_r4),
[p_x]"+r"(p_x), [imag_sum_2]"+f"(imag_sum_2)
:
: "memory"
);
for (i = 0; i < 12; i++) {
__asm__ volatile (
"lwc1 %[temp0], 8(%[p_x]) \n\t"
"lwc1 %[temp1], 12(%[p_x]) \n\t"
"lwc1 %[temp2], 16(%[p_x]) \n\t"
"lwc1 %[temp3], 20(%[p_x]) \n\t"
"lwc1 %[temp4], 24(%[p_x]) \n\t"
"lwc1 %[temp5], 28(%[p_x]) \n\t"
"mul.s %[temp_r], %[temp1], %[temp1] \n\t"
"mul.s %[temp_r1], %[temp1], %[temp3] \n\t"
"mul.s %[temp_r2], %[temp1], %[temp2] \n\t"
"mul.s %[temp_r3], %[temp1], %[temp5] \n\t"
"mul.s %[temp_r4], %[temp1], %[temp4] \n\t"
"madd.s %[temp_r], %[temp_r], %[temp0], %[temp0] \n\t"
"madd.s %[temp_r1], %[temp_r1], %[temp0], %[temp2] \n\t"
"msub.s %[temp_r2], %[temp_r2], %[temp0], %[temp3] \n\t"
"madd.s %[temp_r3], %[temp_r3], %[temp0], %[temp4] \n\t"
"msub.s %[temp_r4], %[temp_r4], %[temp0], %[temp5] \n\t"
"add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
"add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
"add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
"add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
"add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
"lwc1 %[temp0], 32(%[p_x]) \n\t"
"lwc1 %[temp1], 36(%[p_x]) \n\t"
"mul.s %[temp_r], %[temp3], %[temp3] \n\t"
"mul.s %[temp_r1], %[temp3], %[temp5] \n\t"
"mul.s %[temp_r2], %[temp3], %[temp4] \n\t"
"mul.s %[temp_r3], %[temp3], %[temp1] \n\t"
"mul.s %[temp_r4], %[temp3], %[temp0] \n\t"
"madd.s %[temp_r], %[temp_r], %[temp2], %[temp2] \n\t"
"madd.s %[temp_r1], %[temp_r1], %[temp2], %[temp4] \n\t"
"msub.s %[temp_r2], %[temp_r2], %[temp2], %[temp5] \n\t"
"madd.s %[temp_r3], %[temp_r3], %[temp2], %[temp0] \n\t"
"msub.s %[temp_r4], %[temp_r4], %[temp2], %[temp1] \n\t"
"add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
"add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
"add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
"add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
"add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
"lwc1 %[temp2], 40(%[p_x]) \n\t"
"lwc1 %[temp3], 44(%[p_x]) \n\t"
"mul.s %[temp_r], %[temp5], %[temp5] \n\t"
"mul.s %[temp_r1], %[temp5], %[temp1] \n\t"
"mul.s %[temp_r2], %[temp5], %[temp0] \n\t"
"mul.s %[temp_r3], %[temp5], %[temp3] \n\t"
"mul.s %[temp_r4], %[temp5], %[temp2] \n\t"
"madd.s %[temp_r], %[temp_r], %[temp4], %[temp4] \n\t"
"madd.s %[temp_r1], %[temp_r1], %[temp4], %[temp0] \n\t"
"msub.s %[temp_r2], %[temp_r2], %[temp4], %[temp1] \n\t"
"madd.s %[temp_r3], %[temp_r3], %[temp4], %[temp2] \n\t"
"msub.s %[temp_r4], %[temp_r4], %[temp4], %[temp3] \n\t"
"add.s %[real_sum_0], %[real_sum_0], %[temp_r] \n\t"
"add.s %[real_sum_1], %[real_sum_1], %[temp_r1] \n\t"
"add.s %[imag_sum_1], %[imag_sum_1], %[temp_r2] \n\t"
"add.s %[real_sum_2], %[real_sum_2], %[temp_r3] \n\t"
"add.s %[imag_sum_2], %[imag_sum_2], %[temp_r4] \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[p_x], %[p_x], 24 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
: [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
[real_sum_0]"+f"(real_sum_0), [real_sum_1]"+f"(real_sum_1),
[imag_sum_1]"+f"(imag_sum_1), [real_sum_2]"+f"(real_sum_2),
[temp_r]"=&f"(temp_r), [temp_r1]"=&f"(temp_r1),
[temp_r2]"=&f"(temp_r2), [temp_r3]"=&f"(temp_r3),
[temp_r4]"=&f"(temp_r4), [p_x]"+r"(p_x),
[imag_sum_2]"+f"(imag_sum_2)
:
: "memory"
);
}
__asm__ volatile (
"lwc1 %[temp0], -296(%[p_x]) \n\t"
