arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
/*
|
|
|
|
|
* Copyright (c) 2016 Google Inc.
|
|
|
|
|
*
|
|
|
|
|
* 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/arm/asm.S"
|
|
|
|
|
#include "neon.S"
|
|
|
|
|
|
|
|
|
|
const itxfm4_coeffs, align=4
|
2016-12-31 12:05:44 +00:00
|
|
|
.short 11585, 0, 6270, 15137
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
iadst4_coeffs:
|
|
|
|
|
.short 5283, 15212, 9929, 13377
|
|
|
|
|
endconst
|
|
|
|
|
|
|
|
|
|
const iadst8_coeffs, align=4
|
|
|
|
|
.short 16305, 1606, 14449, 7723, 10394, 12665, 4756, 15679
|
|
|
|
|
idct_coeffs:
|
2016-12-31 12:05:44 +00:00
|
|
|
.short 11585, 0, 6270, 15137, 3196, 16069, 13623, 9102
|
|
|
|
|
.short 1606, 16305, 12665, 10394, 7723, 14449, 15679, 4756
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.short 804, 16364, 12140, 11003, 7005, 14811, 15426, 5520
|
|
|
|
|
.short 3981, 15893, 14053, 8423, 9760, 13160, 16207, 2404
|
|
|
|
|
endconst
|
|
|
|
|
|
|
|
|
|
const iadst16_coeffs, align=4
|
2016-12-31 20:27:13 +00:00
|
|
|
.short 16364, 804, 15893, 3981, 11003, 12140, 8423, 14053
|
|
|
|
|
.short 14811, 7005, 13160, 9760, 5520, 15426, 2404, 16207
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endconst
|
|
|
|
|
|
|
|
|
|
@ Do four 4x4 transposes, using q registers for the subtransposes that don't
|
|
|
|
|
@ need to address the individual d registers.
|
|
|
|
|
@ r0,r1 == rq1, r2,r3 == rq1, etc
|
|
|
|
|
.macro transpose16_q_4x_4x4 rq0, rq1, rq2, rq3, rq4, rq5, rq6, rq7, r0, r1, r2, r3, r4, r5, r6, r7, r8, r9, r10, r11, r12, r13, r14, r15
|
|
|
|
|
vtrn.32 \rq0, \rq1
|
|
|
|
|
vtrn.32 \rq2, \rq3
|
|
|
|
|
vtrn.32 \rq4, \rq5
|
|
|
|
|
vtrn.32 \rq6, \rq7
|
|
|
|
|
vtrn.16 \r0, \r1
|
|
|
|
|
vtrn.16 \r2, \r3
|
|
|
|
|
vtrn.16 \r4, \r5
|
|
|
|
|
vtrn.16 \r6, \r7
|
|
|
|
|
vtrn.16 \r8, \r9
|
|
|
|
|
vtrn.16 \r10, \r11
|
|
|
|
|
vtrn.16 \r12, \r13
|
|
|
|
|
vtrn.16 \r14, \r15
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1 = ((in1 + in2) * d0[0] + (1 << 13)) >> 14
|
|
|
|
|
@ out2 = ((in1 - in2) * d0[0] + (1 << 13)) >> 14
|
|
|
|
|
@ in/out are d registers
|
|
|
|
|
.macro mbutterfly0 out1, out2, in1, in2, tmpd1, tmpd2, tmpq3, tmpq4, neg=0
|
|
|
|
|
vadd.s16 \tmpd1, \in1, \in2
|
|
|
|
|
vsub.s16 \tmpd2, \in1, \in2
|
|
|
|
|
vmull.s16 \tmpq3, \tmpd1, d0[0]
|
|
|
|
|
vmull.s16 \tmpq4, \tmpd2, d0[0]
|
|
|
|
|
.if \neg > 0
|
|
|
|
|
vneg.s32 \tmpq3, \tmpq3
|
|
|
|
|
.endif
|
|
|
|
|
vrshrn.s32 \out1, \tmpq3, #14
|
|
|
|
|
vrshrn.s32 \out2, \tmpq4, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
@ Same as mbutterfly0 above, but treating the input in in2 as zero,
|
|
|
|
|
@ writing the same output into both out1 and out2.
|
|
|
|
|
.macro mbutterfly0_h out1, out2, in1, in2, tmpd1, tmpd2, tmpq3, tmpq4
|
|
|
|
|
vmull.s16 \tmpq3, \in1, d0[0]
|
|
|
|
|
vrshrn.s32 \out1, \tmpq3, #14
|
|
|
|
|
vrshrn.s32 \out2, \tmpq3, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ out1,out2 = ((in1 + in2) * d0[0] + (1 << 13)) >> 14
|
|
|
|
|
@ out3,out4 = ((in1 - in2) * d0[0] + (1 << 13)) >> 14
|
|
|
|
|
@ Same as mbutterfly0, but with input being 2 q registers, output
|
|
|
|
|
@ being 4 d registers.
|
|
|
|
|
@ This can do with either 4 or 6 temporary q registers.
|
|
|
|
|
.macro dmbutterfly0 out1, out2, out3, out4, in1, in2, tmpq1, tmpq2, tmpd11, tmpd12, tmpd21, tmpd22, tmpq3, tmpq4, tmpq5, tmpq6
|
|
|
|
|
vadd.s16 \tmpq1, \in1, \in2
|
|
|
|
|
vsub.s16 \tmpq2, \in1, \in2
|
|
|
|
|
vmull.s16 \tmpq3, \tmpd11, d0[0]
|
|
|
|
|
vmull.s16 \tmpq4, \tmpd12, d0[0]
|
|
|
|
|
.ifb \tmpq5
|
|
|
|
|
vrshrn.s32 \out1, \tmpq3, #14
|
|
|
|
|
vrshrn.s32 \out2, \tmpq4, #14
|
|
|
|
|
vmull.s16 \tmpq3, \tmpd21, d0[0]
|
|
|
|
|
vmull.s16 \tmpq4, \tmpd22, d0[0]
|
|
|
|
|
vrshrn.s32 \out3, \tmpq3, #14
|
|
|
|
|
vrshrn.s32 \out4, \tmpq4, #14
|
|
|
|
|
.else
|
|
|
|
|
vmull.s16 \tmpq5, \tmpd21, d0[0]
|
|
|
|
|
vmull.s16 \tmpq6, \tmpd22, d0[0]
|
|
|
|
|
vrshrn.s32 \out1, \tmpq3, #14
|
|
|
|
|
vrshrn.s32 \out2, \tmpq4, #14
|
|
|
|
|
vrshrn.s32 \out3, \tmpq5, #14
|
|
|
|
|
vrshrn.s32 \out4, \tmpq6, #14
|
|
|
|
|
.endif
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1 = in1 * coef1 - in2 * coef2
|
|
|
|
|
@ out2 = in1 * coef2 + in2 * coef1
|
|
|
|
|
@ out are 2 q registers, in are 2 d registers
|
|
|
|
|
.macro mbutterfly_l out1, out2, in1, in2, coef1, coef2
|
|
|
|
|
vmull.s16 \out1, \in1, \coef1
|
|
|
|
|
vmlsl.s16 \out1, \in2, \coef2
|
|
|
|
|
vmull.s16 \out2, \in1, \coef2
|
|
|
|
|
vmlal.s16 \out2, \in2, \coef1
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1,out2 = in1,in2 * coef1 - in3,in4 * coef2
|
|
|
|
|
@ out3,out4 = in1,in2 * coef2 + in3,in4 * coef1
|
|
|
|
|
@ out are 4 q registers, in are 4 d registers
|
|
|
|
|
.macro dmbutterfly_l out1, out2, out3, out4, in1, in2, in3, in4, coef1, coef2
|
|
|
|
|
vmull.s16 \out1, \in1, \coef1
|
|
|
|
|
vmull.s16 \out2, \in2, \coef1
|
|
|
|
|
vmull.s16 \out3, \in1, \coef2
|
|
|
|
|
vmull.s16 \out4, \in2, \coef2
|
|
|
|
|
vmlsl.s16 \out1, \in3, \coef2
|
|
|
|
|
vmlsl.s16 \out2, \in4, \coef2
|
|
|
|
|
vmlal.s16 \out3, \in3, \coef1
|
|
|
|
|
vmlal.s16 \out4, \in4, \coef1
|
|
|
|
|
.endm
|
|
|
|
|
|
2017-01-09 22:15:12 +00:00
|
|
|
@ inout1 = (inout1 * coef1 - inout2 * coef2 + (1 << 13)) >> 14
|
|
|
|
|
@ inout2 = (inout1 * coef2 + inout2 * coef1 + (1 << 13)) >> 14
|
|
|
|
|
@ inout are 2 d registers, tmp are 2 q registers
|
|
|
|
|
.macro mbutterfly inout1, inout2, coef1, coef2, tmp1, tmp2, neg=0
|
|
|
|
|
mbutterfly_l \tmp1, \tmp2, \inout1, \inout2, \coef1, \coef2
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.if \neg > 0
|
|
|
|
|
vneg.s32 \tmp2, \tmp2
|
|
|
|
|
.endif
|
2017-01-09 22:15:12 +00:00
|
|
|
vrshrn.s32 \inout1, \tmp1, #14
|
|
|
|
|
vrshrn.s32 \inout2, \tmp2, #14
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.endm
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
@ Same as mbutterfly above, but treating the input in inout2 as zero
|
|
|
|
|
.macro mbutterfly_h1 inout1, inout2, coef1, coef2, tmp1, tmp2
|
|
|
|
|
vmull.s16 \tmp1, \inout1, \coef1
|
|
|
|
|
vmull.s16 \tmp2, \inout1, \coef2
|
|
|
|
|
vrshrn.s32 \inout1, \tmp1, #14
|
|
|
|
|
vrshrn.s32 \inout2, \tmp2, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ Same as mbutterfly above, but treating the input in inout1 as zero
|
|
|
|
|
.macro mbutterfly_h2 inout1, inout2, coef1, coef2, tmp1, tmp2
|
|
|
|
|
vmull.s16 \tmp1, \inout2, \coef2
|
|
|
|
|
vmull.s16 \tmp2, \inout2, \coef1
|
|
|
|
|
vneg.s32 \tmp1, \tmp1
|
|
|
|
|
vrshrn.s32 \inout2, \tmp2, #14
|
|
|
|
|
vrshrn.s32 \inout1, \tmp1, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ inout1,inout2 = (inout1,inout2 * coef1 - inout3,inout4 * coef2 + (1 << 13)) >> 14
|
|
|
|
|
@ inout3,inout4 = (inout1,inout2 * coef2 + inout3,inout4 * coef1 + (1 << 13)) >> 14
|
|
|
|
|
@ inout are 4 d registers, tmp are 4 q registers
|
|
|
|
|
.macro dmbutterfly inout1, inout2, inout3, inout4, coef1, coef2, tmp1, tmp2, tmp3, tmp4
|
|
|
|
|
dmbutterfly_l \tmp1, \tmp2, \tmp3, \tmp4, \inout1, \inout2, \inout3, \inout4, \coef1, \coef2
|
|
|
|
|
vrshrn.s32 \inout1, \tmp1, #14
|
|
|
|
|
vrshrn.s32 \inout2, \tmp2, #14
|
|
|
|
|
vrshrn.s32 \inout3, \tmp3, #14
|
|
|
|
|
vrshrn.s32 \inout4, \tmp4, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1 = in1 + in2
|
|
|
|
|
@ out2 = in1 - in2
|
|
|
|
|
.macro butterfly out1, out2, in1, in2
|
|
|
|
|
vadd.s16 \out1, \in1, \in2
|
|
|
|
|
vsub.s16 \out2, \in1, \in2
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1 = in1 - in2
|
|
|
|
|
@ out2 = in1 + in2
|
|
|
|
|
.macro butterfly_r out1, out2, in1, in2
|
|
|
|
|
vsub.s16 \out1, \in1, \in2
|
|
|
|
|
vadd.s16 \out2, \in1, \in2
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1 = (in1 + in2 + (1 << 13)) >> 14
|
|
|
|
|
@ out2 = (in1 - in2 + (1 << 13)) >> 14
|
|
|
|
|
@ out are 2 d registers, in are 2 q registers, tmp are 2 q registers
|
|
|
|
|
.macro butterfly_n out1, out2, in1, in2, tmp1, tmp2
|
|
|
|
|
vadd.s32 \tmp1, \in1, \in2
|
|
|
|
|
vsub.s32 \tmp2, \in1, \in2
|
|
|
|
|
vrshrn.s32 \out1, \tmp1, #14
|
|
|
|
|
vrshrn.s32 \out2, \tmp2, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ out1,out2 = (in1,in2 + in3,in4 + (1 << 13)) >> 14
|
|
|
|
|
@ out3,out4 = (in1,in2 - in3,in4 + (1 << 13)) >> 14
|
|
|
|
|
@ out are 4 d registers, in are 4 q registers, tmp are 4 q registers
