1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-26 19:01:44 +02:00
FFmpeg/libavcodec/loongarch/h264idct.S
Shiyou Yin e1b6ecd20a
avcodec/la: add LSX optimization for h264 idct.
loongson_asm.S is LoongArch asm optimization helper.
Add functions:
  ff_h264_idct_add_8_lsx
  ff_h264_idct8_add_8_lsx
  ff_h264_idct_dc_add_8_lsx
  ff_h264_idct8_dc_add_8_lsx
  ff_h264_idct_add16_8_lsx
  ff_h264_idct8_add4_8_lsx
  ff_h264_idct_add8_8_lsx
  ff_h264_idct_add8_422_8_lsx
  ff_h264_idct_add16_intra_8_lsx
  ff_h264_luma_dc_dequant_idct_8_lsx
Replaced function(LSX is sufficient for these functions):
  ff_h264_idct_add_lasx
  ff_h264_idct4x4_addblk_dc_lasx
  ff_h264_idct_add16_lasx
  ff_h264_idct8_add4_lasx
  ff_h264_idct_add8_lasx
  ff_h264_idct_add8_422_lasx
  ff_h264_idct_add16_intra_lasx
  ff_h264_deq_idct_luma_dc_lasx
Renamed functions:
  ff_h264_idct8_addblk_lasx ==> ff_h264_idct8_add_8_lasx
  ff_h264_idct8_dc_addblk_lasx ==> ff_h264_idct8_dc_add_8_lasx

./configure --disable-lasx
ffmpeg -i 1_h264_1080p_30fps_3Mbps.mp4 -f rawvideo -y /dev/null -an
before: 155fps
after:  161fps

Reviewed-by: Shiyou Yin <yinshiyou-hf@loongson.cn>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
2023-05-25 21:04:25 +02:00

659 lines
23 KiB
ArmAsm

/*
* Loongson LASX optimized h264idct
*
* Copyright (c) 2023 Loongson Technology Corporation Limited
* Contributed by Shiyou Yin <yinshiyou-hf@loongson.cn>
*
* 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 "loongson_asm.S"
/*
* #define FUNC2(a, b, c) FUNC3(a, b, c)
* #define FUNCC(a) FUNC2(a, BIT_DEPTH, _c)
* void FUNCC(ff_h264_idct_add)(uint8_t *_dst, int16_t *_block, int stride)
* LSX optimization is enough for this function.
*/
function ff_h264_idct_add_8_lsx
fld.d f0, a1, 0
fld.d f1, a1, 8
fld.d f2, a1, 16
fld.d f3, a1, 24
vxor.v vr7, vr7, vr7
add.d t2, a2, a2
add.d t3, t2, a2
vst vr7, a1, 0
vst vr7, a1, 16
vadd.h vr4, vr0, vr2
vsub.h vr5, vr0, vr2
vsrai.h vr6, vr1, 1
vsrai.h vr7, vr3, 1
vsub.h vr6, vr6, vr3
vadd.h vr7, vr1, vr7
LSX_BUTTERFLY_4_H vr4, vr5, vr6, vr7, vr0, vr1, vr2, vr3
LSX_TRANSPOSE4x4_H vr0, vr1, vr2, vr3, vr0, vr1, vr2, vr3, vr4, vr5
vadd.h vr4, vr0, vr2
vsub.h vr5, vr0, vr2
vsrai.h vr6, vr1, 1
vsrai.h vr7, vr3, 1
vsub.h vr6, vr6, vr3
vadd.h vr7, vr1, vr7
LSX_BUTTERFLY_4_H vr4, vr5, vr6, vr7, vr0, vr1, vr2, vr3
fld.s f4, a0, 0
fldx.s f5, a0, a2
fldx.s f6, a0, t2