"lwc1 %[temp1], -292(%[p_x]) \n\t"
"lwc1 %[temp2], 8(%[p_x]) \n\t"
"lwc1 %[temp3], 12(%[p_x]) \n\t"
"lwc1 %[temp4], -288(%[p_x]) \n\t"
"lwc1 %[temp5], -284(%[p_x]) \n\t"
"lwc1 %[temp6], -280(%[p_x]) \n\t"
"lwc1 %[temp7], -276(%[p_x]) \n\t"
"madd.s %[temp_r], %[real_sum_0], %[temp0], %[temp0] \n\t"
"madd.s %[temp_r1], %[real_sum_0], %[temp2], %[temp2] \n\t"
"madd.s %[temp_r2], %[real_sum_1], %[temp0], %[temp4] \n\t"
"madd.s %[temp_r3], %[imag_sum_1], %[temp0], %[temp5] \n\t"
"madd.s %[temp_r], %[temp_r], %[temp1], %[temp1] \n\t"
"madd.s %[temp_r1], %[temp_r1], %[temp3], %[temp3] \n\t"
"madd.s %[temp_r2], %[temp_r2], %[temp1], %[temp5] \n\t"
"nmsub.s %[temp_r3], %[temp_r3], %[temp1], %[temp4] \n\t"
"lwc1 %[temp4], 16(%[p_x]) \n\t"
"lwc1 %[temp5], 20(%[p_x]) \n\t"
"swc1 %[temp_r], 40(%[p_phi]) \n\t"
"swc1 %[temp_r1], 16(%[p_phi]) \n\t"
"swc1 %[temp_r2], 24(%[p_phi]) \n\t"
"swc1 %[temp_r3], 28(%[p_phi]) \n\t"
"madd.s %[temp_r], %[real_sum_1], %[temp2], %[temp4] \n\t"
"madd.s %[temp_r1], %[imag_sum_1], %[temp2], %[temp5] \n\t"
"madd.s %[temp_r2], %[real_sum_2], %[temp0], %[temp6] \n\t"
"madd.s %[temp_r3], %[imag_sum_2], %[temp0], %[temp7] \n\t"
"madd.s %[temp_r], %[temp_r], %[temp3], %[temp5] \n\t"
"nmsub.s %[temp_r1], %[temp_r1], %[temp3], %[temp4] \n\t"
"madd.s %[temp_r2], %[temp_r2], %[temp1], %[temp7] \n\t"
"nmsub.s %[temp_r3], %[temp_r3], %[temp1], %[temp6] \n\t"
"swc1 %[temp_r], 0(%[p_phi]) \n\t"
"swc1 %[temp_r1], 4(%[p_phi]) \n\t"
"swc1 %[temp_r2], 8(%[p_phi]) \n\t"
"swc1 %[temp_r3], 12(%[p_phi]) \n\t"
: [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
[temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp_r]"=&f"(temp_r),
[real_sum_0]"+f"(real_sum_0), [real_sum_1]"+f"(real_sum_1),
[real_sum_2]"+f"(real_sum_2), [imag_sum_1]"+f"(imag_sum_1),
[temp_r2]"=&f"(temp_r2), [temp_r3]"=&f"(temp_r3),
[temp_r1]"=&f"(temp_r1), [p_phi]"+r"(p_phi),
[imag_sum_2]"+f"(imag_sum_2)
: [p_x]"r"(p_x)
: "memory"
);
}
static void sbr_hf_gen_mips(float (*X_high)[2], const float (*X_low)[2],
const float alpha0[2], const float alpha1[2],
float bw, int start, int end)
{
float alpha[4];
int i;
float *p_x_low = (float*)&X_low[0][0] + 2*start;
float *p_x_high = &X_high[0][0] + 2*start;
float temp0, temp1, temp2, temp3, temp4, temp5, temp6;
float temp7, temp8, temp9, temp10, temp11, temp12;
alpha[0] = alpha1[0] * bw * bw;
alpha[1] = alpha1[1] * bw * bw;
alpha[2] = alpha0[0] * bw;
alpha[3] = alpha0[1] * bw;
for (i = start; i < end; i++) {
__asm__ volatile (
"lwc1 %[temp0], -16(%[p_x_low]) \n\t"
"lwc1 %[temp1], -12(%[p_x_low]) \n\t"
"lwc1 %[temp2], -8(%[p_x_low]) \n\t"
"lwc1 %[temp3], -4(%[p_x_low]) \n\t"
"lwc1 %[temp5], 0(%[p_x_low]) \n\t"
"lwc1 %[temp6], 4(%[p_x_low]) \n\t"
"lwc1 %[temp7], 0(%[alpha]) \n\t"
"lwc1 %[temp8], 4(%[alpha]) \n\t"
"lwc1 %[temp9], 8(%[alpha]) \n\t"
"lwc1 %[temp10], 12(%[alpha]) \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[p_x_high], %[p_x_high], 8 \n\t"
PTR_ADDIU "%[p_x_low], %[p_x_low], 8 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"mul.s %[temp11], %[temp1], %[temp8] \n\t"
"msub.s %[temp11], %[temp11], %[temp0], %[temp7] \n\t"
"madd.s %[temp11], %[temp11], %[temp2], %[temp9] \n\t"
"nmsub.s %[temp11], %[temp11], %[temp3], %[temp10] \n\t"
"add.s %[temp11], %[temp11], %[temp5] \n\t"
"swc1 %[temp11], -8(%[p_x_high]) \n\t"
