Menu
▾
▴
[mpeg2-cvs] CVS: mpeg2dec/libmpeg2 idct_altivec.c,NONE,1.1 Makefile.am,1.1.1.1,1.2 idct.c,1.1.1.1,1.2 mpeg2_internal.h,1.2,1.3
|
From: Michel L. <wa...@us...> - 2001-09-12 05:50:53
|
Update of /cvsroot/libmpeg2/mpeg2dec/libmpeg2
In directory usw-pr-cvs1:/tmp/cvs-serv28781/libmpeg2
Modified Files:
Makefile.am idct.c mpeg2_internal.h
Added Files:
idct_altivec.c
Log Message:
First pass of checkins for altivec IDCT.
This should build on any compiler; however you will need a fairly
recent version of binutils with support for altivec assembly.
This is very untested, so if it breaks you get to keep both pieces :)
--- NEW FILE: idct_altivec.c ---
/*
* idct_altivec.c
* Copyright (C) 1999-2001 Aaron Holtzman <aho...@es...>
*
* This file is part of mpeg2dec, a free MPEG-2 video stream decoder.
*
* mpeg2dec is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* mpeg2dec 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 General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/
#include "config.h"
#ifdef ARCH_PPC
#include <inttypes.h>
#include "mpeg2_internal.h"
#if 0 /* C version of the altivec idct, requires compiler extensions */
#define vector_t vector signed short
#define IDCT_HALF \
/* 1st stage */ \
t1 = vec_mradds (a1, vx7, vx1 ); \
t8 = vec_mradds (a1, vx1, vec_subs (zero, vx7)); \
t7 = vec_mradds (a2, vx5, vx3); \
t3 = vec_mradds (ma2, vx3, vx5); \
\
/* 2nd stage */ \
t5 = vec_adds (vx0, vx4); \
t0 = vec_subs (vx0, vx4); \
t2 = vec_mradds (a0, vx6, vx2); \
t4 = vec_mradds (a0, vx2, vec_subs (zero,vx6)); \
t6 = vec_adds (t8, t3); \
t3 = vec_subs (t8, t3); \
t8 = vec_subs (t1, t7); \
t1 = vec_adds (t1, t7); \
\
/* 3rd stage */ \
t7 = vec_adds (t5, t2); \
t2 = vec_subs (t5, t2); \
t5 = vec_adds (t0, t4); \
t0 = vec_subs (t0, t4); \
t4 = vec_subs (t8, t3); \
t3 = vec_adds (t8, t3); \
\
/* 4th stage */ \
vy0 = vec_adds (t7, t1); \
vy7 = vec_subs (t7, t1); \
vy1 = vec_mradds (c4, t3, t5); \
vy6 = vec_mradds (mc4, t3, t5); \
vy2 = vec_mradds (c4, t4, t0); \
vy5 = vec_mradds (mc4, t4, t0); \
vy3 = vec_adds (t2, t6); \
vy4 = vec_subs (t2, t6);
#define IDCT \
vector_t vx0, vx1, vx2, vx3, vx4, vx5, vx6, vx7; \
vector_t vy0, vy1, vy2, vy3, vy4, vy5, vy6, vy7; \
vector_t a0, a1, a2, ma2, c4, mc4, zero, bias; \
vector_t t0, t1, t2, t3, t4, t5, t6, t7, t8; \
vector unsigned short shift; \
\
c4 = vec_splat (constants[0], 0); \
a0 = vec_splat (constants[0], 1); \
a1 = vec_splat (constants[0], 2); \
a2 = vec_splat (constants[0], 3); \
mc4 = vec_splat (constants[0], 4); \
ma2 = vec_splat (constants[0], 5); \
bias = (vector_t)vec_splat ((vector signed int)constants[0], 3); \