|
|
|
|
|
.macro dbutterfly_n out1, out2, out3, out4, in1, in2, in3, in4, tmp1, tmp2, tmp3, tmp4
|
|
|
|
|
vadd.s32 \tmp1, \in1, \in3
|
|
|
|
|
vadd.s32 \tmp2, \in2, \in4
|
|
|
|
|
vsub.s32 \tmp3, \in1, \in3
|
|
|
|
|
vsub.s32 \tmp4, \in2, \in4
|
|
|
|
|
vrshrn.s32 \out1, \tmp1, #14
|
|
|
|
|
vrshrn.s32 \out2, \tmp2, #14
|
|
|
|
|
vrshrn.s32 \out3, \tmp3, #14
|
|
|
|
|
vrshrn.s32 \out4, \tmp4, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
.macro iwht4 c0, c1, c2, c3
|
|
|
|
|
vadd.i16 \c0, \c0, \c1
|
|
|
|
|
vsub.i16 d17, \c2, \c3
|
|
|
|
|
vsub.i16 d16, \c0, d17
|
|
|
|
|
vshr.s16 d16, d16, #1
|
|
|
|
|
vsub.i16 \c2, d16, \c1
|
|
|
|
|
vsub.i16 \c1, d16, \c3
|
|
|
|
|
vadd.i16 \c3, d17, \c2
|
|
|
|
|
vsub.i16 \c0, \c0, \c1
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
.macro idct4 c0, c1, c2, c3
|
2016-12-31 12:05:44 +00:00
|
|
|
vmull.s16 q13, \c1, d0[3]
|
|
|
|
|
vmull.s16 q11, \c1, d0[2]
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
vadd.i16 d16, \c0, \c2
|
|
|
|
|
vsub.i16 d17, \c0, \c2
|
2016-12-31 12:05:44 +00:00
|
|
|
vmlal.s16 q13, \c3, d0[2]
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
vmull.s16 q9, d16, d0[0]
|
|
|
|
|
vmull.s16 q10, d17, d0[0]
|
2016-12-31 12:05:44 +00:00
|
|
|
vmlsl.s16 q11, \c3, d0[3]
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
vrshrn.s32 d26, q13, #14
|
|
|
|
|
vrshrn.s32 d18, q9, #14
|
|
|
|
|
vrshrn.s32 d20, q10, #14
|
|
|
|
|
vrshrn.s32 d22, q11, #14
|
|
|
|
|
vadd.i16 \c0, d18, d26
|
|
|
|
|
vsub.i16 \c3, d18, d26
|
|
|
|
|
vadd.i16 \c1, d20, d22
|
|
|
|
|
vsub.i16 \c2, d20, d22
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
.macro iadst4 c0, c1, c2, c3
|
|
|
|
|
vmull.s16 q10, \c0, d1[0]
|
|
|
|
|
vmlal.s16 q10, \c2, d1[1]
|
|
|
|
|
vmlal.s16 q10, \c3, d1[2]
|
|
|
|
|
vmull.s16 q11, \c0, d1[2]
|
|
|
|
|
vmlsl.s16 q11, \c2, d1[0]
|
|
|
|
|
vsub.s16 \c0, \c0, \c2
|
|
|
|
|
vmlsl.s16 q11, \c3, d1[1]
|
|
|
|
|
vadd.s16 \c0, \c0, \c3
|
|
|
|
|
vmull.s16 q13, \c1, d1[3]
|
|
|
|
|
vmull.s16 q12, \c0, d1[3]
|
|
|
|
|
vadd.s32 q14, q10, q13
|
|
|
|
|
vadd.s32 q1, q11, q13
|
|
|
|
|
vrshrn.s32 \c0, q14, #14
|
|
|
|
|
vadd.s32 q10, q10, q11
|
|
|
|
|
vrshrn.s32 \c1, q1, #14
|
|
|
|
|
vsub.s32 q10, q10, q13
|
|
|
|
|
vrshrn.s32 \c2, q12, #14
|
|
|
|
|
vrshrn.s32 \c3, q10, #14
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
@ The public functions in this file have got the following signature:
|
|
|
|
|
@ void itxfm_add(uint8_t *dst, ptrdiff_t stride, int16_t *block, int eob);
|
|
|
|
|
|
|
|
|
|
.macro itxfm_func4x4 txfm1, txfm2
|
|
|
|
|
function ff_vp9_\txfm1\()_\txfm2\()_4x4_add_neon, export=1
|
|
|
|
|
.ifc \txfm1,\txfm2
|
|
|
|
|
.ifc \txfm1,idct
|
|
|
|
|
movrel r12, itxfm4_coeffs
|
|
|
|
|
vld1.16 {d0}, [r12,:64]
|
|
|
|
|
.endif
|
|
|
|
|
.ifc \txfm1,iadst
|
|
|
|
|
movrel r12, iadst4_coeffs
|
|
|
|
|
vld1.16 {d1}, [r12,:64]
|
|
|
|
|
.endif
|
|
|
|
|
.else
|
|
|
|
|
movrel r12, itxfm4_coeffs
|
|
|
|
|
vld1.16 {q0}, [r12,:128]
|
|
|
|
|
.endif
|
|
|
|
|
|
|
|
|
|
vmov.i16 q15, #0
|
|
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
cmp r3, #1
|
|
|
|
|
bne 1f
|
|
|
|
|
@ DC-only for idct/idct
|
|
|
|
|
vld1.16 {d4[]}, [r2,:16]
|
|
|
|
|
vmull.s16 q2, d4, d0[0]
|
|
|
|
|
vrshrn.s32 d4, q2, #14
|
|
|
|
|
vmull.s16 q2, d4, d0[0]
|
|
|
|
|
vrshrn.s32 d4, q2, #14
|
|
|
|
|
vst1.16 {d30[0]}, [r2,:16]
|
|
|
|
|
vdup.16 q2, d4[0]
|
|
|
|
|
vmov q3, q2
|
|
|
|
|
b 2f
|
|
|
|
|
.endif
|
|
|
|
|
|
|
|
|
|
1:
|
|
|
|
|
vld1.16 {d4-d7}, [r2,:128]
|
|
|
|
|
vst1.16 {q15}, [r2,:128]!
|
|
|
|
|
|
|
|
|
|
.ifc \txfm1,iwht
|
|
|
|
|
vshr.s16 q2, q2, #2
|
|
|
|
|
vshr.s16 q3, q3, #2
|
|
|
|
|
.endif
|
|
|
|
|
|
|
|
|
|
\txfm1\()4 d4, d5, d6, d7
|
|
|
|
|
|
|
|
|
|
vst1.16 {q15}, [r2,:128]!
|
|
|
|
|
@ Transpose 4x4 with 16 bit elements
|
|
|
|
|
vtrn.16 d4, d5
|
|
|
|
|
vtrn.16 d6, d7
|
|
|
|
|
vtrn.32 q2, q3
|
|
|
|
|
|
|
|
|
|
\txfm2\()4 d4, d5, d6, d7
|
|
|
|
|
2:
|
|
|
|
|
vld1.32 {d0[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d0[1]}, [r0,:32], r1
|
|
|
|
|
.ifnc \txfm1,iwht
|
|
|
|
|
vrshr.s16 q2, q2, #4
|
|
|
|
|
vrshr.s16 q3, q3, #4
|
|
|
|
|
.endif
|
|
|
|
|
vaddw.u8 q2, q2, d0
|
|
|
|
|
vld1.32 {d1[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d1[1]}, [r0,:32], r1
|
|
|
|
|
vqmovun.s16 d0, q2
|
|
|
|
|
sub r0, r0, r1, lsl #2
|
|
|
|
|
|
|
|
|
|
vaddw.u8 q3, q3, d1
|
|
|
|
|
vst1.32 {d0[0]}, [r0,:32], r1
|
|
|
|
|
vqmovun.s16 d1, q3
|
|
|
|
|
|
|
|
|
|
vst1.32 {d0[1]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d1[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d1[1]}, [r0,:32], r1
|
|
|
|
|
|
|
|
|
|
bx lr
|
|
|
|
|
endfunc
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
itxfm_func4x4 idct, idct
|
|
|
|
|
itxfm_func4x4 iadst, idct
|
|
|
|
|
itxfm_func4x4 idct, iadst
|
|
|
|
|
itxfm_func4x4 iadst, iadst
|
|
|
|
|
itxfm_func4x4 iwht, iwht
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
.macro idct8
|
|
|
|
|
dmbutterfly0 d16, d17, d24, d25, q8, q12, q2, q4, d4, d5, d8, d9, q3, q2, q5, q4 @ q8 = t0a, q12 = t1a
|
2016-12-31 12:05:44 +00:00
|
|
|
dmbutterfly d20, d21, d28, d29, d0[2], d0[3], q2, q3, q4, q5 @ q10 = t2a, q14 = t3a
|
|
|
|
|
dmbutterfly d18, d19, d30, d31, d1[0], d1[1], q2, q3, q4, q5 @ q9 = t4a, q15 = t7a
|
|
|
|
|
dmbutterfly d26, d27, d22, d23, d1[2], d1[3], q2, q3, q4, q5 @ q13 = t5a, q11 = t6a
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
butterfly q2, q14, q8, q14 @ q2 = t0, q14 = t3
|
|
|
|
|
butterfly q3, q10, q12, q10 @ q3 = t1, q10 = t2
|
|
|
|
|
butterfly q4, q13, q9, q13 @ q4 = t4, q13 = t5a
|
|
|
|
|
butterfly q5, q11, q15, q11 @ q5 = t7, q11 = t6a
|
|
|
|
|
|
|
|
|
|
butterfly q8, q15, q2, q5 @ q8 = out[0], q15 = out[7]
|
|
|
|
|
|
|
|
|
|
dmbutterfly0 d4, d5, d10, d11, q11, q13, q9, q13, d18, d19, d26, d27, q2, q5, q11, q12 @ q2 = t6, q5 = t5
|
|
|
|
|
|
|
|
|
|
butterfly q11, q12, q14, q4 @ q11 = out[3], q12 = out[4]
|
|
|
|
|
butterfly q9, q14, q3, q2 @ q9 = out[1], q14 = out[6]
|
|
|
|
|
butterfly_r q13, q10, q10, q5 @ q13 = out[5], q10 = out[2]
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
.macro iadst8
|
|
|
|
|
dmbutterfly_l q4, q5, q2, q3, d30, d31, d16, d17, d2[1], d2[0] @ q4,q5 = t1a, q2,q3 = t0a
|
|
|
|
|
dmbutterfly_l q8, q15, q6, q7, d22, d23, d24, d25, d3[1], d3[0] @ q8,q15 = t5a, q6,q7 = t4a
|
|
|
|
|
|
|
|
|
|
dbutterfly_n d22, d23, d4, d5, q2, q3, q6, q7, q11, q12, q2, q3 @ q11 = t0, q2 = t4
|
|
|
|
|
|
|
|
|
|
dbutterfly_n d24, d25, d6, d7, q4, q5, q8, q15, q12, q3, q6, q7 @ q12 = t1, q3 = t5
|
|
|
|
|
|
|
|
|
|
dmbutterfly_l q6, q7, q4, q5, d26, d27, d20, d21, d2[3], d2[2] @ q6,q7 = t3a, q4,q5 = t2a
|
|
|
|
|
dmbutterfly_l q10, q13, q8, q15, d18, d19, d28, d29, d3[3], d3[2] @ q10,q13 = t7a, q8,q15 = t6a
|
|
|
|
|
|
|
|
|
|
dbutterfly_n d18, d19, d8, d9, q4, q5, q8, q15, q9, q14, q4, q5 @ q9 = t2, q4 = t6
|
|
|
|
|
dbutterfly_n d16, d17, d12, d13, q6, q7, q10, q13, q8, q15, q6, q7 @ q8 = t3, q6 = t7
|
|
|
|
|
|
|
|
|
|
butterfly q15, q12, q12, q8 @ q15 = -out[7], q12 = t3
|
|
|
|
|
vneg.s16 q15, q15 @ q15 = out[7]
|
|
|
|
|
butterfly q8, q9, q11, q9 @ q8 = out[0], q9 = t2
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
dmbutterfly_l q10, q11, q5, q7, d4, d5, d6, d7, d0[2], d0[3] @ q10,q11 = t5a, q5,q7 = t4a
|
|
|
|
|
dmbutterfly_l q2, q3, q13, q14, d12, d13, d8, d9, d0[3], d0[2] @ q2,q3 = t6a, q13,q14 = t7a
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
dbutterfly_n d28, d29, d8, d9, q10, q11, q13, q14, q4, q6, q10, q11 @ q14 = out[6], q4 = t7
|
|
|
|
|
|
|
|
|
|
dmbutterfly0 d22, d23, d24, d25, q9, q12, q6, q13, d12, d13, d26, d27, q9, q10 @ q11 = -out[3], q12 = out[4]
|
|
|
|
|
vneg.s16 q11, q11 @ q11 = out[3]
|
|
|
|
|
|
|
|
|
|
dbutterfly_n d18, d19, d4, d5, q5, q7, q2, q3, q9, q10, q2, q3 @ q9 = -out[1], q2 = t6
|
|
|
|
|
vneg.s16 q9, q9 @ q9 = out[1]
|
|
|
|
|
|
|
|
|
|
dmbutterfly0 d20, d21, d26, d27, q2, q4, q3, q5, d6, d7, d10, d11, q6, q7 @ q10 = out[2], q13 = -out[5]
|
|
|
|
|
vneg.s16 q13, q13 @ q13 = out[5]
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
.macro itxfm_func8x8 txfm1, txfm2
|
|
|
|
|
function ff_vp9_\txfm1\()_\txfm2\()_8x8_add_neon, export=1
|
|
|
|
|
@ Push q4-q7 if iadst is used, idct requires
|
|
|
|
|
@ a few scratch registers less, so only push q4-q5
|
|
|
|
|
@ if only idct is involved.