fldx.s f7, a0, t3
vsrari.h vr0, vr0, 6
vsrari.h vr1, vr1, 6
vsrari.h vr2, vr2, 6
vsrari.h vr3, vr3, 6
vsllwil.hu.bu vr4, vr4, 0
vsllwil.hu.bu vr5, vr5, 0
vsllwil.hu.bu vr6, vr6, 0
vsllwil.hu.bu vr7, vr7, 0
vadd.h vr0, vr0, vr4
vadd.h vr1, vr1, vr5
vadd.h vr2, vr2, vr6
vadd.h vr3, vr3, vr7
vssrarni.bu.h vr1, vr0, 0
vssrarni.bu.h vr3, vr2, 0
vbsrl.v vr0, vr1, 8
vbsrl.v vr2, vr3, 8
fst.s f1, a0, 0
fstx.s f0, a0, a2
fstx.s f3, a0, t2
fstx.s f2, a0, t3
endfunc
/*
* #define FUNC2(a, b, c) FUNC3(a, b, c)
* #define FUNCC(a) FUNC2(a, BIT_DEPTH, _c)
* void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, int16_t *_block, int stride)
*/
function ff_h264_idct8_add_8_lsx
ld.h t0, a1, 0
add.d t2, a2, a2
add.d t3, t2, a2
add.d t4, t3, a2
add.d t5, t4, a2
add.d t6, t5, a2
add.d t7, t6, a2
addi.w t0, t0, 32
st.h t0, a1, 0
vld vr0, a1, 0
vld vr1, a1, 16
vld vr2, a1, 32
vld vr3, a1, 48
vld vr4, a1, 64
vld vr5, a1, 80
vld vr6, a1, 96
vld vr7, a1, 112
vxor.v vr8, vr8, vr8
vst vr8, a1, 0
vst vr8, a1, 16
vst vr8, a1, 32
vst vr8, a1, 48
vst vr8, a1, 64
vst vr8, a1, 80
vst vr8, a1, 96
vst vr8, a1, 112
vadd.h vr18, vr0, vr4
vsub.h vr19, vr0, vr4
vsrai.h vr20, vr2, 1
vsrai.h vr21, vr6, 1
vsub.h vr20, vr20, vr6
vadd.h vr21, vr21, vr2
LSX_BUTTERFLY_4_H vr18, vr19, vr20, vr21, vr10, vr12, vr14, vr16
vsrai.h vr11, vr7, 1
vsrai.h vr13, vr3, 1
vsrai.h vr15, vr5, 1
vsrai.h vr17, vr1, 1
vsub.h vr11, vr5, vr11
vsub.h vr13, vr7, vr13
vadd.h vr15, vr7, vr15
vadd.h vr17, vr5, vr17
vsub.h vr11, vr11, vr7
vsub.h vr13, vr13, vr3
vadd.h vr15, vr15, vr5
vadd.h vr17, vr17, vr1
vsub.h vr11, vr11, vr3
vadd.h vr13, vr13, vr1
vsub.h vr15, vr15, vr1
vadd.h vr17, vr17, vr3
vsrai.h vr18, vr11, 2
vsrai.h vr19, vr13, 2
vsrai.h vr20, vr15, 2
vsrai.h vr21, vr17, 2
vadd.h vr11, vr11, vr21
vadd.h vr13, vr13, vr20
vsub.h vr15, vr19, vr15
vsub.h vr17, vr17, vr18
LSX_BUTTERFLY_8_H vr10, vr16, vr12, vr14, vr13, vr15, vr11, vr17, \
vr0, vr3, vr1, vr2, vr5, vr6, vr4, vr7
LSX_TRANSPOSE8x8_H vr0, vr1, vr2, vr3, vr4, vr5, vr6, vr7, \
vr0, vr1, vr2, vr3, vr4, vr5, vr6, vr7, \
vr10, vr11, vr12, vr13, vr14, vr15, vr16, vr17
vexth.w.h vr20, vr0
vexth.w.h vr21, vr1
vexth.w.h vr22, vr2
vexth.w.h vr23, vr3
vexth.w.h vr8, vr4
vexth.w.h vr9, vr5
vexth.w.h vr18, vr6
vexth.w.h vr19, vr7
vsllwil.w.h vr0, vr0, 0
vsllwil.w.h vr1, vr1, 0
vsllwil.w.h vr2, vr2, 0
vsllwil.w.h vr3, vr3, 0