\
zero = vec_splat_s16 (0); \
shift = vec_splat_u16 (4); \
\
vx0 = vec_mradds (vec_sl (block[0], shift), constants[1], zero); \
vx1 = vec_mradds (vec_sl (block[1], shift), constants[2], zero); \
vx2 = vec_mradds (vec_sl (block[2], shift), constants[3], zero); \
vx3 = vec_mradds (vec_sl (block[3], shift), constants[4], zero); \
vx4 = vec_mradds (vec_sl (block[4], shift), constants[1], zero); \
vx5 = vec_mradds (vec_sl (block[5], shift), constants[4], zero); \
vx6 = vec_mradds (vec_sl (block[6], shift), constants[3], zero); \
vx7 = vec_mradds (vec_sl (block[7], shift), constants[2], zero); \
\
IDCT_HALF \
\
vx0 = vec_mergeh (vy0, vy4); \
vx1 = vec_mergel (vy0, vy4); \
vx2 = vec_mergeh (vy1, vy5); \
vx3 = vec_mergel (vy1, vy5); \
vx4 = vec_mergeh (vy2, vy6); \
vx5 = vec_mergel (vy2, vy6); \
vx6 = vec_mergeh (vy3, vy7); \
vx7 = vec_mergel (vy3, vy7); \
\
vy0 = vec_mergeh (vx0, vx4); \
vy1 = vec_mergel (vx0, vx4); \
vy2 = vec_mergeh (vx1, vx5); \
vy3 = vec_mergel (vx1, vx5); \
vy4 = vec_mergeh (vx2, vx6); \
vy5 = vec_mergel (vx2, vx6); \
vy6 = vec_mergeh (vx3, vx7); \
vy7 = vec_mergel (vx3, vx7); \
\
vx0 = vec_adds (vec_mergeh (vy0, vy4), bias); \
vx1 = vec_mergel (vy0, vy4); \
vx2 = vec_mergeh (vy1, vy5); \
vx3 = vec_mergel (vy1, vy5); \
vx4 = vec_mergeh (vy2, vy6); \
vx5 = vec_mergel (vy2, vy6); \
vx6 = vec_mergeh (vy3, vy7); \
vx7 = vec_mergel (vy3, vy7); \
\
IDCT_HALF \
\
shift = vec_splat_u16 (6); \
vx0 = vec_sra (vy0, shift); \
vx1 = vec_sra (vy1, shift); \
vx2 = vec_sra (vy2, shift); \
vx3 = vec_sra (vy3, shift); \
vx4 = vec_sra (vy4, shift); \
vx5 = vec_sra (vy5, shift); \
vx6 = vec_sra (vy6, shift); \
vx7 = vec_sra (vy7, shift);
static vector_t constants[5] = {
(vector_t)(23170, 13573, 6518, 21895, -23170, -21895, 32, 31),
(vector_t)(16384, 22725, 21407, 19266, 16384, 19266, 21407, 22725),
(vector_t)(22725, 31521, 29692, 26722, 22725, 26722, 29692, 31521),
(vector_t)(21407, 29692, 27969, 25172, 21407, 25172, 27969, 29692),
(vector_t)(19266, 26722, 25172, 22654, 19266, 22654, 25172, 26722)
};
void idct_block_add_altivec (vector_t * block, uint8_t * dest, int stride)
{
vector_t tmp, tmp2;
IDCT
#define ADD(dest,src) \
*(uint64_t *)&tmp = *(uint64_t *)dest; \
tmp2 = vec_adds (vec_mergeh (zero, tmp), src); \
tmp = (vector_t)vec_packsu (tmp2, tmp2); \
*(uint64_t *)dest = *(uint64_t *)&tmp;
ADD (dest, vx0) dest += stride;
ADD (dest, vx1) dest += stride;
ADD (dest, vx2) dest += stride;
ADD (dest, vx3) dest += stride;
ADD (dest, vx4) dest += stride;
ADD (dest, vx5) dest += stride;
ADD (dest, vx6) dest += stride;
ADD (dest, vx7)