|
|
|
|
|
@ The iadst also uses a few coefficients from
|
|
|
|
|
@ idct, so those always need to be loaded.
|
|
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vpush {q4-q5}
|
|
|
|
|
.else
|
|
|
|
|
movrel r12, iadst8_coeffs
|
|
|
|
|
vld1.16 {q1}, [r12,:128]!
|
|
|
|
|
vpush {q4-q7}
|
|
|
|
|
.endif
|
2017-01-03 14:38:56 +00:00
|
|
|
vld1.16 {q0}, [r12,:128]
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
vmov.i16 q2, #0
|
|
|
|
|
vmov.i16 q3, #0
|
|
|
|
|
|
|
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
cmp r3, #1
|
|
|
|
|
bne 1f
|
|
|
|
|
@ DC-only for idct/idct
|
|
|
|
|
vld1.16 {d16[]}, [r2,:16]
|
|
|
|
|
vmull.s16 q8, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d16, q8, #14
|
|
|
|
|
vmull.s16 q8, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d16, q8, #14
|
|
|
|
|
vdup.16 q8, d16[0]
|
|
|
|
|
vmov q9, q8
|
|
|
|
|
vmov q10, q8
|
|
|
|
|
vmov q11, q8
|
|
|
|
|
vmov q12, q8
|
|
|
|
|
vmov q13, q8
|
|
|
|
|
vmov q14, q8
|
|
|
|
|
vmov q15, q8
|
|
|
|
|
vst1.16 {d4[0]}, [r2,:16]
|
|
|
|
|
b 2f
|
|
|
|
|
.endif
|
|
|
|
|
1:
|
|
|
|
|
vld1.16 {q8-q9}, [r2,:128]!
|
|
|
|
|
vld1.16 {q10-q11}, [r2,:128]!
|
|
|
|
|
vld1.16 {q12-q13}, [r2,:128]!
|
|
|
|
|
vld1.16 {q14-q15}, [r2,:128]!
|
|
|
|
|
sub r2, r2, #128
|
|
|
|
|
vst1.16 {q2-q3}, [r2,:128]!
|
|
|
|
|
vst1.16 {q2-q3}, [r2,:128]!
|
|
|
|
|
vst1.16 {q2-q3}, [r2,:128]!
|
|
|
|
|
vst1.16 {q2-q3}, [r2,:128]!
|
|
|
|
|
|
|
|
|
|
\txfm1\()8
|
|
|
|
|
|
|
|
|
|
@ Transpose 8x8 with 16 bit elements
|
|
|
|
|
vswp d17, d24
|
|
|
|
|
vswp d19, d26
|
|
|
|
|
vswp d21, d28
|
|
|
|
|
vswp d23, d30
|
|
|
|
|
transpose16_4x4 q8, q9, q10, q11, q12, q13, q14, q15
|
|
|
|
|
|
|
|
|
|
\txfm2\()8
|
|
|
|
|
2:
|
|
|
|
|
mov r3, r0
|
|
|
|
|
@ Add into the destination
|
|
|
|
|
vld1.8 {d4}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q8, q8, #5
|
|
|
|
|
vld1.8 {d5}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q9, q9, #5
|
|
|
|
|
vld1.8 {d6}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q10, q10, #5
|
|
|
|
|
vaddw.u8 q8, q8, d4
|
|
|
|
|
vld1.8 {d7}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q11, q11, #5
|
|
|
|
|
vaddw.u8 q9, q9, d5
|
|
|
|
|
vld1.8 {d8}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q12, q12, #5
|
|
|
|
|
vaddw.u8 q10, q10, d6
|
|
|
|
|
vqmovun.s16 d4, q8
|
|
|
|
|
vld1.8 {d9}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q13, q13, #5
|
|
|
|
|
vaddw.u8 q11, q11, d7
|
|
|
|
|
vqmovun.s16 d5, q9
|
|
|
|
|
vld1.8 {d10}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q14, q14, #5
|
|
|
|
|
vaddw.u8 q12, q12, d8
|
|
|
|
|
vqmovun.s16 d6, q10
|
|
|
|
|
vld1.8 {d11}, [r0,:64], r1
|
|
|
|
|
vrshr.s16 q15, q15, #5
|
|
|
|
|
vaddw.u8 q13, q13, d9
|
|
|
|
|
vqmovun.s16 d7, q11
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
vst1.8 {d4}, [r3,:64], r1
|
|
|
|
|
vaddw.u8 q14, q14, d10
|
|
|
|
|
vst1.8 {d5}, [r3,:64], r1
|
|
|
|
|
vqmovun.s16 d8, q12
|
|
|
|
|
vst1.8 {d6}, [r3,:64], r1
|
|
|
|
|
vaddw.u8 q15, q15, d11
|
|
|
|
|
vst1.8 {d7}, [r3,:64], r1
|
|
|
|
|
vqmovun.s16 d9, q13
|
|
|
|
|
vst1.8 {d8}, [r3,:64], r1
|
|
|
|
|
vqmovun.s16 d10, q14
|
|
|
|
|
vst1.8 {d9}, [r3,:64], r1
|
|
|
|
|
vqmovun.s16 d11, q15
|
|
|
|
|
|
|
|
|
|
vst1.8 {d10}, [r3,:64], r1
|
|
|
|
|
vst1.8 {d11}, [r3,:64], r1
|
|
|
|
|
|
|
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
vpop {q4-q5}
|
|
|
|
|
.else
|
|
|
|
|
vpop {q4-q7}
|
|
|
|
|
.endif
|
|
|
|
|
bx lr
|
|
|
|
|
endfunc
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
itxfm_func8x8 idct, idct
|
|
|
|
|
itxfm_func8x8 iadst, idct
|
|
|
|
|
.ltorg
|
|
|
|
|
itxfm_func8x8 idct, iadst
|
|
|
|
|
itxfm_func8x8 iadst, iadst
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
function idct16x16_dc_add_neon
|
|
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vld1.16 {d0}, [r12,:64]
|
|
|
|
|
|
2017-01-08 22:04:19 +00:00
|
|
|
vmov.i16 q2, #0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
vld1.16 {d16[]}, [r2,:16]
|
|
|
|
|
vmull.s16 q8, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d16, q8, #14
|
|
|
|
|
vmull.s16 q8, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d16, q8, #14
|
|
|
|
|
vdup.16 q8, d16[0]
|
|
|
|
|
vst1.16 {d4[0]}, [r2,:16]
|
|
|
|
|
|
|
|
|
|
vrshr.s16 q8, q8, #6
|
|
|
|
|
|
2017-01-04 11:08:51 +00:00
|
|
|
mov r3, r0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
mov r12, #16
|
|
|
|
|
1:
|
|
|
|
|
@ Loop to add the constant from q8 into all 16x16 outputs
|
2017-01-04 11:08:51 +00:00
|
|
|
subs r12, r12, #2
|
|
|
|
|
vld1.8 {q2}, [r0,:128], r1
|
|
|
|
|
vaddw.u8 q10, q8, d4
|
|
|
|
|
vld1.8 {q3}, [r0,:128], r1
|
|
|
|
|
vaddw.u8 q11, q8, d5
|
|
|
|
|
vaddw.u8 q12, q8, d6
|
|
|
|
|
vaddw.u8 q13, q8, d7
|
|
|
|
|
vqmovun.s16 d4, q10
|
|
|
|
|
vqmovun.s16 d5, q11
|
|
|
|
|
vqmovun.s16 d6, q12
|
|
|
|
|
vst1.8 {q2}, [r3,:128], r1
|
|
|
|
|
vqmovun.s16 d7, q13
|
|
|
|
|
vst1.8 {q3}, [r3,:128], r1
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
bne 1b
|
|
|
|
|
|
|
|
|
|
bx lr
|
|
|
|
|
endfunc
|
|
|
|
|
.ltorg
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
.macro idct16_end
|
|
|
|
|
butterfly d18, d7, d4, d7 @ d18 = t0a, d7 = t7a
|
|
|
|
|
butterfly d19, d22, d5, d22 @ d19 = t1a, d22 = t6
|
|
|
|
|
butterfly d4, d26, d20, d26 @ d4 = t2a, d26 = t5
|
|
|
|
|
butterfly d5, d6, d28, d6 @ d5 = t3a, d6 = t4
|
|
|
|
|
butterfly d20, d28, d16, d24 @ d20 = t8a, d28 = t11a
|
|
|
|
|
butterfly d24, d21, d23, d21 @ d24 = t9, d21 = t10
|
|
|
|
|
butterfly d23, d27, d25, d27 @ d23 = t14, d27 = t13
|
|
|
|
|
butterfly d25, d29, d29, d17 @ d25 = t15a, d29 = t12a
|
|
|
|
|
|
|
|
|
|
mbutterfly0 d27, d21, d27, d21, d16, d30, q8, q15 @ d27 = t13a, d21 = t10a
|
|
|
|
|
mbutterfly0 d29, d28, d29, d28, d16, d30, q8, q15 @ d29 = t12, d28 = t11
|
|
|
|
|
|
|
|
|
|
vswp d27, d29 @ d27 = t12, d29 = t13a
|
|
|
|
|
vswp d28, d27 @ d28 = t12, d27 = t11
|
|
|
|
|
butterfly d16, d31, d18, d25 @ d16 = out[0], d31 = out[15]
|
|
|
|
|
butterfly d17, d30, d19, d23 @ d17 = out[1], d30 = out[14]
|
|
|
|
|
butterfly_r d25, d22, d22, d24 @ d25 = out[9], d22 = out[6]
|
|
|
|
|
butterfly d23, d24, d7, d20 @ d23 = out[7], d24 = out[8]
|
|
|
|
|
butterfly d18, d29, d4, d29 @ d18 = out[2], d29 = out[13]
|
|
|
|
|
butterfly d19, d28, d5, d28 @ d19 = out[3], d28 = out[12]
|
|
|
|
|
vmov d4, d21 @ d4 = t10a
|
|
|
|
|
butterfly d20, d27, d6, d27 @ d20 = out[4], d27 = out[11]
|
|
|
|
|
butterfly d21, d26, d26, d4 @ d21 = out[5], d26 = out[10]
|
|
|
|
|
bx lr
|
|
|
|
|
.endm
|
|
|
|
|
|
2016-11-23 08:56:12 +00:00
|
|
|
function idct16
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
mbutterfly0 d16, d24, d16, d24, d4, d6, q2, q3 @ d16 = t0a, d24 = t1a
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly d20, d28, d0[2], d0[3], q2, q3 @ d20 = t2a, d28 = t3a
|
|
|
|
|
mbutterfly d18, d30, d1[0], d1[1], q2, q3 @ d18 = t4a, d30 = t7a
|
|
|
|
|
mbutterfly d26, d22, d1[2], d1[3], q2, q3 @ d26 = t5a, d22 = t6a
|
|
|
|
|
mbutterfly d17, d31, d2[0], d2[1], q2, q3 @ d17 = t8a, d31 = t15a
|
|
|
|
|
mbutterfly d25, d23, d2[2], d2[3], q2, q3 @ d25 = t9a, d23 = t14a
|
|
|
|
|
mbutterfly d21, d27, d3[0], d3[1], q2, q3 @ d21 = t10a, d27 = t13a
|
|
|
|
|
mbutterfly d29, d19, d3[2], d3[3], q2, q3 @ d29 = t11a, d19 = t12a
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
butterfly d4, d28, d16, d28 @ d4 = t0, d28 = t3
|
|
|
|
|
butterfly d5, d20, d24, d20 @ d5 = t1, d20 = t2
|
|
|
|
|
butterfly d6, d26, d18, d26 @ d6 = t4, d26 = t5
|
|
|
|
|
butterfly d7, d22, d30, d22 @ d7 = t7, d22 = t6
|
|
|
|
|
butterfly d16, d25, d17, d25 @ d16 = t8, d25 = t9
|
|
|
|
|
butterfly d24, d21, d29, d21 @ d24 = t11, d21 = t10
|
|
|
|
|
butterfly d17, d27, d19, d27 @ d17 = t12, d27 = t13
|
|
|
|
|
butterfly d29, d23, d31, d23 @ d29 = t15, d23 = t14
|
|
|
|
|
|
|
|
|
|
mbutterfly0 d22, d26, d22, d26, d18, d30, q9, q15 @ d22 = t6a, d26 = t5a
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly d23, d25, d0[2], d0[3], q9, q15 @ d23 = t9a, d25 = t14a
|
|
|
|
|
mbutterfly d27, d21, d0[2], d0[3], q9, q15, neg=1 @ d27 = t13a, d21 = t10a
|
2016-11-22 09:07:38 +00:00
|
|
|
idct16_end
|
|
|
|
|
endfunc
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
function idct16_half
|
|
|
|
|
mbutterfly0_h d16, d24, d16, d24, d4, d6, q2, q3 @ d16 = t0a, d24 = t1a
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_h1 d20, d28, d0[2], d0[3], q2, q3 @ d20 = t2a, d28 = t3a
|
|
|
|
|
mbutterfly_h1 d18, d30, d1[0], d1[1], q2, q3 @ d18 = t4a, d30 = t7a
|
|
|
|
|
mbutterfly_h2 d26, d22, d1[2], d1[3], q2, q3 @ d26 = t5a, d22 = t6a
|
|
|
|
|
