/*
* Optimization of some functions from mpegvideo.c for armv5te
* Copyright (c) 2007 Siarhei Siamashka
*
* 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 "config.h"
#include "libavutil/arm/asm.S"
/*
* Special optimized version of dct_unquantize_h263_helper_c, it
* requires the block to be at least 8 bytes aligned, and may process
* more elements than requested. But it is guaranteed to never
* process more than 64 elements provided that count argument is <= 64,
* so it is safe. This function is optimized for a common distribution
* of values for nCoeffs (they are mostly multiple of 8 plus one or
* two extra elements). So this function processes data as 8 elements
* per loop iteration and contains optional 2 elements processing in
* the end.
*
* Inner loop should take 6 cycles per element on arm926ej-s (Nokia 770)
*/
.macro dequant_t dst, src, mul, add, tmp
rsbs \tmp, ip, \src, asr #16
it gt
addgt \tmp, \add, #0
it lt
rsblt \tmp, \add, #0
it ne
smlatbne \dst, \src, \mul, \tmp
.endm
.macro dequant_b dst, src, mul, add, tmp
rsbs \tmp, ip, \src, lsl #16
it gt
addgt \tmp, \add, #0
it lt
rsblt \tmp, \add, #0
it ne
smlabbne \dst, \src, \mul, \tmp
.endm
function ff_dct_unquantize_h263_armv5te, export=1
push {r4-r9,lr}
mov ip, #0
subs r3, r3, #2
ble 2f
ldrd r4, r5, [r0, #0]
1:
ldrd r6, r7, [r0, #8]
dequant_t r9, r4, r1, r2, r9
dequant_t lr, r5, r1, r2, lr
dequant_b r4, r4, r1, r2, r8
dequant_b r5, r5, r1, r2, r8
strh r4, [r0], #2
strh r9, [r0], #2
strh r5, [r0], #2
strh lr, [r0], #2
dequant_t r9, r6, r1, r2, r9
dequant_t lr, r7, r1, r2, lr
dequant_b r6, r6, r1, r2, r8
dequant_b r7, r7, r1, r2, r8
strh r6, [r0], #2
strh r9, [r0], #2
strh r7, [r0], #2
strh lr, [r0], #2
subs r3, r3, #8
it gt
ldrdgt r4, r5, [r0, #0] /* load data early to avoid load/use pipeline stall */
bgt 1b
adds r3, r3, #2
it le
pople {r4-r9,pc}
2:
ldrsh r9, [r0, #0]
ldrsh lr, [r0, #2]
mov r8, r2
cmp r9, #0
it lt
rsblt r8, r2, #0
it ne
smlabbne r9, r9, r1, r8
mov r8, r2
cmp lr, #0
it lt
rsblt r8, r2, #0
it ne
smlabbne lr, lr, r1, r8
strh r9, [r0], #2
strh lr, [r0], #2
pop {r4-r9,pc}
endfunc