}
void idct_block_copy_altivec (vector_t * block, uint8_t * dest, int stride)
{
vector_t tmp;
IDCT
#define COPY(dest,src) \
tmp = (vector_t)vec_packsu (src, src); \
*((uint64_t *)dest) = *(uint64_t *)&tmp;
COPY (dest, vx0) dest += stride;
COPY (dest, vx1) dest += stride;
COPY (dest, vx2) dest += stride;
COPY (dest, vx3) dest += stride;
COPY (dest, vx4) dest += stride;
COPY (dest, vx5) dest += stride;
COPY (dest, vx6) dest += stride;
COPY (dest, vx7)
}
#else /* asm version of the altivec idct, converted from the above */
asm (" \n"
" .section \".data\" \n"
" .align 4 \n"
" .type prescale,@object \n"
" .size prescale,64 \n"
"prescale: \n"
" .long 1073764549 \n"
" .long 1402948418 \n"
" .long 1073761090 \n"
" .long 1402951877 \n"
" .long 1489337121 \n"
" .long 1945921634 \n"
" .long 1489332322 \n"
" .long 1945926433 \n"
" .long 1402958844 \n"
" .long 1833001556 \n"
" .long 1402954324 \n"
" .long 1833006076 \n"
" .long 1262643298 \n"
" .long 1649694846 \n"
" .long 1262639230 \n"
" .long 1649698914 \n"
" .align 4 \n"
" .type constants,@object \n"
" .size constants,16 \n"
"constants: \n"
" .long 1518482693 \n"
" .long 427185543 \n"
" .long -1518425479 \n"
" .long 2097183 \n"
" .section \".text\" \n"
" .align 2 \n"
" .globl idct_block_add_altivec \n"
" .type idct_block_add_altivec,@function \n"
"idct_block_add_altivec: \n"
" .extern _savev23 \n"
" .extern _restv23 \n"
" stwu 1,-176(1) \n"
" mflr 0 \n"
" stw 0,180(1) \n"
" addi 0,1,176 \n"
" bl _savev23 \n"
" addi 9,3,112 \n"
" lvx 0,0,3 \n"
" vspltish 25,4 \n"
" lvx 12,0,9 \n"
" lis 10,prescale@ha \n"
" vspltisw 13,0 \n"
" addi 9,3,16 \n"
" lvx 10,0,9 \n"
" la 10,prescale@l(10) \n"
" lvx 5,0,10 \n"
" addi 11,10,16 \n"
" vslh 0,0,25 \n"
" addi 8,3,48 \n"
" lvx 7,0,11 \n"
" vslh 12,12,25 \n"
" addi 9,3,80 \n"
" lvx 11,0,8 \n"
" lvx 1,0,9 \n"
" addi 11,10,48 \n"
" vslh 10,10,25 \n"
" lvx 8,0,11 \n"
" vmhraddshs 18,0,5,13 \n"
" addi 8,3,96 \n"
" addi 11,3,32 \n"
" lvx 0,0,8 \n"
" vmhraddshs 27,12,7,13 \n"
" lis 9,constants@ha \n"
" lvx 9,0,11 \n"
" vslh 11,11,25 \n"
" vslh 1,1,25 \n"
" la 9,constants@l(9) \n"
" lvx 12,0,9 \n"
" vmhraddshs 14,10,7,13 \n"
" addi 10,10,32 \n"
" lvx 10,0,10 \n"
" vmhraddshs 23,1,8,13 \n"
" addi 3,3,64 \n"
" vmhraddshs 24,11,8,13 \n"
" lvx 1,0,3 \n"
" lfd 0,0(4) \n"
" vslh 9,9,25 \n"
" addi 9,1,16 \n"
" vslh 0,0,25 \n"
" vsplth 8,12,2 \n"
" stfd 0,16(1) \n"
" vsubshs 11,13,27 \n"
" vsplth 3,12,5 \n"
" lvx 6,0,9 \n"
" vmhraddshs 19,0,10,13 \n"
" vsplth 4,12,3 \n"
" vmhraddshs 28,9,10,13 \n"
" vsplth 7,12,1 \n"
" vmhraddshs 31,8,14,11 \n"
" vsplth 10,12,0 \n"