mbutterfly_h1 d17, d31, d2[0], d2[1], q2, q3 @ d17 = t8a, d31 = t15a
|
|
|
|
|
mbutterfly_h2 d25, d23, d2[2], d2[3], q2, q3 @ d25 = t9a, d23 = t14a
|
|
|
|
|
mbutterfly_h1 d21, d27, d3[0], d3[1], q2, q3 @ d21 = t10a, d27 = t13a
|
|
|
|
|
mbutterfly_h2 d29, d19, d3[2], d3[3], q2, q3 @ d29 = t11a, d19 = t12a
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d4, d28, d16, d28 @ d4 = t0, d28 = t3
|
|
|
|
|
butterfly d5, d20, d24, d20 @ d5 = t1, d20 = t2
|
|
|
|
|
butterfly d6, d26, d18, d26 @ d6 = t4, d26 = t5
|
|
|
|
|
butterfly d7, d22, d30, d22 @ d7 = t7, d22 = t6
|
|
|
|
|
butterfly d16, d25, d17, d25 @ d16 = t8, d25 = t9
|
|
|
|
|
butterfly d24, d21, d29, d21 @ d24 = t11, d21 = t10
|
|
|
|
|
butterfly d17, d27, d19, d27 @ d17 = t12, d27 = t13
|
|
|
|
|
butterfly d29, d23, d31, d23 @ d29 = t15, d23 = t14
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
mbutterfly0 d22, d26, d22, d26, d18, d30, q9, q15 @ d22 = t6a, d26 = t5a
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly d23, d25, d0[2], d0[3], q9, q15 @ d23 = t9a, d25 = t14a
|
|
|
|
|
mbutterfly d27, d21, d0[2], d0[3], q9, q15, neg=1 @ d27 = t13a, d21 = t10a
|
2016-11-22 09:07:38 +00:00
|
|
|
idct16_end
|
|
|
|
|
endfunc
|
|
|
|
|
|
|
|
|
|
function idct16_quarter
|
2016-12-31 12:05:44 +00:00
|
|
|
vmull.s16 q12, d19, d3[3]
|
|
|
|
|
vmull.s16 q2, d17, d2[0]
|
|
|
|
|
vmull.s16 q3, d18, d1[1]
|
|
|
|
|
vmull.s16 q15, d18, d1[0]
|
2016-11-22 09:07:38 +00:00
|
|
|
vneg.s32 q12, q12
|
2016-12-31 12:05:44 +00:00
|
|
|
vmull.s16 q14, d17, d2[1]
|
|
|
|
|
vmull.s16 q13, d19, d3[2]
|
2016-11-22 09:07:38 +00:00
|
|
|
vmull.s16 q11, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d24, q12, #14
|
|
|
|
|
vrshrn.s32 d16, q2, #14
|
|
|
|
|
vrshrn.s32 d7, q3, #14
|
|
|
|
|
vrshrn.s32 d6, q15, #14
|
|
|
|
|
vrshrn.s32 d29, q14, #14
|
|
|
|
|
vrshrn.s32 d17, q13, #14
|
|
|
|
|
vrshrn.s32 d28, q11, #14
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q10, q11, d17, d24, d0[2], d0[3]
|
|
|
|
|
mbutterfly_l q9, q15, d29, d16, d0[2], d0[3]
|
2016-11-22 09:07:38 +00:00
|
|
|
vneg.s32 q11, q11
|
|
|
|
|
vrshrn.s32 d27, q10, #14
|
|
|
|
|
vrshrn.s32 d21, q11, #14
|
|
|
|
|
vrshrn.s32 d23, q9, #14
|
|
|
|
|
vrshrn.s32 d25, q15, #14
|
|
|
|
|
vmov d4, d28
|
|
|
|
|
vmov d5, d28
|
|
|
|
|
mbutterfly0 d22, d26, d7, d6, d18, d30, q9, q15
|
|
|
|
|
vmov d20, d28
|
|
|
|
|
idct16_end
|
2016-11-23 08:56:12 +00:00
|
|
|
endfunc
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-23 08:56:12 +00:00
|
|
|
function iadst16
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
movrel r12, iadst16_coeffs
|
|
|
|
|
vld1.16 {q0-q1}, [r12,:128]
|
|
|
|
|
|
|
|
|
|
mbutterfly_l q3, q2, d31, d16, d0[1], d0[0] @ q3 = t1, q2 = t0
|
2016-12-31 20:27:13 +00:00
|
|
|
mbutterfly_l q5, q4, d23, d24, d1[1], d1[0] @ q5 = t9, q4 = t8
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_n d31, d24, q3, q5, q6, q5 @ d31 = t1a, d24 = t9a
|
|
|
|
|
mbutterfly_l q7, q6, d29, d18, d0[3], d0[2] @ q7 = t3, q6 = t2
|
|
|
|
|
butterfly_n d16, d23, q2, q4, q3, q4 @ d16 = t0a, d23 = t8a
|
|
|
|
|
|
2016-12-31 20:27:13 +00:00
|
|
|
mbutterfly_l q3, q2, d21, d26, d1[3], d1[2] @ q3 = t11, q2 = t10
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_n d29, d26, q7, q3, q4, q3 @ d29 = t3a, d26 = t11a
|
2016-12-31 20:27:13 +00:00
|
|
|
mbutterfly_l q5, q4, d27, d20, d2[1], d2[0] @ q5 = t5, q4 = t4
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_n d18, d21, q6, q2, q3, q2 @ d18 = t2a, d21 = t10a
|
|
|
|
|
|
|
|
|
|
mbutterfly_l q7, q6, d19, d28, d3[1], d3[0] @ q7 = t13, q6 = t12
|
|
|
|
|
butterfly_n d20, d28, q5, q7, q2, q7 @ d20 = t5a, d28 = t13a
|
2016-12-31 20:27:13 +00:00
|
|
|
mbutterfly_l q3, q2, d25, d22, d2[3], d2[2] @ q3 = t7, q2 = t6
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_n d27, d19, q4, q6, q5, q6 @ d27 = t4a, d19 = t12a
|
|
|
|
|
|
|
|
|
|
mbutterfly_l q5, q4, d17, d30, d3[3], d3[2] @ q5 = t15, q4 = t14
|
|
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vld1.16 {q0}, [r12,:128]
|
|
|
|
|
butterfly_n d22, d30, q3, q5, q6, q5 @ d22 = t7a, d30 = t15a
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q7, q6, d23, d24, d1[0], d1[1] @ q7 = t9, q6 = t8
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_n d25, d17, q2, q4, q3, q4 @ d25 = t6a, d17 = t14a
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q2, q3, d28, d19, d1[1], d1[0] @ q2 = t12, q3 = t13
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_n d23, d19, q6, q2, q4, q2 @ d23 = t8a, d19 = t12a
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q5, q4, d21, d26, d1[2], d1[3] @ q5 = t11, q4 = t10
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_r d4, d27, d16, d27 @ d4 = t4, d27 = t0
|
|
|
|
|
butterfly_n d24, d28, q7, q3, q6, q3 @ d24 = t9a, d28 = t13a
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q6, q7, d30, d17, d1[3], d1[2] @ q6 = t14, q7 = t15
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly_r d5, d20, d31, d20 @ d5 = t5, d20 = t1
|
|
|
|
|
butterfly_n d21, d17, q4, q6, q3, q6 @ d21 = t10a, d17 = t14a
|
|
|
|
|
butterfly_n d26, d30, q5, q7, q4, q7 @ d26 = t11a, d30 = t15a
|
|
|
|
|
|
|
|
|
|
butterfly_r d6, d25, d18, d25 @ d6 = t6, d25 = t2
|
|
|
|
|
butterfly_r d7, d22, d29, d22 @ d7 = t7, d22 = t3
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q5, q4, d19, d28, d0[2], d0[3] @ q5 = t13, q4 = t12
|
|
|
|
|
mbutterfly_l q6, q7, d30, d17, d0[3], d0[2] @ q6 = t14, q7 = t15
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
butterfly_n d18, d30, q4, q6, q8, q6 @ d18 = out[2], d30 = t14a
|
|
|
|
|
butterfly_n d29, d17, q5, q7, q6, q7 @ d29 = -out[13], d17 = t15a
|
|
|
|
|
vneg.s16 d29, d29 @ d29 = out[13]
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q5, q4, d4, d5, d0[2], d0[3] @ q5 = t5a, q4 = t4a
|
|
|
|
|
mbutterfly_l q6, q7, d7, d6, d0[3], d0[2] @ q6 = t6a, q7 = t7a
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
butterfly d2, d6, d27, d25 @ d2 = out[0], d6 = t2a
|
|
|
|
|
butterfly d3, d7, d23, d21 @ d3 =-out[1], d7 = t10
|
|
|
|
|
|
|
|
|
|
butterfly_n d19, d31, q4, q6, q2, q4 @ d19 = -out[3], d31 = t6
|
|
|
|
|
vneg.s16 d19, d19 @ d19 = out[3]
|
|
|
|
|
butterfly_n d28, d16, q5, q7, q2, q5 @ d28 = out[12], d16 = t7
|
|
|
|
|
|
|
|
|
|
butterfly d5, d8, d20, d22 @ d5 =-out[15],d8 = t3a
|
|
|
|
|
butterfly d4, d9, d24, d26 @ d4 = out[14],d9 = t11
|
|
|
|
|
|
|
|
|
|
mbutterfly0 d23, d24, d6, d8, d10, d11, q6, q7, 1 @ d23 = out[7], d24 = out[8]
|
|
|
|
|
mbutterfly0 d20, d27, d16, d31, d10, d11, q6, q7 @ d20 = out[4], d27 = out[11]
|
|
|
|
|
mbutterfly0 d22, d25, d9, d7, d10, d11, q6, q7 @ d22 = out[6], d25 = out[9]
|
|
|
|
|
mbutterfly0 d21, d26, d30, d17, d10, d11, q6, q7, 1 @ d21 = out[5], d26 = out[10]
|
|
|
|
|
|
|
|
|
|
vneg.s16 d31, d5 @ d31 = out[15]
|
|
|
|
|
vneg.s16 d17, d3 @ d17 = out[1]
|
|
|
|
|
|
|
|
|
|
vmov d16, d2
|
|
|
|
|
vmov d30, d4
|
2016-11-23 08:56:12 +00:00
|
|
|
bx lr
|
|
|
|
|
endfunc
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2017-02-05 20:55:20 +00:00
|
|
|
.macro load_add_store coef0, coef1, coef2, coef3
|
|
|
|
|
vrshr.s16 \coef0, \coef0, #6
|
|
|
|
|
vrshr.s16 \coef1, \coef1, #6
|
|
|
|
|
|
|
|
|
|
vld1.32 {d4[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d4[1]}, [r3,:32], r1
|
|
|
|
|
vrshr.s16 \coef2, \coef2, #6
|
|
|
|
|
vrshr.s16 \coef3, \coef3, #6
|
|
|
|
|
vld1.32 {d5[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d5[1]}, [r3,:32], r1
|
|
|
|
|
vaddw.u8 \coef0, \coef0, d4
|
|
|
|
|
vld1.32 {d6[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d6[1]}, [r3,:32], r1
|
|
|
|
|
vaddw.u8 \coef1, \coef1, d5
|
|
|
|
|
vld1.32 {d7[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d7[1]}, [r3,:32], r1
|
|
|
|
|
|
|
|
|
|
vqmovun.s16 d4, \coef0
|
|
|
|
|
vqmovun.s16 d5, \coef1
|
|
|
|
|
sub r0, r0, r1, lsl #2
|
|
|
|
|
sub r3, r3, r1, lsl #2
|
|
|
|
|
vaddw.u8 \coef2, \coef2, d6
|
|
|
|
|
vaddw.u8 \coef3, \coef3, d7
|
|
|
|
|
vst1.32 {d4[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d4[1]}, [r3,:32], r1
|
|
|
|
|
vqmovun.s16 d6, \coef2
|
|
|
|
|
vst1.32 {d5[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d5[1]}, [r3,:32], r1
|
|
|
|
|
vqmovun.s16 d7, \coef3
|
|
|
|
|
|
|
|
|
|
vst1.32 {d6[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d6[1]}, [r3,:32], r1
|
|
|
|
|
vst1.32 {d7[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d7[1]}, [r3,:32], r1
|
|
|
|
|
.endm
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.macro itxfm16_1d_funcs txfm
|
|
|
|
|
@ Read a vertical 4x16 slice out of a 16x16 matrix, do a transform on it,
|
|
|
|
|
@ transpose into a horizontal 16x4 slice and store.