vsllwil.w.h vr4, vr4, 0
vsllwil.w.h vr5, vr5, 0
vsllwil.w.h vr6, vr6, 0
vsllwil.w.h vr7, vr7, 0
vadd.w vr11, vr0, vr4
vsub.w vr13, vr0, vr4
vsrai.w vr15, vr2, 1
vsrai.w vr17, vr6, 1
vsub.w vr15, vr15, vr6
vadd.w vr17, vr17, vr2
LSX_BUTTERFLY_4_W vr11, vr13, vr15, vr17, vr10, vr12, vr14, vr16
vsrai.w vr11, vr7, 1
vsrai.w vr13, vr3, 1
vsrai.w vr15, vr5, 1
vsrai.w vr17, vr1, 1
vsub.w vr11, vr5, vr11
vsub.w vr13, vr7, vr13
vadd.w vr15, vr7, vr15
vadd.w vr17, vr5, vr17
vsub.w vr11, vr11, vr7
vsub.w vr13, vr13, vr3
vadd.w vr15, vr15, vr5
vadd.w vr17, vr17, vr1
vsub.w vr11, vr11, vr3
vadd.w vr13, vr13, vr1
vsub.w vr15, vr15, vr1
vadd.w vr17, vr17, vr3
vsrai.w vr0, vr11, 2
vsrai.w vr1, vr13, 2
vsrai.w vr2, vr15, 2
vsrai.w vr3, vr17, 2
vadd.w vr11, vr11, vr3
vadd.w vr13, vr13, vr2
vsub.w vr15, vr1, vr15
vsub.w vr17, vr17, vr0
LSX_BUTTERFLY_8_W vr10, vr12, vr14, vr16, vr11, vr13, vr15, vr17, \
vr0, vr1, vr2, vr3, vr4, vr5, vr6, vr7
vadd.w vr11, vr20, vr8
vsub.w vr13, vr20, vr8
vsrai.w vr15, vr22, 1
vsrai.w vr17, vr18, 1
vsub.w vr15, vr15, vr18
vadd.w vr17, vr17, vr22
LSX_BUTTERFLY_4_W vr11, vr13, vr15, vr17, vr10, vr12, vr14, vr16
vsrai.w vr11, vr19, 1
vsrai.w vr13, vr23, 1
vsrai.w vr15, vr9, 1
vsrai.w vr17, vr21, 1
vsub.w vr11, vr9, vr11
vsub.w vr13, vr19, vr13
vadd.w vr15, vr19, vr15
vadd.w vr17, vr9, vr17
vsub.w vr11, vr11, vr19
vsub.w vr13, vr13, vr23
vadd.w vr15, vr15, vr9
vadd.w vr17, vr17, vr21
vsub.w vr11, vr11, vr23
vadd.w vr13, vr13, vr21
vsub.w vr15, vr15, vr21
vadd.w vr17, vr17, vr23
vsrai.w vr20, vr11, 2
vsrai.w vr21, vr13, 2
vsrai.w vr22, vr15, 2
vsrai.w vr23, vr17, 2
vadd.w vr11, vr11, vr23
vadd.w vr13, vr13, vr22
vsub.w vr15, vr21, vr15
vsub.w vr17, vr17, vr20
LSX_BUTTERFLY_8_W vr10, vr12, vr14, vr16, vr11, vr13, vr15, vr17, \
vr20, vr21, vr22, vr23, vr8, vr9, vr18, vr19
vld vr10, a0, 0
vldx vr11, a0, a2
vldx vr12, a0, t2
vldx vr13, a0, t3
vldx vr14, a0, t4
vldx vr15, a0, t5
vldx vr16, a0, t6
vldx vr17, a0, t7
vsrani.h.w vr20, vr0, 6
vsrani.h.w vr21, vr1, 6
vsrani.h.w vr22, vr2, 6
vsrani.h.w vr23, vr3, 6
vsrani.h.w vr8, vr4, 6
vsrani.h.w vr9, vr5, 6
vsrani.h.w vr18, vr6, 6
vsrani.h.w vr19, vr7, 6
vsllwil.hu.bu vr10, vr10, 0
vsllwil.hu.bu vr11, vr11, 0
vsllwil.hu.bu vr12, vr12, 0
vsllwil.hu.bu vr13, vr13, 0
vsllwil.hu.bu vr14, vr14, 0
vsllwil.hu.bu vr15, vr15, 0
vsllwil.hu.bu vr16, vr16, 0