" vmhraddshs 2,3,24,23 \n"
" vsplth 11,12,4 \n"
" vslh 1,1,25 \n"
" vspltw 12,12,3 \n"
" vmhraddshs 26,4,23,24 \n"
" vspltish 25,6 \n"
" vmhraddshs 9,1,5,13 \n"
" vmrghh 6,13,6 \n"
" vmhraddshs 27,8,27,14 \n"
" vsubshs 0,13,19 \n"
" vmhraddshs 14,7,19,28 \n"
" vaddshs 23,31,2 \n"
" vmhraddshs 1,7,28,0 \n"
" vsubshs 2,31,2 \n"
" vaddshs 19,18,9 \n"
" vsubshs 31,27,26 \n"
" vaddshs 27,27,26 \n"
" vsubshs 9,18,9 \n"
" vaddshs 26,19,14 \n"
" vsubshs 14,19,14 \n"
" vaddshs 19,9,1 \n"
" vsubshs 9,9,1 \n"
" vsubshs 1,31,2 \n"
" vaddshs 2,31,2 \n"
" vmhraddshs 15,11,1,9 \n"
" vmhraddshs 29,11,2,19 \n"
" vmhraddshs 16,10,2,19 \n"
" vmhraddshs 17,10,1,9 \n"
" vaddshs 30,14,23 \n"
" vsubshs 18,14,23 \n"
" vaddshs 5,26,27 \n"
" vsubshs 1,26,27 \n"
" vmrglh 24,16,15 \n"
" vmrglh 14,5,18 \n"
" vmrglh 23,17,29 \n"
" vmrglh 27,30,1 \n"
" vmrghh 19,30,1 \n"
" vmrghh 9,17,29 \n"
" vmrghh 18,5,18 \n"
" vmrglh 30,14,23 \n"
" vmrglh 1,24,27 \n"
" vmrghh 28,16,15 \n"
" vmrglh 16,18,9 \n"
" vmrghh 5,18,9 \n"
" vmrghh 29,24,27 \n"
" vmrglh 27,30,1 \n"
" vmrghh 18,28,19 \n"
" vmrglh 15,28,19 \n"
" vsubshs 0,13,27 \n"
" vmrghh 17,14,23 \n"
" vmrglh 14,5,18 \n"
" vmrglh 23,17,29 \n"
" vmrglh 24,16,15 \n"
" vmhraddshs 31,8,14,0 \n"
" vmrghh 19,30,1 \n"
" vmhraddshs 27,8,27,14 \n"
" vmhraddshs 2,3,24,23 \n"
" vmrghh 28,16,15 \n"
" vmhraddshs 26,4,23,24 \n"
" vmrghh 0,5,18 \n"
" vmhraddshs 14,7,19,28 \n"
" vmrghh 9,17,29 \n"
" vaddshs 18,0,12 \n"
" vsubshs 1,13,19 \n"
" vaddshs 23,31,2 \n"
" vsubshs 2,31,2 \n"
" vaddshs 19,18,9 \n"
" vsubshs 31,27,26 \n"
" vaddshs 27,27,26 \n"
" vaddshs 26,19,14 \n"
" vsubshs 9,18,9 \n"
" vaddshs 5,26,27 \n"
" vmhraddshs 1,7,28,1 \n"
" vsrah 18,5,25 \n"
" vsubshs 14,19,14 \n"
" vaddshs 0,6,18 \n"
" vaddshs 30,14,23 \n"
" vpkshus 0,0,0 \n"
" vaddshs 19,9,1 \n"
" stvx 0,0,9 \n"
" vsubshs 9,9,1 \n"
" lfd 0,16(1) \n"
" vsubshs 1,31,2 \n"
" stfd 0,0(4) \n"
" vaddshs 2,31,2 \n"
" add 4,4,5 \n"
" vsubshs 18,14,23 \n"
" lfd 0,0(4) \n"
" vmhraddshs 16,10,2,19 \n"
" stfd 0,16(1) \n"
" vmhraddshs 17,10,1,9 \n"
" lvx 0,0,9 \n"
" vmhraddshs 15,11,1,9 \n"
" vsubshs 1,26,27 \n"
" vsrah 24,30,25 \n"
" vsrah 14,16,25 \n"
" vmrghh 0,13,0 \n"
" vsrah 28,17,25 \n"
" vsrah 27,1,25 \n"
" vaddshs 0,0,14 \n"
" vsrah 9,18,25 \n"
" vpkshus 0,0,0 \n"
" vsrah 23,15,25 \n"
" stvx 0,0,9 \n"
" vmhraddshs 29,11,2,19 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" add 4,4,5 \n"
" lfd 0,0(4) \n"
" stfd 0,16(1) \n"
" vsrah 19,29,25 \n"
" lvx 0,0,9 \n"
" vmrghh 0,13,0 \n"
" vaddshs 0,0,28 \n"
" vpkshus 1,0,0 \n"
" stvx 1,0,9 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" add 4,4,5 \n"