|
|
|
|
|
@ r0 = dst (temp buffer)
|
2017-01-09 22:15:15 +00:00
|
|
|
@ r1 = slice offset
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ r2 = src
|
|
|
|
|
function \txfm\()16_1d_4x16_pass1_neon
|
2016-11-23 08:56:12 +00:00
|
|
|
push {lr}
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
mov r12, #32
|
2017-01-08 22:04:19 +00:00
|
|
|
vmov.s16 q2, #0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
|
|
|
|
vst1.16 {d4}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
|
2016-11-23 08:56:12 +00:00
|
|
|
bl \txfm\()16
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
@ Do four 4x4 transposes. Originally, d16-d31 contain the
|
|
|
|
|
@ 16 rows. Afterwards, d16-d19, d20-d23, d24-d27, d28-d31
|
|
|
|
|
@ contain the transposed 4x4 blocks.
|
|
|
|
|
transpose16_q_4x_4x4 q8, q9, q10, q11, q12, q13, q14, q15, d16, d17, d18, d19, d20, d21, d22, d23, d24, d25, d26, d27, d28, d29, d30, d31
|
|
|
|
|
|
|
|
|
|
@ Store the transposed 4x4 blocks horizontally.
|
2017-01-09 22:15:15 +00:00
|
|
|
cmp r1, #12
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
beq 1f
|
|
|
|
|
.irp i, 16, 20, 24, 28, 17, 21, 25, 29, 18, 22, 26, 30, 19, 23, 27, 31
|
|
|
|
|
vst1.16 {d\i}, [r0,:64]!
|
|
|
|
|
.endr
|
2016-11-23 08:56:12 +00:00
|
|
|
pop {pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
1:
|
2017-01-09 22:15:15 +00:00
|
|
|
@ Special case: For the last input column (r1 == 12),
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ which would be stored as the last row in the temp buffer,
|
|
|
|
|
@ don't store the first 4x4 block, but keep it in registers
|
|
|
|
|
@ for the first slice of the second pass (where it is the
|
|
|
|
|
@ last 4x4 block).
|
|
|
|
|
add r0, r0, #8
|
2017-02-04 20:16:09 +00:00
|
|
|
vst1.16 {d20}, [r0,:64]!
|
|
|
|
|
vst1.16 {d24}, [r0,:64]!
|
|
|
|
|
vst1.16 {d28}, [r0,:64]!
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
add r0, r0, #8
|
2017-02-04 20:16:09 +00:00
|
|
|
vst1.16 {d21}, [r0,:64]!
|
|
|
|
|
vst1.16 {d25}, [r0,:64]!
|
|
|
|
|
vst1.16 {d29}, [r0,:64]!
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
add r0, r0, #8
|
2017-02-04 20:16:09 +00:00
|
|
|
vst1.16 {d22}, [r0,:64]!
|
|
|
|
|
vst1.16 {d26}, [r0,:64]!
|
|
|
|
|
vst1.16 {d30}, [r0,:64]!
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
add r0, r0, #8
|
2017-02-04 20:16:09 +00:00
|
|
|
vst1.16 {d23}, [r0,:64]!
|
|
|
|
|
vst1.16 {d27}, [r0,:64]!
|
|
|
|
|
vst1.16 {d31}, [r0,:64]!
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
vmov d28, d16
|
|
|
|
|
vmov d29, d17
|
|
|
|
|
vmov d30, d18
|
|
|
|
|
vmov d31, d19
|
2016-11-23 08:56:12 +00:00
|
|
|
pop {pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endfunc
|
|
|
|
|
|
|
|
|
|
@ Read a vertical 4x16 slice out of a 16x16 matrix, do a transform on it,
|
|
|
|
|
@ load the destination pixels (from a similar 4x16 slice), add and store back.
|
|
|
|
|
@ r0 = dst
|
|
|
|
|
@ r1 = dst stride
|
|
|
|
|
@ r2 = src (temp buffer)
|
|
|
|
|
@ r3 = slice offset
|
|
|
|
|
function \txfm\()16_1d_4x16_pass2_neon
|
2016-11-23 08:56:12 +00:00
|
|
|
push {lr}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
mov r12, #32
|
|
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
cmp r3, #0
|
|
|
|
|
beq 1f
|
|
|
|
|
.irp i, 28, 29, 30, 31
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
1:
|
|
|
|
|
|
|
|
|
|
add r3, r0, r1
|
|
|
|
|
lsl r1, r1, #1
|
2016-11-23 08:56:12 +00:00
|
|
|
bl \txfm\()16
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
load_add_store q8, q9, q10, q11
|
|
|
|
|
load_add_store q12, q13, q14, q15
|
|
|
|
|
|
2016-11-23 08:56:12 +00:00
|
|
|
pop {pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endfunc
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
itxfm16_1d_funcs idct
|
|
|
|
|
itxfm16_1d_funcs iadst
|
|
|
|
|
|
2017-01-09 22:15:15 +00:00
|
|
|
@ This is the minimum eob value for each subpartition, in increments of 4
|
|
|
|
|
const min_eob_idct_idct_16, align=4
|
|
|
|
|
.short 0, 10, 38, 89
|
|
|
|
|
endconst
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.macro itxfm_func16x16 txfm1, txfm2
|
|
|
|
|
function ff_vp9_\txfm1\()_\txfm2\()_16x16_add_neon, export=1
|
|
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
cmp r3, #1
|
|
|
|
|
beq idct16x16_dc_add_neon
|
|
|
|
|
.endif
|
2017-01-09 22:15:15 +00:00
|
|
|
push {r4-r8,lr}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.ifnc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
vpush {q4-q7}
|
|
|
|
|
.endif
|
|
|
|
|
|
|
|
|
|
@ Align the stack, allocate a temp buffer
|
2017-01-09 22:15:09 +00:00
|
|
|
T mov r7, sp
|
|
|
|
|
T and r7, r7, #15
|
|
|
|
|
A and r7, sp, #15
|
|
|
|
|
add r7, r7, #512
|
|
|
|
|
sub sp, sp, r7
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
mov r4, r0
|
|
|
|
|
mov r5, r1
|
|
|
|
|
mov r6, r2
|
|
|
|
|
|
|
|
|
|
.ifc \txfm1,idct
|
|
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vld1.16 {q0-q1}, [r12,:128]
|
|
|
|
|
.endif
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
cmp r3, #10
|
|
|
|
|
ble idct16x16_quarter_add_neon
|
|
|
|
|
cmp r3, #38
|
|
|
|
|
ble idct16x16_half_add_neon
|
2017-02-26 20:13:10 +00:00
|
|
|
|
|
|
|
|
movrel r8, min_eob_idct_idct_16 + 2
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 0, 4, 8, 12
|
|
|
|
|
add r0, sp, #(\i*32)
|
2017-01-09 22:15:15 +00:00
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
.if \i > 0
|
|
|
|
|
ldrh_post r1, r8, #2
|
|
|
|
|
cmp r3, r1
|
|
|
|
|
it le
|
|
|
|
|
movle r1, #(16 - \i)/4
|
|
|
|
|
ble 1f
|
|
|
|
|
.endif
|
|
|
|
|
.endif
|
|
|
|
|
mov r1, #\i
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
add r2, r6, #(\i*2)
|
|
|
|
|
bl \txfm1\()16_1d_4x16_pass1_neon
|
|
|
|
|
.endr
|
2017-01-09 22:15:15 +00:00
|
|
|
|
|
|
|
|
.ifc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
b 3f
|
|
|
|
|
1:
|
|
|
|
|
@ For all-zero slices in pass 1, set d28-d31 to zero, for the in-register
|
|
|
|
|
@ passthrough of coefficients to pass 2 and clear the end of the temp buffer
|
|
|
|
|
vmov.i16 q14, #0
|
|
|
|
|
vmov.i16 q15, #0
|
|
|
|
|
2:
|
|
|
|
|
subs r1, r1, #1
|
|
|
|
|
.rept 4
|
|
|
|
|
vst1.16 {q14-q15}, [r0,:128]!
|
|
|
|
|
.endr
|
|
|
|
|
bne 2b
|
|
|
|
|
3:
|
|
|
|
|
.endif
|
|
|
|
|
|
2017-01-09 22:15:14 +00:00
|
|
|
.ifc \txfm1\()_\txfm2,iadst_idct
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vld1.16 {q0-q1}, [r12,:128]
|
|
|
|
|
.endif
|
|
|
|
|
.irp i, 0, 4, 8, 12
|
|
|
|
|
add r0, r4, #(\i)
|
|
|
|
|
mov r1, r5
|
|
|
|
|
add r2, sp, #(\i*2)
|
|
|
|
|
mov r3, #\i
|
|
|
|
|
bl \txfm2\()16_1d_4x16_pass2_neon
|
|
|
|
|
.endr
|
|
|
|
|
|
2017-01-09 22:15:09 +00:00
|
|
|
add sp, sp, r7
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.ifnc \txfm1\()_\txfm2,idct_idct
|
|
|
|
|
vpop {q4-q7}
|
|
|
|
|
.endif
|
2017-01-09 22:15:15 +00:00
|
|
|
pop {r4-r8,pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endfunc
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
itxfm_func16x16 idct, idct
|
|
|
|
|
itxfm_func16x16 iadst, idct
|
|
|
|
|
itxfm_func16x16 idct, iadst
|
|
|
|
|
itxfm_func16x16 iadst, iadst
|
|
|
|
|
.ltorg
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
function idct16_1d_4x16_pass1_quarter_neon
|
|
|
|
|
push {lr}
|
|
|
|
|
mov r12, #32
|
|
|
|
|
vmov.s16 q2, #0
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
|
|
|
|
vst1.16 {d4}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
|
|
|
|
|
bl idct16_quarter
|
|
|
|
|
|
|
|
|
|
@ Do four 4x4 transposes. Originally, d16-d31 contain the
|
|
|
|
|
@ 16 rows. Afterwards, d16-d19, d20-d23, d24-d27, d28-d31
|
|
|
|
|
@ contain the transposed 4x4 blocks.
|
|
|
|
|
transpose16_q_4x_4x4 q8, q9, q10, q11, q12, q13, q14, q15, d16, d17, d18, d19, d20, d21, d22, d23, d24, d25, d26, d27, d28, d29, d30, d31
|
|
|
|
|
|
|
|
|
|
@ Store the transposed 4x4 blocks horizontally.
|
|
|
|
|
@ The first 4x4 block is kept in registers for the second pass,
|
|
|
|
|
@ store the rest in the temp buffer.
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d20}, [r0,:64]!
|
|
|
|
|
vst1.16 {d24}, [r0,:64]!
|
|
|
|
|
vst1.16 {d28}, [r0,:64]!
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d21}, [r0,:64]!
|
|
|
|
|
vst1.16 {d25}, [r0,:64]!
|
|
|
|
|
vst1.16 {d29}, [r0,:64]!
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d22}, [r0,:64]!
|
|
|
|
|
vst1.16 {d26}, [r0,:64]!
|
|
|
|
|
vst1.16 {d30}, [r0,:64]!