vsllwil.hu.bu vr17, vr17, 0
vadd.h vr0, vr20, vr10
vadd.h vr1, vr21, vr11
vadd.h vr2, vr22, vr12
vadd.h vr3, vr23, vr13
vadd.h vr4, vr8, vr14
vadd.h vr5, vr9, vr15
vadd.h vr6, vr18, vr16
vadd.h vr7, vr19, vr17
vssrarni.bu.h vr1, vr0, 0
vssrarni.bu.h vr3, vr2, 0
vssrarni.bu.h vr5, vr4, 0
vssrarni.bu.h vr7, vr6, 0
vbsrl.v vr0, vr1, 8
vbsrl.v vr2, vr3, 8
vbsrl.v vr4, vr5, 8
vbsrl.v vr6, vr7, 8
fst.d f1, a0, 0
fstx.d f0, a0, a2
fstx.d f3, a0, t2
fstx.d f2, a0, t3
fstx.d f5, a0, t4
fstx.d f4, a0, t5
fstx.d f7, a0, t6
fstx.d f6, a0, t7
endfunc
/*
* #define FUNC2(a, b, c) FUNC3(a, b, c)
* #define FUNCC(a) FUNC2(a, BIT_DEPTH, _c)
* void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, int16_t *_block, int stride)
*/
function ff_h264_idct8_add_8_lasx
ld.h t0, a1, 0
add.d t2, a2, a2
add.d t3, t2, a2
add.d t4, t3, a2
add.d t5, t4, a2
add.d t6, t5, a2
add.d t7, t6, a2
addi.w t0, t0, 32
st.h t0, a1, 0
vld vr0, a1, 0
vld vr1, a1, 16
vld vr2, a1, 32
vld vr3, a1, 48
vld vr4, a1, 64
vld vr5, a1, 80
vld vr6, a1, 96
vld vr7, a1, 112
xvxor.v xr8, xr8, xr8
xvst xr8, a1, 0
xvst xr8, a1, 32
xvst xr8, a1, 64
xvst xr8, a1, 96
vadd.h vr18, vr0, vr4
vsub.h vr19, vr0, vr4
vsrai.h vr20, vr2, 1
vsrai.h vr21, vr6, 1
vsub.h vr20, vr20, vr6
vadd.h vr21, vr21, vr2
LSX_BUTTERFLY_4_H vr18, vr19, vr20, vr21, vr10, vr12, vr14, vr16
vsrai.h vr11, vr7, 1
vsrai.h vr13, vr3, 1
vsrai.h vr15, vr5, 1
vsrai.h vr17, vr1, 1
vsub.h vr11, vr5, vr11
vsub.h vr13, vr7, vr13
vadd.h vr15, vr7, vr15
vadd.h vr17, vr5, vr17
vsub.h vr11, vr11, vr7
vsub.h vr13, vr13, vr3
vadd.h vr15, vr15, vr5
vadd.h vr17, vr17, vr1
vsub.h vr11, vr11, vr3
vadd.h vr13, vr13, vr1
vsub.h vr15, vr15, vr1
vadd.h vr17, vr17, vr3
vsrai.h vr18, vr11, 2
vsrai.h vr19, vr13, 2
vsrai.h vr20, vr15, 2
vsrai.h vr21, vr17, 2
vadd.h vr11, vr11, vr21
vadd.h vr13, vr13, vr20
vsub.h vr15, vr19, vr15
vsub.h vr17, vr17, vr18
LSX_BUTTERFLY_8_H vr10, vr16, vr12, vr14, vr13, vr15, vr11, vr17, \
vr0, vr3, vr1, vr2, vr5, vr6, vr4, vr7
LSX_TRANSPOSE8x8_H vr0, vr1, vr2, vr3, vr4, vr5, vr6, vr7, \
vr0, vr1, vr2, vr3, vr4, vr5, vr6, vr7, \
vr10, vr11, vr12, vr13, vr14, vr15, vr16, vr17
vext2xv.w.h xr0, xr0
vext2xv.w.h xr1, xr1
vext2xv.w.h xr2, xr2
vext2xv.w.h xr3, xr3
vext2xv.w.h xr4, xr4
vext2xv.w.h xr5, xr5
vext2xv.w.h xr6, xr6
vext2xv.w.h xr7, xr7
xvadd.w xr11, xr0, xr4
xvsub.w xr13, xr0, xr4
xvsrai.w xr15, xr2, 1