" lfd 0,0(4) \n"
" stfd 0,16(1) \n"
" lvx 0,0,9 \n"
" vmrghh 0,13,0 \n"
" vaddshs 0,0,24 \n"
" vpkshus 1,0,0 \n"
" stvx 1,0,9 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" add 4,4,5 \n"
" lfd 0,0(4) \n"
" stfd 0,16(1) \n"
" lvx 0,0,9 \n"
" vmrghh 0,13,0 \n"
" vaddshs 0,0,9 \n"
" vpkshus 1,0,0 \n"
" stvx 1,0,9 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" add 4,4,5 \n"
" lfd 0,0(4) \n"
" stfd 0,16(1) \n"
" lvx 0,0,9 \n"
" vmrghh 0,13,0 \n"
" vaddshs 0,0,23 \n"
" vpkshus 1,0,0 \n"
" stvx 1,0,9 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" add 4,4,5 \n"
" lfd 0,0(4) \n"
" stfd 0,16(1) \n"
" lvx 0,0,9 \n"
" vmrghh 0,13,0 \n"
" vaddshs 0,0,19 \n"
" vpkshus 1,0,0 \n"
" stvx 1,0,9 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" add 4,4,5 \n"
" lfd 0,0(4) \n"
" stfd 0,16(1) \n"
" lvx 0,0,9 \n"
" vmrghh 13,13,0 \n"
" vaddshs 0,13,27 \n"
" vpkshus 0,0,0 \n"
" stvx 0,0,9 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" addi 0,1,176 \n"
" bl _restv23 \n"
" lwz 0,180(1) \n"
" mtlr 0 \n"
" la 1,176(1) \n"
" blr \n"
".Lfe2: \n"
" .size idct_block_add_altivec,.Lfe2-idct_block_add_altivec \n"
" .align 2 \n"
" .globl idct_block_copy_altivec \n"
" .type idct_block_copy_altivec,@function \n"
"idct_block_copy_altivec: \n"
" .extern _savev24 \n"
" .extern _restv24 \n"
" stwu 1,-160(1) \n"
" mflr 0 \n"
" stw 0,164(1) \n"
" addi 0,1,160 \n"
" bl _savev24 \n"
" addi 9,3,112 \n"
" lvx 0,0,3 \n"
" vspltish 25,4 \n"
" lvx 13,0,9 \n"
" lis 10,prescale@ha \n"
" vspltisw 12,0 \n"
" addi 9,3,16 \n"
" lvx 10,0,9 \n"
" la 10,prescale@l(10) \n"
" lvx 6,0,10 \n"
" addi 11,10,16 \n"
" vslh 0,0,25 \n"
" addi 8,3,48 \n"
" lvx 7,0,11 \n"
" vslh 13,13,25 \n"
" addi 9,3,80 \n"
" lvx 11,0,8 \n"
" lvx 1,0,9 \n"
" addi 11,10,48 \n"
" vslh 10,10,25 \n"
" lvx 8,0,11 \n"
" vmhraddshs 19,0,6,12 \n"
" addi 8,3,96 \n"
" addi 11,3,32 \n"
" lvx 0,0,8 \n"
" vmhraddshs 29,13,7,12 \n"
" lis 9,constants@ha \n"
" lvx 9,0,11 \n"
" vslh 11,11,25 \n"
" vslh 1,1,25 \n"
" la 9,constants@l(9) \n"
" lvx 13,0,9 \n"
" vmhraddshs 31,10,7,12 \n"
" addi 10,10,32 \n"
" lvx 10,0,10 \n"
" vmhraddshs 2,1,8,12 \n"
" addi 3,3,64 \n"
" vmhraddshs 17,11,8,12 \n"
" lvx 1,0,3 \n"
" addi 9,1,16 \n"
" vslh 9,9,25 \n"
" vslh 0,0,25 \n"
" vsplth 8,13,2 \n"
" vsubshs 11,12,29 \n"
" vsplth 5,13,5 \n"
" vmhraddshs 3,0,10,12 \n"
" vsplth 4,13,3 \n"
" vmhraddshs 18,9,10,12 \n"
" vsplth 7,13,1 \n"
" vmhraddshs 9,5,17,2 \n"
" vsplth 10,13,0 \n"
" vmhraddshs 16,8,31,11 \n"
" vslh 1,1,25 \n"
" vsplth 11,13,4 \n"
" vmhraddshs 27,4,2,17 \n"
" vspltw 13,13,3 \n"
" vmhraddshs 6,1,6,12 \n"