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d23}, [r0,:64]!
|
|
|
|
|
vst1.16 {d27}, [r0,:64]!
|
|
|
|
|
vst1.16 {d31}, [r0,:64]!
|
|
|
|
|
pop {pc}
|
|
|
|
|
endfunc
|
|
|
|
|
|
|
|
|
|
function idct16_1d_4x16_pass2_quarter_neon
|
|
|
|
|
push {lr}
|
|
|
|
|
@ Only load the top 4 lines, and only do it for the later slices.
|
|
|
|
|
@ For the first slice, d16-d19 is kept in registers from the first pass.
|
|
|
|
|
cmp r3, #0
|
|
|
|
|
beq 1f
|
|
|
|
|
mov r12, #32
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
1:
|
|
|
|
|
|
|
|
|
|
add r3, r0, r1
|
|
|
|
|
lsl r1, r1, #1
|
|
|
|
|
bl idct16_quarter
|
|
|
|
|
|
|
|
|
|
load_add_store q8, q9, q10, q11
|
|
|
|
|
load_add_store q12, q13, q14, q15
|
|
|
|
|
|
|
|
|
|
pop {pc}
|
|
|
|
|
endfunc
|
|
|
|
|
|
|
|
|
|
function idct16_1d_4x16_pass1_half_neon
|
|
|
|
|
push {lr}
|
|
|
|
|
mov r12, #32
|
|
|
|
|
vmov.s16 q2, #0
|
|
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
|
|
|
|
vst1.16 {d4}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
|
|
|
|
|
bl idct16_half
|
|
|
|
|
|
|
|
|
|
@ Do four 4x4 transposes. Originally, d16-d31 contain the
|
|
|
|
|
@ 16 rows. Afterwards, d16-d19, d20-d23, d24-d27, d28-d31
|
|
|
|
|
@ contain the transposed 4x4 blocks.
|
|
|
|
|
transpose16_q_4x_4x4 q8, q9, q10, q11, q12, q13, q14, q15, d16, d17, d18, d19, d20, d21, d22, d23, d24, d25, d26, d27, d28, d29, d30, d31
|
|
|
|
|
|
|
|
|
|
@ Store the transposed 4x4 blocks horizontally.
|
|
|
|
|
cmp r1, #4
|
|
|
|
|
beq 1f
|
|
|
|
|
.irp i, 16, 20, 24, 28, 17, 21, 25, 29, 18, 22, 26, 30, 19, 23, 27, 31
|
|
|
|
|
vst1.16 {d\i}, [r0,:64]!
|
|
|
|
|
.endr
|
|
|
|
|
pop {pc}
|
|
|
|
|
1:
|
|
|
|
|
@ Special case: For the second input column (r1 == 4),
|
|
|
|
|
@ which would be stored as the second row in the temp buffer,
|
|
|
|
|
@ don't store the first 4x4 block, but keep it in registers
|
|
|
|
|
@ for the first slice of the second pass (where it is the
|
|
|
|
|
@ second 4x4 block).
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d20}, [r0,:64]!
|
|
|
|
|
vst1.16 {d24}, [r0,:64]!
|
|
|
|
|
vst1.16 {d28}, [r0,:64]!
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d21}, [r0,:64]!
|
|
|
|
|
vst1.16 {d25}, [r0,:64]!
|
|
|
|
|
vst1.16 {d29}, [r0,:64]!
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d22}, [r0,:64]!
|
|
|
|
|
vst1.16 {d26}, [r0,:64]!
|
|
|
|
|
vst1.16 {d30}, [r0,:64]!
|
|
|
|
|
add r0, r0, #8
|
|
|
|
|
vst1.16 {d23}, [r0,:64]!
|
|
|
|
|
vst1.16 {d27}, [r0,:64]!
|
|
|
|
|
vst1.16 {d31}, [r0,:64]!
|
|
|
|
|
vmov d20, d16
|
|
|
|
|
vmov d21, d17
|
|
|
|
|
vmov d22, d18
|
|
|
|
|
vmov d23, d19
|
|
|
|
|
pop {pc}
|
|
|
|
|
endfunc
|
|
|
|
|
|
|
|
|
|
function idct16_1d_4x16_pass2_half_neon
|
|
|
|
|
push {lr}
|
|
|
|
|
mov r12, #32
|
|
|
|
|
cmp r3, #0
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
beq 1f
|
|
|
|
|
.irp i, 20, 21, 22, 23
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
1:
|
|
|
|
|
|
|
|
|
|
add r3, r0, r1
|
|
|
|
|
lsl r1, r1, #1
|
|
|
|
|
bl idct16_half
|
|
|
|
|
|
|
|
|
|
load_add_store q8, q9, q10, q11
|
|
|
|
|
load_add_store q12, q13, q14, q15
|
|
|
|
|
|
|
|
|
|
pop {pc}
|
|
|
|
|
endfunc
|
|
|
|
|
.purgem load_add_store
|
|
|
|
|
|
|
|
|
|
.macro idct16_partial size
|
|
|
|
|
function idct16x16_\size\()_add_neon
|
|
|
|
|
add r0, sp, #(0*32)
|
|
|
|
|
mov r1, #0
|
|
|
|
|
add r2, r6, #(0*2)
|
|
|
|
|
bl idct16_1d_4x16_pass1_\size\()_neon
|
|
|
|
|
.ifc \size,half
|
|
|
|
|
add r0, sp, #(4*32)
|
|
|
|
|
mov r1, #4
|
|
|
|
|
add r2, r6, #(4*2)
|
|
|
|
|
bl idct16_1d_4x16_pass1_\size\()_neon
|
|
|
|
|
.endif
|
|
|
|
|
.irp i, 0, 4, 8, 12
|
|
|
|
|
add r0, r4, #(\i)
|
|
|
|
|
mov r1, r5
|
|
|
|
|
add r2, sp, #(\i*2)
|
|
|
|
|
mov r3, #\i
|
|
|
|
|
bl idct16_1d_4x16_pass2_\size\()_neon
|
|
|
|
|
.endr
|
|
|
|
|
|
|
|
|
|
add sp, sp, r7
|
|
|
|
|
pop {r4-r8,pc}
|
|
|
|
|
endfunc
|
|
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
idct16_partial quarter
|
|
|
|
|
idct16_partial half
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
function idct32x32_dc_add_neon
|
|
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vld1.16 {d0}, [r12,:64]
|
|
|
|
|
|
2017-01-08 22:04:19 +00:00
|
|
|
vmov.i16 q2, #0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
vld1.16 {d16[]}, [r2,:16]
|
|
|
|
|
vmull.s16 q8, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d16, q8, #14
|
|
|
|
|
vmull.s16 q8, d16, d0[0]
|
|
|
|
|
vrshrn.s32 d16, q8, #14
|
|
|
|
|
vdup.16 q8, d16[0]
|
|
|
|
|
vst1.16 {d4[0]}, [r2,:16]
|
|
|
|
|
|
|
|
|
|
vrshr.s16 q8, q8, #6
|
|
|
|
|
|
2017-01-04 11:08:51 +00:00
|
|
|
mov r3, r0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
mov r12, #32
|
|
|
|
|
1:
|
|
|
|
|
@ Loop to add the constant from q8 into all 32x32 outputs
|
2017-01-04 11:08:51 +00:00
|
|
|
subs r12, r12, #2
|
|
|
|
|
vld1.8 {q0-q1}, [r0,:128], r1
|
|
|
|
|
vaddw.u8 q9, q8, d0
|
|
|
|
|
vaddw.u8 q10, q8, d1
|
|
|
|
|
vld1.8 {q2-q3}, [r0,:128], r1
|
|
|
|
|
vaddw.u8 q11, q8, d2
|
|
|
|
|
vaddw.u8 q12, q8, d3
|
|
|
|
|
vaddw.u8 q13, q8, d4
|
|
|
|
|
vaddw.u8 q14, q8, d5
|
|
|
|
|
vaddw.u8 q15, q8, d6
|
|
|
|
|
vqmovun.s16 d0, q9
|
|
|
|
|
vaddw.u8 q9, q8, d7
|
|
|
|
|
vqmovun.s16 d1, q10
|
|
|
|
|
vqmovun.s16 d2, q11
|
|
|
|
|
vqmovun.s16 d3, q12
|
|
|
|
|
vqmovun.s16 d4, q13
|
|
|
|
|
vqmovun.s16 d5, q14
|
|
|
|
|
vst1.8 {q0-q1}, [r3,:128], r1
|
|
|
|
|
vqmovun.s16 d6, q15
|
|
|
|
|
vqmovun.s16 d7, q9
|
|
|
|
|
vst1.8 {q2-q3}, [r3,:128], r1
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
bne 1b
|
|
|
|
|
|
|
|
|
|
bx lr
|
|
|
|
|
endfunc
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
.macro idct32_end
|
2017-01-02 20:50:38 +00:00
|
|
|
butterfly d16, d9, d8, d9 @ d16 = t16a, d9 = t19a
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d17, d20, d23, d20 @ d17 = t17, d20 = t18
|
2017-01-02 20:50:38 +00:00
|
|
|
butterfly d18, d10, d11, d10 @ d18 = t23a, d10 = t20a
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d19, d21, d22, d21 @ d19 = t22, d21 = t21
|
2017-01-02 20:50:38 +00:00
|
|
|
butterfly d8, d28, d28, d30 @ d8 = t24a, d28 = t27a
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d23, d26, d25, d26 @ d23 = t25, d26 = t26
|
2017-01-02 20:50:38 +00:00
|
|
|
butterfly d11, d29, d29, d31 @ d11 = t31a, d29 = t28a
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d22, d27, d24, d27 @ d22 = t30, d27 = t29
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly d27, d20, d0[2], d0[3], q12, q15 @ d27 = t18a, d20 = t29a
|
|
|
|
|
mbutterfly d29, d9, d0[2], d0[3], q12, q15 @ d29 = t19, d5 = t28
|
|
|
|
|
mbutterfly d28, d10, d0[2], d0[3], q12, q15, neg=1 @ d28 = t27, d6 = t20
|
|
|
|
|
mbutterfly d26, d21, d0[2], d0[3], q12, q15, neg=1 @ d26 = t26a, d21 = t21a
|
2016-11-22 09:07:38 +00:00
|
|
|
|
2017-01-02 20:50:38 +00:00
|
|
|
butterfly d31, d24, d11, d8 @ d31 = t31, d24 = t24
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d30, d25, d22, d23 @ d30 = t30a, d25 = t25a
|
|
|
|
|
butterfly_r d23, d16, d16, d18 @ d23 = t23, d16 = t16
|
|
|
|
|
butterfly_r d22, d17, d17, d19 @ d22 = t22a, d17 = t17a
|
|
|
|
|
butterfly d18, d21, d27, d21 @ d18 = t18, d21 = t21
|
2017-01-02 20:50:38 +00:00
|
|
|
butterfly_r d27, d28, d9, d28 @ d27 = t27a, d28 = t28a
|
|
|
|
|
butterfly d8, d26, d20, d26 @ d8 = t29, d26 = t26
|
|
|
|
|
butterfly d19, d20, d29, d10 @ d19 = t19a, d20 = t20
|
|
|
|
|
vmov d29, d8 @ d29 = t29
|
|
|
|
|
|
|
|
|
|
mbutterfly0 d27, d20, d27, d20, d8, d10, q4, q5 @ d27 = t27, d20 = t20
|
|
|
|
|
mbutterfly0 d26, d21, d26, d21, d8, d10, q4, q5 @ d26 = t26a, d21 = t21a
|
|
|
|
|
mbutterfly0 d25, d22, d25, d22, d8, d10, q4, q5 @ d25 = t25, d22 = t22
|
|
|
|
|
mbutterfly0 d24, d23, d24, d23, d8, d10, q4, q5 @ d24 = t24a, d23 = t23a
|
2016-11-22 09:07:38 +00:00
|
|
|
bx lr
|
|
|
|
|
.endm
|
|
|
|
|
|
2016-11-23 08:56:12 +00:00
|
|
|
function idct32_odd
|
2017-01-02 20:50:38 +00:00
|
|
|
mbutterfly d16, d31, d4[0], d4[1], q4, q5 @ d16 = t16a, d31 = t31a
|
|
|
|
|
mbutterfly d24, d23, d4[2], d4[3], q4, q5 @ d24 = t17a, d23 = t30a
|
|
|
|
|
mbutterfly d20, d27, d5[0], d5[1], q4, q5 @ d20 = t18a, d27 = t29a
|
|
|
|
|
mbutterfly d28, d19, d5[2], d5[3], q4, q5 @ d28 = t19a, d19 = t28a
|
|
|
|
|
mbutterfly d18, d29, d6[0], d6[1], q4, q5 @ d18 = t20a, d29 = t27a
|
|
|
|
|
mbutterfly d26, d21, d6[2], d6[3], q4, q5 @ d26 = t21a, d21 = t26a
|
|
|
|
|
mbutterfly d22, d25, d7[0], d7[1], q4, q5 @ d22 = t22a, d25 = t25a
|
|
|
|
|
mbutterfly d30, d17, d7[2], d7[3], q4, q5 @ d30 = t23a, d17 = t24a
|
|
|
|
|
|
|
|
|
|
butterfly d8, d24, d16, d24 @ d8 = t16, d24 = t17
|
|
|
|
|
butterfly d9, d20, d28, d20 @ d9 = t19, d20 = t18
|
|
|
|
|
butterfly d10, d26, d18, d26 @ d10 = t20, d26 = t21
|
|
|
|
|
butterfly d11, d22, d30, d22 @ d11 = t23, d22 = t22
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
butterfly d28, d25, d17, d25 @ d28 = t24, d25 = t25
|
|
|
|
|
butterfly d30, d21, d29, d21 @ d30 = t27, d21 = t26
|
|
|
|
|
butterfly d29, d23, d31, d23 @ d29 = t31, d23 = t30
|
|
|
|
|