xvsrai.w xr17, xr6, 1
xvsub.w xr15, xr15, xr6
xvadd.w xr17, xr17, xr2
LASX_BUTTERFLY_4_W xr11, xr13, xr15, xr17, xr10, xr12, xr14, xr16
xvsrai.w xr11, xr7, 1
xvsrai.w xr13, xr3, 1
xvsrai.w xr15, xr5, 1
xvsrai.w xr17, xr1, 1
xvsub.w xr11, xr5, xr11
xvsub.w xr13, xr7, xr13
xvadd.w xr15, xr7, xr15
xvadd.w xr17, xr5, xr17
xvsub.w xr11, xr11, xr7
xvsub.w xr13, xr13, xr3
xvadd.w xr15, xr15, xr5
xvadd.w xr17, xr17, xr1
xvsub.w xr11, xr11, xr3
xvadd.w xr13, xr13, xr1
xvsub.w xr15, xr15, xr1
xvadd.w xr17, xr17, xr3
xvsrai.w xr0, xr11, 2
xvsrai.w xr1, xr13, 2
xvsrai.w xr2, xr15, 2
xvsrai.w xr3, xr17, 2
xvadd.w xr11, xr11, xr3
xvadd.w xr13, xr13, xr2
xvsub.w xr15, xr1, xr15
xvsub.w xr17, xr17, xr0
LASX_BUTTERFLY_8_W xr10, xr12, xr14, xr16, xr11, xr13, xr15, xr17, \
xr0, xr1, xr2, xr3, xr4, xr5, xr6, xr7
vld vr10, a0, 0
vldx vr11, a0, a2
vldx vr12, a0, t2
vldx vr13, a0, t3
vldx vr14, a0, t4
vldx vr15, a0, t5
vldx vr16, a0, t6
vldx vr17, a0, t7
xvldi xr8, 0x806 //"xvldi.w xr8 6"
xvsran.h.w xr0, xr0, xr8
xvsran.h.w xr1, xr1, xr8
xvsran.h.w xr2, xr2, xr8
xvsran.h.w xr3, xr3, xr8
xvsran.h.w xr4, xr4, xr8
xvsran.h.w xr5, xr5, xr8
xvsran.h.w xr6, xr6, xr8
xvsran.h.w xr7, xr7, xr8
xvpermi.d xr0, xr0, 0x08
xvpermi.d xr1, xr1, 0x08
xvpermi.d xr2, xr2, 0x08
xvpermi.d xr3, xr3, 0x08
xvpermi.d xr4, xr4, 0x08
xvpermi.d xr5, xr5, 0x08
xvpermi.d xr6, xr6, 0x08
xvpermi.d xr7, xr7, 0x08
vsllwil.hu.bu vr10, vr10, 0
vsllwil.hu.bu vr11, vr11, 0
vsllwil.hu.bu vr12, vr12, 0
vsllwil.hu.bu vr13, vr13, 0
vsllwil.hu.bu vr14, vr14, 0
vsllwil.hu.bu vr15, vr15, 0
vsllwil.hu.bu vr16, vr16, 0
vsllwil.hu.bu vr17, vr17, 0
vadd.h vr0, vr0, vr10
vadd.h vr1, vr1, vr11
vadd.h vr2, vr2, vr12
vadd.h vr3, vr3, vr13
vadd.h vr4, vr4, vr14
vadd.h vr5, vr5, vr15
vadd.h vr6, vr6, vr16
vadd.h vr7, vr7, vr17
vssrarni.bu.h vr1, vr0, 0
vssrarni.bu.h vr3, vr2, 0
vssrarni.bu.h vr5, vr4, 0
vssrarni.bu.h vr7, vr6, 0
vbsrl.v vr0, vr1, 8
vbsrl.v vr2, vr3, 8
vbsrl.v vr4, vr5, 8
vbsrl.v vr6, vr7, 8
fst.d f1, a0, 0
fstx.d f0, a0, a2
fstx.d f3, a0, t2
fstx.d f2, a0, t3
fstx.d f5, a0, t4
fstx.d f4, a0, t5
fstx.d f7, a0, t6
fstx.d f6, a0, t7
endfunc
/*
* #define FUNC2(a, b, c) FUNC3(a, b, c)
* #define FUNCC(a) FUNC2(a, BIT_DEPTH, _c)
* void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, int16_t *_block, int stride)
* LSX optimization is enough for this function.