" vspltish 25,6 \n"
" vmhraddshs 29,8,29,31 \n"
" vmhraddshs 17,7,3,18 \n"
" vsubshs 0,12,3 \n"
" vaddshs 24,16,9 \n"
" vmhraddshs 3,7,18,0 \n"
" vsubshs 9,16,9 \n"
" vaddshs 2,19,6 \n"
" vsubshs 16,29,27 \n"
" vaddshs 29,29,27 \n"
" vsubshs 1,19,6 \n"
" vaddshs 27,2,17 \n"
" vsubshs 17,2,17 \n"
" vaddshs 2,1,3 \n"
" vsubshs 1,1,3 \n"
" vsubshs 3,16,9 \n"
" vaddshs 9,16,9 \n"
" vmhraddshs 6,10,3,1 \n"
" vmhraddshs 26,11,9,2 \n"
" vmhraddshs 28,11,3,1 \n"
" vmhraddshs 15,10,9,2 \n"
" vsubshs 1,17,24 \n"
" vaddshs 19,27,29 \n"
" vaddshs 30,17,24 \n"
" vmrglh 2,6,26 \n"
" vsubshs 14,27,29 \n"
" vmrglh 31,19,1 \n"
" vmrglh 17,15,28 \n"
" vmrghh 6,6,26 \n"
" vmrglh 29,30,14 \n"
" vmrghh 19,19,1 \n"
" vmrghh 3,30,14 \n"
" vmrghh 18,15,28 \n"
" vmrglh 15,19,6 \n"
" vmrglh 30,31,2 \n"
" vmrglh 14,17,29 \n"
" vmrghh 19,19,6 \n"
" vmrghh 26,17,29 \n"
" vmrghh 1,18,3 \n"
" vmrglh 29,30,14 \n"
" vmrglh 28,18,3 \n"
" vmrghh 6,31,2 \n"
" vsubshs 0,12,29 \n"
" vmrglh 31,19,1 \n"
" vmrglh 2,6,26 \n"
" vmrglh 17,15,28 \n"
" vmhraddshs 16,8,31,0 \n"
" vmrghh 3,30,14 \n"
" vmhraddshs 29,8,29,31 \n"
" vmhraddshs 9,5,17,2 \n"
" vmrghh 18,15,28 \n"
" vmhraddshs 27,4,2,17 \n"
" vmrghh 0,19,1 \n"
" vmhraddshs 17,7,3,18 \n"
" vmrghh 6,6,26 \n"
" vsubshs 12,12,3 \n"
" vaddshs 19,0,13 \n"
" vaddshs 24,16,9 \n"
" vmhraddshs 3,7,18,12 \n"
" vsubshs 9,16,9 \n"
" vaddshs 2,19,6 \n"
" vsubshs 16,29,27 \n"
" vaddshs 29,29,27 \n"
" vsubshs 1,19,6 \n"
" vaddshs 27,2,17 \n"
" vsubshs 17,2,17 \n"
" vaddshs 2,1,3 \n"
" vsubshs 1,1,3 \n"
" vsubshs 3,16,9 \n"
" vaddshs 9,16,9 \n"
" vaddshs 19,27,29 \n"
" vmhraddshs 15,10,9,2 \n"
" vmhraddshs 6,10,3,1 \n"
" vsrah 19,19,25 \n"
" vmhraddshs 28,11,3,1 \n"
" vpkshus 0,19,19 \n"
" vaddshs 30,17,24 \n"
" stvx 0,0,9 \n"
" vsrah 31,15,25 \n"
" lfd 0,16(1) \n"
" vsrah 18,6,25 \n"
" stfd 0,0(4) \n"
" vpkshus 0,31,31 \n"
" vsubshs 1,17,24 \n"
" stvx 0,0,9 \n"
" vsrah 17,30,25 \n"
" lfd 0,16(1) \n"
" vpkshus 0,18,18 \n"
" vsrah 6,1,25 \n"
" stvx 0,0,9 \n"
" vpkshus 1,17,17 \n"
" vmhraddshs 26,11,9,2 \n"
" add 4,4,5 \n"
" vsrah 2,28,25 \n"
" vpkshus 0,6,6 \n"
" stfd 0,0(4) \n"
" vsubshs 14,27,29 \n"
" lfd 0,16(1) \n"
" vpkshus 13,2,2 \n"
" stvx 1,0,9 \n"
" vsrah 29,14,25 \n"
" add 4,4,5 \n"
" vsrah 3,26,25 \n"
" stfd 0,0(4) \n"
" vpkshus 12,29,29 \n"
" lfd 0,16(1) \n"
" vpkshus 1,3,3 \n"
" stvx 0,0,9 \n"
" add 4,4,5 \n"
" stfd 0,0(4) \n"
" lfd 13,16(1) \n"
" stvx 13,0,9 \n"
" add 4,4,5 \n"
" stfd 13,0(4) \n"
" lfd 0,16(1) \n"
" add 4,4,5 \n"
" stfd 0,0(4) \n"
" stvx 1,0,9 \n"
" add 4,4,5 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" stvx 12,0,9 \n"