butterfly d31, d27, d19, d27 @ d31 = t28, d27 = t29
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly d23, d24, d1[0], d1[1], q8, q9 @ d23 = t17a, d24 = t30a
|
|
|
|
|
mbutterfly d27, d20, d1[0], d1[1], q8, q9, neg=1 @ d27 = t29a, d20 = t18a
|
|
|
|
|
mbutterfly d21, d26, d1[2], d1[3], q8, q9 @ d21 = t21a, d26 = t26a
|
|
|
|
|
mbutterfly d25, d22, d1[2], d1[3], q8, q9, neg=1 @ d25 = t25a, d22 = t22a
|
2016-11-22 09:07:38 +00:00
|
|
|
idct32_end
|
|
|
|
|
endfunc
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
function idct32_odd_half
|
2017-01-02 20:50:38 +00:00
|
|
|
mbutterfly_h1 d16, d31, d4[0], d4[1], q4, q5 @ d16 = t16a, d31 = t31a
|
|
|
|
|
mbutterfly_h2 d24, d23, d4[2], d4[3], q4, q5 @ d24 = t17a, d23 = t30a
|
|
|
|
|
mbutterfly_h1 d20, d27, d5[0], d5[1], q4, q5 @ d20 = t18a, d27 = t29a
|
|
|
|
|
mbutterfly_h2 d28, d19, d5[2], d5[3], q4, q5 @ d28 = t19a, d19 = t28a
|
|
|
|
|
mbutterfly_h1 d18, d29, d6[0], d6[1], q4, q5 @ d18 = t20a, d29 = t27a
|
|
|
|
|
mbutterfly_h2 d26, d21, d6[2], d6[3], q4, q5 @ d26 = t21a, d21 = t26a
|
|
|
|
|
mbutterfly_h1 d22, d25, d7[0], d7[1], q4, q5 @ d22 = t22a, d25 = t25a
|
|
|
|
|
mbutterfly_h2 d30, d17, d7[2], d7[3], q4, q5 @ d30 = t23a, d17 = t24a
|
|
|
|
|
|
|
|
|
|
butterfly d8, d24, d16, d24 @ d8 = t16, d24 = t17
|
|
|
|
|
butterfly d9, d20, d28, d20 @ d9 = t19, d20 = t18
|
|
|
|
|
butterfly d10, d26, d18, d26 @ d10 = t20, d26 = t21
|
|
|
|
|
butterfly d11, d22, d30, d22 @ d11 = t23, d22 = t22
|
2016-11-22 09:07:38 +00:00
|
|
|
butterfly d28, d25, d17, d25 @ d28 = t24, d25 = t25
|
|
|
|
|
butterfly d30, d21, d29, d21 @ d30 = t27, d21 = t26
|
|
|
|
|
butterfly d29, d23, d31, d23 @ d29 = t31, d23 = t30
|
|
|
|
|
butterfly d31, d27, d19, d27 @ d31 = t28, d27 = t29
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly d23, d24, d1[0], d1[1], q8, q9 @ d23 = t17a, d24 = t30a
|
|
|
|
|
mbutterfly d27, d20, d1[0], d1[1], q8, q9, neg=1 @ d27 = t29a, d20 = t18a
|
|
|
|
|
mbutterfly d21, d26, d1[2], d1[3], q8, q9 @ d21 = t21a, d26 = t26a
|
|
|
|
|
mbutterfly d25, d22, d1[2], d1[3], q8, q9, neg=1 @ d25 = t25a, d22 = t22a
|
2016-11-22 09:07:38 +00:00
|
|
|
|
|
|
|
|
idct32_end
|
2016-11-23 08:56:12 +00:00
|
|
|
endfunc
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
function idct32_odd_quarter
|
2017-01-02 20:50:38 +00:00
|
|
|
vmull.s16 q4, d16, d4[0]
|
|
|
|
|
vmull.s16 q14, d19, d5[3]
|
|
|
|
|
vmull.s16 q15, d16, d4[1]
|
|
|
|
|
vmull.s16 q11, d17, d7[2]
|
|
|
|
|
vmull.s16 q5, d17, d7[3]
|
|
|
|
|
vmull.s16 q13, d19, d5[2]
|
|
|
|
|
vmull.s16 q10, d18, d6[0]
|
|
|
|
|
vmull.s16 q12, d18, d6[1]
|
2016-11-22 09:07:38 +00:00
|
|
|
|
|
|
|
|
vneg.s32 q14, q14
|
2017-01-02 20:50:38 +00:00
|
|
|
vneg.s32 q5, q5
|
2016-11-22 09:07:38 +00:00
|
|
|
|
2017-01-02 20:50:38 +00:00
|
|
|
vrshrn.s32 d8, q4, #14
|
|
|
|
|
vrshrn.s32 d9, q14, #14
|
2016-11-22 09:07:38 +00:00
|
|
|
vrshrn.s32 d29, q15, #14
|
|
|
|
|
vrshrn.s32 d28, q11, #14
|
2017-01-02 20:50:38 +00:00
|
|
|
vrshrn.s32 d11, q5, #14
|
2016-11-22 09:07:38 +00:00
|
|
|
vrshrn.s32 d31, q13, #14
|
2017-01-02 20:50:38 +00:00
|
|
|
vrshrn.s32 d10, q10, #14
|
2016-11-22 09:07:38 +00:00
|
|
|
vrshrn.s32 d30, q12, #14
|
|
|
|
|
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q8, q9, d29, d8, d1[0], d1[1]
|
|
|
|
|
mbutterfly_l q13, q10, d31, d9, d1[0], d1[1]
|
2016-11-22 09:07:38 +00:00
|
|
|
vrshrn.s32 d23, q8, #14
|
|
|
|
|
vrshrn.s32 d24, q9, #14
|
|
|
|
|
vneg.s32 q10, q10
|
|
|
|
|
vrshrn.s32 d27, q13, #14
|
|
|
|
|
vrshrn.s32 d20, q10, #14
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q8, q9, d30, d10, d1[2], d1[3]
|
2016-11-22 09:07:38 +00:00
|
|
|
vrshrn.s32 d21, q8, #14
|
|
|
|
|
vrshrn.s32 d26, q9, #14
|
2016-12-31 12:05:44 +00:00
|
|
|
mbutterfly_l q8, q9, d28, d11, d1[2], d1[3]
|
2016-11-22 09:07:38 +00:00
|
|
|
vrshrn.s32 d25, q8, #14
|
|
|
|
|
vneg.s32 q9, q9
|
|
|
|
|
vrshrn.s32 d22, q9, #14
|
|
|
|
|
|
|
|
|
|
idct32_end
|
|
|
|
|
endfunc
|
|
|
|
|
|
|
|
|
|
.macro idct32_funcs suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ Do an 32-point IDCT of a 4x32 slice out of a 32x32 matrix.
|
|
|
|
|
@ We don't have register space to do a single pass IDCT of 4x32 though,
|
|
|
|
|
@ but the 32-point IDCT can be decomposed into two 16-point IDCTs;
|
|
|
|
|
@ a normal IDCT16 with every other input component (the even ones, with
|
|
|
|
|
@ each output written twice), followed by a separate 16-point IDCT
|
|
|
|
|
@ of the odd inputs, added/subtracted onto the outputs of the first idct16.
|
|
|
|
|
@ r0 = dst (temp buffer)
|
|
|
|
|
@ r1 = unused
|
|
|
|
|
@ r2 = src
|
2016-11-22 09:07:38 +00:00
|
|
|
function idct32_1d_4x32_pass1\suffix\()_neon
|
2016-11-23 08:56:12 +00:00
|
|
|
push {lr}
|
|
|
|
|
|
2017-01-02 20:50:38 +00:00
|
|
|
@ idct16 clobbers q2-q3 (since it doesn't clobber q4-q7 at all
|
|
|
|
|
@ when doing the normal 16x16 idct), so move the idct32_odd coeffs
|
|
|
|
|
@ to q4-q5
|
|
|
|
|
vmov q4, q2
|
|
|
|
|
vmov q5, q3
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
@ Double stride of the input, since we only read every other line
|
|
|
|
|
mov r12, #128
|
2017-01-08 22:04:19 +00:00
|
|
|
vmov.s16 d4, #0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
@ d16 = IN(0), d17 = IN(2) ... d31 = IN(30)
|
2016-11-22 09:07:38 +00:00
|
|
|
.ifb \suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
|
|
|
|
vst1.16 {d4}, [r2,:64], r12
|
|
|
|
|
.endr
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_quarter
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
|
|
|
|
vst1.16 {d4}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_half
|
|
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
|
|
|
|
vst1.16 {d4}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
.endif
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
bl idct16\suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2017-01-02 20:50:38 +00:00
|
|
|
@ Move the idct32_odd coeffs back into q2-q3 for idct32_odd;
|
|
|
|
|
@ the constants for a vmul with a lane must be in q0-q3.
|
|
|
|
|
vmov q2, q4
|
|
|
|
|
vmov q3, q5
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ Do four 4x4 transposes. Originally, d16-d31 contain the
|
|
|
|
|
@ 16 rows. Afterwards, d16-d19, d20-d23, d24-d27, d28-d31
|
|
|
|
|
@ contain the transposed 4x4 blocks.
|
|
|
|
|
transpose16_q_4x_4x4 q8, q9, q10, q11, q12, q13, q14, q15, d16, d17, d18, d19, d20, d21, d22, d23, d24, d25, d26, d27, d28, d29, d30, d31
|
|
|
|
|
@ Store the registers a, b, c, d horizontally, followed
|
|
|
|
|
@ by the same registers d, c, b, a mirrored.
|
|
|
|
|
.macro store_rev a, b, c, d
|
|
|
|
|
.irp i, \a, \b, \c, \d
|
|
|
|
|
vst1.16 {d\i}, [r0,:64]!
|
|
|
|
|
vrev64.16 d\i, d\i
|
|
|
|
|
.endr
|
|
|
|
|
.irp i, \d, \c, \b, \a
|
|
|
|
|
vst1.16 {d\i}, [r0,:64]!
|
|
|
|
|
.endr
|
|
|
|
|
.endm
|
|
|
|
|
store_rev 16, 20, 24, 28
|
|
|
|
|
store_rev 17, 21, 25, 29
|
|
|
|
|
store_rev 18, 22, 26, 30
|
|
|
|
|
store_rev 19, 23, 27, 31
|
|
|
|
|
sub r0, r0, #256
|
|
|
|
|
.purgem store_rev
|
|
|
|
|
|
|
|
|
|
@ Move r2 back to the start of the input, and move
|
|
|
|
|
@ to the first odd row
|
2016-11-22 09:07:38 +00:00
|
|
|
.ifb \suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
sub r2, r2, r12, lsl #4
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_quarter
|
|
|
|
|
sub r2, r2, r12, lsl #2
|
|
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_half
|
|
|
|
|
sub r2, r2, r12, lsl #3
|
|
|
|
|
.endif
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
add r2, r2, #64
|
|
|
|
|
|
2017-01-08 22:04:19 +00:00
|
|
|
vmov.s16 d8, #0
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
@ d16 = IN(1), d17 = IN(3) ... d31 = IN(31)
|
2016-11-22 09:07:38 +00:00
|
|
|
.ifb \suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
2017-01-02 20:50:38 +00:00
|
|
|
vst1.16 {d8}, [r2,:64], r12
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.endr
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_quarter
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
2017-01-02 20:50:38 +00:00
|
|
|
vst1.16 {d8}, [r2,:64], r12
|
2016-11-22 09:07:38 +00:00
|
|
|
.endr
|
|
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_half
|
|
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23
|
|
|
|
|
vld1.16 {d\i}, [r2,:64]
|
2017-01-02 20:50:38 +00:00
|
|
|
vst1.16 {d8}, [r2,:64], r12
|
2016-11-22 09:07:38 +00:00
|
|
|
.endr
|
|
|
|
|
.endif
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
bl idct32_odd\suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
transpose16_q_4x_4x4 q15, q14, q13, q12, q11, q10, q9, q8, d31, d30, d29, d28, d27, d26, d25, d24, d23, d22, d21, d20, d19, d18, d17, d16
|
|
|
|
|
|
|
|
|
|
@ Store the registers a, b, c, d horizontally,
|
|
|
|
|
@ adding into the output first, and then mirrored, subtracted
|
|
|
|
|
@ from the output.
|
|
|
|
|
.macro store_rev a, b, c, d
|
|
|
|
|
.irp i, \a, \b, \c, \d
|
2017-01-02 20:50:38 +00:00
|
|
|
vld1.16 {d8}, [r0,:64]
|
|
|
|
|
vadd.s16 d8, d8, d\i
|
|
|
|
|
vst1.16 {d8}, [r0,:64]!