"mul.s %[temp12], %[temp1], %[temp7] \n\t"
"madd.s %[temp12], %[temp12], %[temp0], %[temp8] \n\t"
"madd.s %[temp12], %[temp12], %[temp3], %[temp9] \n\t"
"madd.s %[temp12], %[temp12], %[temp2], %[temp10] \n\t"
"add.s %[temp12], %[temp12], %[temp6] \n\t"
"swc1 %[temp12], -4(%[p_x_high]) \n\t"
: [temp0]"=&f"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&f"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5),
[temp6]"=&f"(temp6), [temp7]"=&f"(temp7), [temp8]"=&f"(temp8),
[temp9]"=&f"(temp9), [temp10]"=&f"(temp10), [temp11]"=&f"(temp11),
[temp12]"=&f"(temp12), [p_x_high]"+r"(p_x_high),
[p_x_low]"+r"(p_x_low)
: [alpha]"r"(alpha)
: "memory"
);
}
}
static void sbr_hf_g_filt_mips(float (*Y)[2], const float (*X_high)[40][2],
const float *g_filt, int m_max, intptr_t ixh)
{
mips: use float* to hold pointer instead of int This is obviously needed for 64-bit support. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-02-26 15:42:51 +02:00
const float *p_x, *p_g, *loop_end;
float *p_y;
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
float temp0, temp1, temp2;
mips: use float* to hold pointer instead of int This is obviously needed for 64-bit support. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-02-26 15:42:51 +02:00
p_g = &g_filt[0];
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
p_y = &Y[0][0];
mips: use float* to hold pointer instead of int This is obviously needed for 64-bit support. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-02-26 15:42:51 +02:00
p_x = &X_high[0][ixh][0];
loop_end = p_g + m_max;
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
__asm__ volatile(
".set push \n\t"
".set noreorder \n\t"
"1: \n\t"
"lwc1 %[temp0], 0(%[p_g]) \n\t"
"lwc1 %[temp1], 0(%[p_x]) \n\t"
"lwc1 %[temp2], 4(%[p_x]) \n\t"
"mul.s %[temp1], %[temp1], %[temp0] \n\t"
"mul.s %[temp2], %[temp2], %[temp0] \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[p_g], %[p_g], 4 \n\t"
PTR_ADDIU "%[p_x], %[p_x], 320 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"swc1 %[temp1], 0(%[p_y]) \n\t"
"swc1 %[temp2], 4(%[p_y]) \n\t"
"bne %[p_g], %[loop_end], 1b \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDIU "%[p_y], %[p_y], 8 \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
".set pop \n\t"
: [temp0]"=&f"(temp0), [temp1]"=&f"(temp1),
[temp2]"=&f"(temp2), [p_x]"+r"(p_x),
[p_y]"+r"(p_y), [p_g]"+r"(p_g)
: [loop_end]"r"(loop_end)
: "memory"
);
}
static void sbr_hf_apply_noise_0_mips(float (*Y)[2], const float *s_m,
const float *q_filt, int noise,
int kx, int m_max)
{
int m;
for (m = 0; m < m_max; m++){
float *Y1=&Y[m][0];
float *ff_table;
float y0,y1, temp1, temp2, temp4, temp5;
int temp0, temp3;
const float *s_m1=&s_m[m];
const float *q_filt1= &q_filt[m];
__asm__ volatile(
"lwc1 %[y0], 0(%[Y1]) \n\t"
"lwc1 %[temp1], 0(%[s_m1]) \n\t"
"addiu %[noise], %[noise], 1 \n\t"
"andi %[noise], %[noise], 0x1ff \n\t"
"sll %[temp0], %[noise], 3 \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDU "%[ff_table],%[ff_sbr_noise_table], %[temp0] \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"add.s %[y0], %[y0], %[temp1] \n\t"
"mfc1 %[temp3], %[temp1] \n\t"
"bne %[temp3], $0, 1f \n\t"
"lwc1 %[y1], 4(%[Y1]) \n\t"
"lwc1 %[temp2], 0(%[q_filt1]) \n\t"
"lwc1 %[temp4], 0(%[ff_table]) \n\t"
"lwc1 %[temp5], 4(%[ff_table]) \n\t"
"madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
"madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
"swc1 %[y1], 4(%[Y1]) \n\t"
"1: \n\t"
"swc1 %[y0], 0(%[Y1]) \n\t"
: [ff_table]"=&r"(ff_table), [y0]"=&f"(y0), [y1]"=&f"(y1),
[temp0]"=&r"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&r"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5)
: [ff_sbr_noise_table]"r"(ff_sbr_noise_table), [noise]"r"(noise),
[Y1]"r"(Y1), [s_m1]"r"(s_m1), [q_filt1]"r"(q_filt1)
: "memory"
);
}
}
static void sbr_hf_apply_noise_1_mips(float (*Y)[2], const float *s_m,
const float *q_filt, int noise,
int kx, int m_max)
{
float y0,y1,temp1, temp2, temp4, temp5;
int temp0, temp3, m;
float phi_sign = 1 - 2 * (kx & 1);
for (m = 0; m < m_max; m++) {
float *ff_table;
float *Y1=&Y[m][0];
const float *s_m1=&s_m[m];
const float *q_filt1= &q_filt[m];
__asm__ volatile(
"lwc1 %[y1], 4(%[Y1]) \n\t"
"lwc1 %[temp1], 0(%[s_m1]) \n\t"
"lw %[temp3], 0(%[s_m1]) \n\t"
"addiu %[noise], %[noise], 1 \n\t"
"andi %[noise], %[noise], 0x1ff \n\t"
"sll %[temp0], %[noise], 3 \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDU "%[ff_table],%[ff_sbr_noise_table],%[temp0] \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"madd.s %[y1], %[y1], %[temp1], %[phi_sign] \n\t"
"bne %[temp3], $0, 1f \n\t"
"lwc1 %[y0], 0(%[Y1]) \n\t"
"lwc1 %[temp2], 0(%[q_filt1]) \n\t"
"lwc1 %[temp4], 0(%[ff_table]) \n\t"
"lwc1 %[temp5], 4(%[ff_table]) \n\t"
"madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
"madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
"swc1 %[y0], 0(%[Y1]) \n\t"
"1: \n\t"
"swc1 %[y1], 4(%[Y1]) \n\t"
: [ff_table] "=&r" (ff_table), [y0] "=&f" (y0), [y1] "=&f" (y1),
[temp0] "=&r" (temp0), [temp1] "=&f" (temp1), [temp2] "=&f" (temp2),
[temp3] "=&r" (temp3), [temp4] "=&f" (temp4), [temp5] "=&f" (temp5)
: [ff_sbr_noise_table] "r" (ff_sbr_noise_table), [noise] "r" (noise),
[Y1] "r" (Y1), [s_m1] "r" (s_m1), [q_filt1] "r" (q_filt1),
[phi_sign] "f" (phi_sign)
: "memory"
);
phi_sign = -phi_sign;
}
}
static void sbr_hf_apply_noise_2_mips(float (*Y)[2], const float *s_m,
const float *q_filt, int noise,
int kx, int m_max)
{
avcodec/mips: Fix register constraint error reported by clang. Clang report following error in aacsbr_mips.c,ac3dsp_mips.c and aacdec_mips.c: "couldn't allocate output register for constraint 'r'" Use 'f' constraint for float variable. Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2020-07-29 12:10:58 +02:00
int m, temp0, temp1;
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
float *ff_table;
avcodec/mips: Fix register constraint error reported by clang. Clang report following error in aacsbr_mips.c,ac3dsp_mips.c and aacdec_mips.c: "couldn't allocate output register for constraint 'r'" Use 'f' constraint for float variable. Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2020-07-29 12:10:58 +02:00
float y0, y1, temp2, temp3, temp4, temp5;
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
for (m = 0; m < m_max; m++) {
float *Y1=&Y[m][0];
const float *s_m1=&s_m[m];
const float *q_filt1= &q_filt[m];
__asm__ volatile(
"lwc1 %[y0], 0(%[Y1]) \n\t"
avcodec/mips: Fix register constraint error reported by clang. Clang report following error in aacsbr_mips.c,ac3dsp_mips.c and aacdec_mips.c: "couldn't allocate output register for constraint 'r'" Use 'f' constraint for float variable. Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2020-07-29 12:10:58 +02:00