" add 4,4,5 \n"
" lfd 0,16(1) \n"
" stfd 0,0(4) \n"
" addi 0,1,160 \n"
" bl _restv24 \n"
" lwz 0,164(1) \n"
" mtlr 0 \n"
" la 1,160(1) \n"
" blr \n"
".Lfe3: \n"
" .size idct_block_copy_altivec,.Lfe3-idct_block_copy_altivec \n"
);
#endif
void idct_mmx_init (void)
{
extern uint8_t scan_norm[64];
extern uint8_t scan_alt[64];
int i, j;
/* the altivec idct uses a transposed input, so we patch scan tables */
for (i = 0; i < 64; i++) {
j = scan_norm[i];
scan_norm[i] = (j >> 3) | ((j & 7) << 3);
j = scan_alt[i];
scan_alt[i] = (j >> 3) | ((j & 7) << 3);
}
}
#endif
Index: Makefile.am
===================================================================
RCS file: /cvsroot/libmpeg2/mpeg2dec/libmpeg2/Makefile.am,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- Makefile.am 2001/07/19 08:21:13 1.1.1.1
+++ Makefile.am 2001/09/12 05:50:46 1.2
@@ -3,7 +3,7 @@
lib_LTLIBRARIES = libmpeg2.la
libmpeg2_la_SOURCES = decode.c header.c stats.c slice.c \
motion_comp.c motion_comp_mlib.c motion_comp_mmx.c \
- idct.c idct_mlib.c idct_mmx.c
+ idct.c idct_mlib.c idct_mmx.c idct_altivec.c
# the source ordering seems to impact the speed.
# On my setup the following ordering is about 2% faster.
# I'll have to investigate...
Index: idct.c
===================================================================
RCS file: /cvsroot/libmpeg2/mpeg2dec/libmpeg2/idct.c,v
retrieving revision 1.1.1.1
retrieving revision 1.2
diff -u -d -r1.1.1.1 -r1.2
--- idct.c 2001/07/19 08:21:13 1.1.1.1
+++ idct.c 2001/09/12 05:50:46 1.2
@@ -77,6 +77,14 @@
idct_mmx_init ();
} else
#endif
+#ifdef ARCH_PPC
+ if (config.flags & MM_ACCEL_PPC_ALTIVEC) {
+ fprintf (stderr, "Using altivec for IDCT transform\n");
+ idct_block_copy = idct_block_copy_altivec;
+ idct_block_add = idct_block_add_altivec;
+ idct_altivec_init ();
+ } else
+#endif
#ifdef LIBMPEG2_MLIB
if (config.flags & MM_ACCEL_MLIB) {
fprintf (stderr, "Using mlib for IDCT transform\n");
Index: mpeg2_internal.h
===================================================================
RCS file: /cvsroot/libmpeg2/mpeg2dec/libmpeg2/mpeg2_internal.h,v
retrieving revision 1.2
retrieving revision 1.3
diff -u -d -r1.2 -r1.3
--- mpeg2_internal.h 2001/07/21 07:52:47 1.2
+++ mpeg2_internal.h 2001/09/12 05:50:46 1.3
@@ -164,6 +164,10 @@
void idct_block_add_mmx (int16_t * block, uint8_t * dest, int stride);
void idct_mmx_init (void);
+/* idct_altivec.c */
+void idct_block_copy_altivec (int16_t * block, uint8_t * dest, int stride);
+void idct_block_add_altivec (int16_t * block, uint8_t * dest, int stride);
+
/* motion_comp.c */
void motion_comp_init (void);
|