*/
function ff_h264_idct_dc_add_8_lsx
vldrepl.h vr4, a1, 0
add.d t2, a2, a2
add.d t3, t2, a2
fld.s f0, a0, 0
fldx.s f1, a0, a2
fldx.s f2, a0, t2
fldx.s f3, a0, t3
st.h zero, a1, 0
vsrari.h vr4, vr4, 6
vilvl.w vr0, vr1, vr0
vilvl.w vr1, vr3, vr2
vsllwil.hu.bu vr0, vr0, 0
vsllwil.hu.bu vr1, vr1, 0
vadd.h vr0, vr0, vr4
vadd.h vr1, vr1, vr4
vssrarni.bu.h vr1, vr0, 0
vbsrl.v vr2, vr1, 4
vbsrl.v vr3, vr1, 8
vbsrl.v vr4, vr1, 12
fst.s f1, a0, 0
fstx.s f2, a0, a2
fstx.s f3, a0, t2
fstx.s f4, a0, t3
endfunc
/*
* #define FUNC2(a, b, c) FUNC3(a, b, c)
* #define FUNCC(a) FUNC2(a, BIT_DEPTH, _c)
* void FUNCC(ff_h264_idct8_dc_add)(uint8_t *_dst, int16_t *_block, int stride)
*/
function ff_h264_idct8_dc_add_8_lsx
vldrepl.h vr8, a1, 0
add.d t2, a2, a2
add.d t3, t2, a2
add.d t4, t3, a2
add.d t5, t4, a2
add.d t6, t5, a2
add.d t7, t6, a2
fld.d f0, a0, 0
fldx.d f1, a0, a2
fldx.d f2, a0, t2
fldx.d f3, a0, t3
fldx.d f4, a0, t4
fldx.d f5, a0, t5
fldx.d f6, a0, t6
fldx.d f7, a0, t7
st.h zero, a1, 0
vsrari.h vr8, vr8, 6
vsllwil.hu.bu vr0, vr0, 0
vsllwil.hu.bu vr1, vr1, 0
vsllwil.hu.bu vr2, vr2, 0
vsllwil.hu.bu vr3, vr3, 0
vsllwil.hu.bu vr4, vr4, 0
vsllwil.hu.bu vr5, vr5, 0
vsllwil.hu.bu vr6, vr6, 0
vsllwil.hu.bu vr7, vr7, 0
vadd.h vr0, vr0, vr8
vadd.h vr1, vr1, vr8
vadd.h vr2, vr2, vr8
vadd.h vr3, vr3, vr8
vadd.h vr4, vr4, vr8
vadd.h vr5, vr5, vr8
vadd.h vr6, vr6, vr8
vadd.h vr7, vr7, vr8
vssrarni.bu.h vr1, vr0, 0
vssrarni.bu.h vr3, vr2, 0
vssrarni.bu.h vr5, vr4, 0
vssrarni.bu.h vr7, vr6, 0
vbsrl.v vr0, vr1, 8
vbsrl.v vr2, vr3, 8
vbsrl.v vr4, vr5, 8
vbsrl.v vr6, vr7, 8
fst.d f1, a0, 0
fstx.d f0, a0, a2
fstx.d f3, a0, t2
fstx.d f2, a0, t3
fstx.d f5, a0, t4
fstx.d f4, a0, t5
fstx.d f7, a0, t6
fstx.d f6, a0, t7
endfunc
function ff_h264_idct8_dc_add_8_lasx
xvldrepl.h xr8, a1, 0
add.d t2, a2, a2
add.d t3, t2, a2
add.d t4, t3, a2
add.d t5, t4, a2
add.d t6, t5, a2
add.d t7, t6, a2
fld.d f0, a0, 0
fldx.d f1, a0, a2
fldx.d f2, a0, t2
fldx.d f3, a0, t3
fldx.d f4, a0, t4
fldx.d f5, a0, t5
fldx.d f6, a0, t6
fldx.d f7, a0, t7
st.h zero, a1, 0