"lwc1 %[temp3], 0(%[s_m1]) \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"addiu %[noise], %[noise], 1 \n\t"
"andi %[noise], %[noise], 0x1ff \n\t"
"sll %[temp0], %[noise], 3 \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDU "%[ff_table],%[ff_sbr_noise_table],%[temp0] \n\t"
avcodec/mips: Fix register constraint error reported by clang. Clang report following error in aacsbr_mips.c,ac3dsp_mips.c and aacdec_mips.c: "couldn't allocate output register for constraint 'r'" Use 'f' constraint for float variable. Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2020-07-29 12:10:58 +02:00
"sub.s %[y0], %[y0], %[temp3] \n\t"
"mfc1 %[temp1], %[temp3] \n\t"
"bne %[temp1], $0, 1f \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"lwc1 %[y1], 4(%[Y1]) \n\t"
"lwc1 %[temp2], 0(%[q_filt1]) \n\t"
"lwc1 %[temp4], 0(%[ff_table]) \n\t"
"lwc1 %[temp5], 4(%[ff_table]) \n\t"
"madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
"madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
"swc1 %[y1], 4(%[Y1]) \n\t"
"1: \n\t"
"swc1 %[y0], 0(%[Y1]) \n\t"
avcodec/mips: Fix register constraint error reported by clang. Clang report following error in aacsbr_mips.c,ac3dsp_mips.c and aacdec_mips.c: "couldn't allocate output register for constraint 'r'" Use 'f' constraint for float variable. Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2020-07-29 12:10:58 +02:00
: [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [y0]"=&f"(y0),
[y1]"=&f"(y1), [ff_table]"=&r"(ff_table),
[temp2]"=&f"(temp2), [temp3]"=&f"(temp3),
[temp4]"=&f"(temp4), [temp5]"=&f"(temp5)
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
: [ff_sbr_noise_table]"r"(ff_sbr_noise_table), [noise]"r"(noise),
[Y1]"r"(Y1), [s_m1]"r"(s_m1), [q_filt1]"r"(q_filt1)
: "memory"
);
}
}
static void sbr_hf_apply_noise_3_mips(float (*Y)[2], const float *s_m,
const float *q_filt, int noise,
int kx, int m_max)
{
float phi_sign = 1 - 2 * (kx & 1);
int m;
for (m = 0; m < m_max; m++) {
float *Y1=&Y[m][0];
float *ff_table;
float y0,y1, temp1, temp2, temp4, temp5;
int temp0, temp3;
const float *s_m1=&s_m[m];
const float *q_filt1= &q_filt[m];
__asm__ volatile(
"lwc1 %[y1], 4(%[Y1]) \n\t"
"lwc1 %[temp1], 0(%[s_m1]) \n\t"
"addiu %[noise], %[noise], 1 \n\t"
"andi %[noise], %[noise], 0x1ff \n\t"
"sll %[temp0], %[noise], 3 \n\t"
mips: port optimizations to mips n64 This mainly consists of replacing all the pointer arithmatic 'addiu' instructions with PTR_ADDIU which will handle the differences in pointer sizes when compiled on 64 bit mips systems. The header asmdefs.h contains the PTR_ macros which expend to the correct mips instructions to manipulate registers containing pointers. Signed-off-by: James Cowgill <james410@cowgill.org.uk> Reviewed-by: Nedeljko Babic <Nedeljko.Babic@imgtec.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-03-05 19:40:15 +02:00
PTR_ADDU "%[ff_table],%[ff_sbr_noise_table], %[temp0] \n\t"
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
"nmsub.s %[y1], %[y1], %[temp1], %[phi_sign] \n\t"
"mfc1 %[temp3], %[temp1] \n\t"
"bne %[temp3], $0, 1f \n\t"
"lwc1 %[y0], 0(%[Y1]) \n\t"
"lwc1 %[temp2], 0(%[q_filt1]) \n\t"
"lwc1 %[temp4], 0(%[ff_table]) \n\t"
"lwc1 %[temp5], 4(%[ff_table]) \n\t"
"madd.s %[y0], %[y0], %[temp2], %[temp4] \n\t"
"madd.s %[y1], %[y1], %[temp2], %[temp5] \n\t"
"swc1 %[y0], 0(%[Y1]) \n\t"
"1: \n\t"
"swc1 %[y1], 4(%[Y1]) \n\t"
: [ff_table]"=&r"(ff_table), [y0]"=&f"(y0), [y1]"=&f"(y1),
[temp0]"=&r"(temp0), [temp1]"=&f"(temp1), [temp2]"=&f"(temp2),
[temp3]"=&r"(temp3), [temp4]"=&f"(temp4), [temp5]"=&f"(temp5)
: [ff_sbr_noise_table]"r"(ff_sbr_noise_table), [noise]"r"(noise),
[Y1]"r"(Y1), [s_m1]"r"(s_m1), [q_filt1]"r"(q_filt1),
[phi_sign]"f"(phi_sign)
: "memory"
);
phi_sign = -phi_sign;
}
}
mips: add support for R6 Understanding the mips32r6 and mips64r6 ISAs in the configure script is not enough. In order to have full support for MIPS R6 in FFmpeg we need to be able to build it, and for that we need to make sure we don't use incompatible assembler code which makes the build fail. Ifdefing the offending code is sufficient to fix the problem. Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2016-02-24 18:38:22 +02:00
#endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
#endif /* HAVE_MIPSFPU */
#endif /* HAVE_INLINE_ASM */
void ff_sbrdsp_init_mips(SBRDSPContext *s)
{
#if HAVE_INLINE_ASM
avcodec/aacdec: fix compilation under soft float MIPS Place HAVE_MIPSFPU further up so that functions that use floating point ASM are defined away. Otherwise compilation failures result when soft float in enabled on the toolchain. Signed-off-by: Rosen Penev <rosenp@gmail.com>
2020-04-06 05:37:10 +02:00
#if HAVE_MIPSFPU
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
s->qmf_pre_shuffle = sbr_qmf_pre_shuffle_mips;
s->qmf_post_shuffle = sbr_qmf_post_shuffle_mips;
mips: add support for R6 Understanding the mips32r6 and mips64r6 ISAs in the configure script is not enough. In order to have full support for MIPS R6 in FFmpeg we need to be able to build it, and for that we need to make sure we don't use incompatible assembler code which makes the build fail. Ifdefing the offending code is sufficient to fix the problem. Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2016-02-24 18:38:22 +02:00
#if !HAVE_MIPS32R6 && !HAVE_MIPS64R6
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
s->sum64x5 = sbr_sum64x5_mips;
s->sum_square = sbr_sum_square_mips;
s->qmf_deint_bfly = sbr_qmf_deint_bfly_mips;
s->autocorrelate = sbr_autocorrelate_mips;
s->hf_gen = sbr_hf_gen_mips;
s->hf_g_filt = sbr_hf_g_filt_mips;
s->hf_apply_noise[0] = sbr_hf_apply_noise_0_mips;
s->hf_apply_noise[1] = sbr_hf_apply_noise_1_mips;
s->hf_apply_noise[2] = sbr_hf_apply_noise_2_mips;
s->hf_apply_noise[3] = sbr_hf_apply_noise_3_mips;
mips: add support for R6 Understanding the mips32r6 and mips64r6 ISAs in the configure script is not enough. In order to have full support for MIPS R6 in FFmpeg we need to be able to build it, and for that we need to make sure we don't use incompatible assembler code which makes the build fail. Ifdefing the offending code is sufficient to fix the problem. Signed-off-by: Vicente Olivert Riera <Vincent.Riera@imgtec.com> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2016-02-24 18:38:22 +02:00
#endif /* !HAVE_MIPS32R6 && !HAVE_MIPS64R6 */
mips: optimization for float aac decoder (sbr module) Signed-off-by: Mirjana Vulin <mvulin@mips.com> Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2013-02-21 16:46:52 +03:00
#endif /* HAVE_MIPSFPU */
#endif /* HAVE_INLINE_ASM */
}
Reference in New Issue Copy Permalink