xvsrari.h xr8, xr8, 6
xvpermi.q xr1, xr0, 0x20
xvpermi.q xr3, xr2, 0x20
xvpermi.q xr5, xr4, 0x20
xvpermi.q xr7, xr6, 0x20
xvsllwil.hu.bu xr1, xr1, 0
xvsllwil.hu.bu xr3, xr3, 0
xvsllwil.hu.bu xr5, xr5, 0
xvsllwil.hu.bu xr7, xr7, 0
xvadd.h xr1, xr1, xr8
xvadd.h xr3, xr3, xr8
xvadd.h xr5, xr5, xr8
xvadd.h xr7, xr7, xr8
xvssrarni.bu.h xr3, xr1, 0
xvssrarni.bu.h xr7, xr5, 0
xvpermi.q xr1, xr3, 0x11
xvpermi.q xr5, xr7, 0x11
xvbsrl.v xr0, xr1, 8
xvbsrl.v xr2, xr3, 8
xvbsrl.v xr4, xr5, 8
xvbsrl.v xr6, xr7, 8
fst.d f3, a0, 0
fstx.d f1, a0, a2
fstx.d f2, a0, t2
fstx.d f0, a0, t3
fstx.d f7, a0, t4
fstx.d f5, a0, t5
fstx.d f6, a0, t6
fstx.d f4, a0, t7
endfunc
/**
* IDCT transforms the 16 dc values and dequantizes them.
* @param qmul quantization parameter
* void FUNCC(ff_h264_luma_dc_dequant_idct)(int16_t *_output, int16_t *_input, int qmul){
* LSX optimization is enough for this function.
*/
function ff_h264_luma_dc_dequant_idct_8_lsx
vld vr0, a1, 0
vld vr1, a1, 8
vld vr2, a1, 16
vld vr3, a1, 24
vreplgr2vr.w vr8, a2
LSX_TRANSPOSE4x4_H vr0, vr1, vr2, vr3, vr4, vr5, vr6, vr7, vr9, vr10
LSX_BUTTERFLY_4_H vr4, vr6, vr7, vr5, vr0, vr3, vr2, vr1
LSX_BUTTERFLY_4_H vr0, vr1, vr2, vr3, vr4, vr7, vr6, vr5
LSX_TRANSPOSE4x4_H vr4, vr5, vr6, vr7, vr0, vr1, vr2, vr3, vr9, vr10
LSX_BUTTERFLY_4_H vr0, vr1, vr3, vr2, vr4, vr7, vr6, vr5
LSX_BUTTERFLY_4_H vr4, vr5, vr6, vr7, vr0, vr1, vr2, vr3
vsllwil.w.h vr0, vr0, 0
vsllwil.w.h vr1, vr1, 0
vsllwil.w.h vr2, vr2, 0
vsllwil.w.h vr3, vr3, 0
vmul.w vr0, vr0, vr8
vmul.w vr1, vr1, vr8
vmul.w vr2, vr2, vr8
vmul.w vr3, vr3, vr8
vsrarni.h.w vr1, vr0, 8
vsrarni.h.w vr3, vr2, 8
vstelm.h vr1, a0, 0, 0
vstelm.h vr1, a0, 32, 4
vstelm.h vr1, a0, 64, 1
vstelm.h vr1, a0, 96, 5
vstelm.h vr3, a0, 128, 0
vstelm.h vr3, a0, 160, 4
vstelm.h vr3, a0, 192, 1
vstelm.h vr3, a0, 224, 5
addi.d a0, a0, 256
vstelm.h vr1, a0, 0, 2
vstelm.h vr1, a0, 32, 6
vstelm.h vr1, a0, 64, 3
vstelm.h vr1, a0, 96, 7
vstelm.h vr3, a0, 128, 2
vstelm.h vr3, a0, 160, 6
vstelm.h vr3, a0, 192, 3
vstelm.h vr3, a0, 224, 7
endfunc