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
vrev64.16 d\i, d\i
|
|
|
|
|
.endr
|
|
|
|
|
.irp i, \d, \c, \b, \a
|
2017-01-02 20:50:38 +00:00
|
|
|
vld1.16 {d8}, [r0,:64]
|
|
|
|
|
vsub.s16 d8, d8, d\i
|
|
|
|
|
vst1.16 {d8}, [r0,:64]!
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.endr
|
|
|
|
|
.endm
|
|
|
|
|
|
2017-01-09 22:15:11 +00:00
|
|
|
store_rev 31, 27, 23, 19
|
|
|
|
|
store_rev 30, 26, 22, 18
|
|
|
|
|
store_rev 29, 25, 21, 17
|
|
|
|
|
store_rev 28, 24, 20, 16
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.purgem store_rev
|
2016-11-23 08:56:12 +00:00
|
|
|
pop {pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endfunc
|
|
|
|
|
.ltorg
|
|
|
|
|
|
|
|
|
|
@ This is mostly the same as 4x32_pass1, but without the transpose,
|
|
|
|
|
@ and use the source as temp buffer between the two idct passes, and
|
|
|
|
|
@ add into the destination.
|
|
|
|
|
@ r0 = dst
|
|
|
|
|
@ r1 = dst stride
|
|
|
|
|
@ r2 = src (temp buffer)
|
2016-11-22 09:07:38 +00:00
|
|
|
function idct32_1d_4x32_pass2\suffix\()_neon
|
2016-11-23 08:56:12 +00:00
|
|
|
push {lr}
|
2017-01-02 20:50:38 +00:00
|
|
|
vmov q4, q2
|
|
|
|
|
vmov q5, q3
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
mov r12, #128
|
|
|
|
|
@ d16 = IN(0), d17 = IN(2) ... d31 = IN(30)
|
2016-11-22 09:07:38 +00:00
|
|
|
.ifb \suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
sub r2, r2, r12, lsl #4
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_quarter
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
sub r2, r2, r12, lsl #2
|
|
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_half
|
|
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
sub r2, r2, r12, lsl #3
|
|
|
|
|
.endif
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
bl idct16\suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2017-01-02 20:50:38 +00:00
|
|
|
vmov q2, q4
|
|
|
|
|
vmov q3, q5
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31
|
|
|
|
|
vst1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
|
|
|
|
|
sub r2, r2, r12, lsl #4
|
|
|
|
|
add r2, r2, #64
|
|
|
|
|
|
|
|
|
|
@ d16 = IN(1), d17 = IN(3) ... d31 = IN(31)
|
2016-11-22 09:07:38 +00:00
|
|
|
.ifb \suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
sub r2, r2, r12, lsl #4
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_quarter
|
|
|
|
|
.irp i, 16, 17, 18, 19
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
sub r2, r2, r12, lsl #2
|
|
|
|
|
.endif
|
|
|
|
|
.ifc \suffix,_half
|
|
|
|
|
.irp i, 16, 17, 18, 19, 20, 21, 22, 23
|
|
|
|
|
vld1.16 {d\i}, [r2,:64], r12
|
|
|
|
|
.endr
|
|
|
|
|
sub r2, r2, r12, lsl #3
|
|
|
|
|
.endif
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
sub r2, r2, #64
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
bl idct32_odd\suffix
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
mov r12, #128
|
|
|
|
|
.macro load_acc_store a, b, c, d, neg=0
|
2017-01-02 20:50:38 +00:00
|
|
|
vld1.16 {d8}, [r2,:64], r12
|
|
|
|
|
vld1.16 {d9}, [r2,:64], r12
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.if \neg == 0
|
2017-01-02 20:50:38 +00:00
|
|
|
vadd.s16 d8, d8, d\a
|
|
|
|
|
vld1.16 {d10}, [r2,:64], r12
|
|
|
|
|
vadd.s16 d9, d9, d\b
|
|
|
|
|
vld1.16 {d11}, [r2,:64], r12
|
|
|
|
|
vadd.s16 d10, d10, d\c
|
|
|
|
|
vadd.s16 d11, d11, d\d
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.else
|
2017-01-02 20:50:38 +00:00
|
|
|
vsub.s16 d8, d8, d\a
|
|
|
|
|
vld1.16 {d10}, [r2,:64], r12
|
|
|
|
|
vsub.s16 d9, d9, d\b
|
|
|
|
|
vld1.16 {d11}, [r2,:64], r12
|
|
|
|
|
vsub.s16 d10, d10, d\c
|
|
|
|
|
vsub.s16 d11, d11, d\d
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.endif
|
2017-01-02 20:50:38 +00:00
|
|
|
vld1.32 {d12[]}, [r0,:32], r1
|
|
|
|
|
vld1.32 {d12[1]}, [r0,:32], r1
|
2017-01-08 22:04:19 +00:00
|
|
|
vrshr.s16 q4, q4, #6
|
2017-01-02 20:50:38 +00:00
|
|
|
vld1.32 {d13[]}, [r0,:32], r1
|
2017-01-08 22:04:19 +00:00
|
|
|
vrshr.s16 q5, q5, #6
|
2017-01-02 20:50:38 +00:00
|
|
|
vld1.32 {d13[1]}, [r0,:32], r1
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
sub r0, r0, r1, lsl #2
|
2017-01-02 20:50:38 +00:00
|
|
|
vaddw.u8 q4, q4, d12
|
|
|
|
|
vaddw.u8 q5, q5, d13
|
|
|
|
|
vqmovun.s16 d8, q4
|
|
|
|
|
vqmovun.s16 d9, q5
|
|
|
|
|
vst1.32 {d8[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d8[1]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d9[0]}, [r0,:32], r1
|
|
|
|
|
vst1.32 {d9[1]}, [r0,:32], r1
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.endm
|
|
|
|
|
load_acc_store 31, 30, 29, 28
|
|
|
|
|
load_acc_store 27, 26, 25, 24
|
|
|
|
|
load_acc_store 23, 22, 21, 20
|
|
|
|
|
load_acc_store 19, 18, 17, 16
|
|
|
|
|
sub r2, r2, r12
|
|
|
|
|
neg r12, r12
|
|
|
|
|
load_acc_store 16, 17, 18, 19, 1
|
|
|
|
|
load_acc_store 20, 21, 22, 23, 1
|
|
|
|
|
load_acc_store 24, 25, 26, 27, 1
|
|
|
|
|
load_acc_store 28, 29, 30, 31, 1
|
|
|
|
|
.purgem load_acc_store
|
2016-11-23 08:56:12 +00:00
|
|
|
pop {pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endfunc
|
2016-11-22 09:07:38 +00:00
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
idct32_funcs
|
|
|
|
|
idct32_funcs _quarter
|
|
|
|
|
idct32_funcs _half
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
2017-01-09 22:15:15 +00:00
|
|
|
const min_eob_idct_idct_32, align=4
|
|
|
|
|
.short 0, 9, 34, 70, 135, 240, 336, 448
|
|
|
|
|
endconst
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
function ff_vp9_idct_idct_32x32_add_neon, export=1
|
|
|
|
|
cmp r3, #1
|
|
|
|
|
beq idct32x32_dc_add_neon
|
2017-01-09 22:15:15 +00:00
|
|
|
push {r4-r8,lr}
|
2017-01-02 20:50:38 +00:00
|
|
|
vpush {q4-q6}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
@ Align the stack, allocate a temp buffer
|
2017-01-09 22:15:09 +00:00
|
|
|
T mov r7, sp
|
|
|
|
|
T and r7, r7, #15
|
|
|
|
|
A and r7, sp, #15
|
|
|
|
|
add r7, r7, #2048
|
|
|
|
|
sub sp, sp, r7
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
|
|
|
|
|
mov r4, r0
|
|
|
|
|
mov r5, r1
|
|
|
|
|
mov r6, r2
|
|
|
|
|
|
2017-01-02 20:50:38 +00:00
|
|
|
movrel r12, idct_coeffs
|
|
|
|
|
vld1.16 {q0-q1}, [r12,:128]!
|
|
|
|
|
vld1.16 {q2-q3}, [r12,:128]
|
|
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
cmp r3, #34
|
|
|
|
|
ble idct32x32_quarter_add_neon
|
|
|
|
|
cmp r3, #135
|
|
|
|
|
ble idct32x32_half_add_neon
|
|
|
|
|
|
2017-02-26 12:02:35 +00:00
|
|
|
movrel r8, min_eob_idct_idct_32 + 2
|
|
|
|
|
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 0, 4, 8, 12, 16, 20, 24, 28
|
|
|
|
|
add r0, sp, #(\i*64)
|
2017-01-09 22:15:15 +00:00
|
|
|
.if \i > 0
|
|
|
|
|
ldrh_post r1, r8, #2
|
|
|
|
|
cmp r3, r1
|
|
|
|
|
it le
|
|
|
|
|
movle r1, #(32 - \i)/2
|
|
|
|
|
ble 1f
|
|
|
|
|
.endif
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
add r2, r6, #(\i*2)
|
|
|
|
|
bl idct32_1d_4x32_pass1_neon
|
|
|
|
|
.endr
|
2017-01-09 22:15:15 +00:00
|
|
|
b 3f
|
|
|
|
|
|
|
|
|
|
1:
|
|
|
|
|
@ Write zeros to the temp buffer for pass 2
|
|
|
|
|
vmov.i16 q14, #0
|
|
|
|
|
vmov.i16 q15, #0
|
|
|
|
|
2:
|
|
|
|
|
subs r1, r1, #1
|
|
|
|
|
.rept 4
|
|
|
|
|
vst1.16 {q14-q15}, [r0,:128]!
|
|
|
|
|
.endr
|
|
|
|
|
bne 2b
|
|
|
|
|
3:
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
.irp i, 0, 4, 8, 12, 16, 20, 24, 28
|
|
|
|
|
add r0, r4, #(\i)
|
|
|
|
|
mov r1, r5
|
|
|
|
|
add r2, sp, #(\i*2)
|
|
|
|
|
bl idct32_1d_4x32_pass2_neon
|
|
|
|
|
.endr
|
|
|
|
|
|
2017-01-09 22:15:09 +00:00
|
|
|
add sp, sp, r7
|
2017-01-02 20:50:38 +00:00
|
|
|
vpop {q4-q6}
|
2017-01-09 22:15:15 +00:00
|
|
|
pop {r4-r8,pc}
|
arm: vp9: Add NEON itxfm routines
This work is sponsored by, and copyright, Google.
For the transforms up to 8x8, we can fit all the data (including
temporaries) in registers and just do a straightforward transform
of all the data. For 16x16, we do a transform of 4x16 pixels in
4 slices, using a temporary buffer. For 32x32, we transform 4x32
pixels at a time, in two steps of 4x16 pixels each.
Examples of relative speedup compared to the C version, from checkasm:
Cortex A7 A8 A9 A53
vp9_inv_adst_adst_4x4_add_neon: 3.39 5.83 4.17 4.01
vp9_inv_adst_adst_8x8_add_neon: 3.79 4.86 4.23 3.98
vp9_inv_adst_adst_16x16_add_neon: 3.33 4.36 4.11 4.16
vp9_inv_dct_dct_4x4_add_neon: 4.06 6.16 4.59 4.46
vp9_inv_dct_dct_8x8_add_neon: 4.61 6.01 4.98 4.86
vp9_inv_dct_dct_16x16_add_neon: 3.35 3.44 3.36 3.79
vp9_inv_dct_dct_32x32_add_neon: 3.89 3.50 3.79 4.42
vp9_inv_wht_wht_4x4_add_neon: 3.22 5.13 3.53 3.77
Thus, the speedup vs C code is around 3-6x.
This is mostly marginally faster than the corresponding routines
in libvpx on most cores, tested with their 32x32 idct (compared to
vpx_idct32x32_1024_add_neon). These numbers are slightly in libvpx's
favour since their version doesn't clear the input buffer like ours
do (although the effect of that on the total runtime probably is
negligible.)
Cortex A7 A8 A9 A53
vp9_inv_dct_dct_32x32_add_neon: 18436.8 16874.1 14235.1 11988.9
libvpx vpx_idct32x32_1024_add_neon 20789.0 13344.3 15049.9 13030.5
Only on the Cortex A8, the libvpx function is faster. On the other cores,
ours is slightly faster even though ours has got source block clearing
integrated.
This is an adapted cherry-pick from libav commits
a67ae67083151f2f9595a1f2d17b601da19b939e and
52d196fb30fb6628921b5f1b31e7bd11eb7e1d9a.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2016-11-14 10:32:22 +00:00
|
|
|
endfunc
|
2016-11-22 09:07:38 +00:00
|
|
|
|
2017-02-26 12:02:35 +00:00
|
|
|
.macro idct32_partial size
|
|
|
|
|
function idct32x32_\size\()_add_neon
|
2016-11-22 09:07:38 +00:00
|
|
|
.irp i, 0, 4
|
|
|
|
|
add r0, sp, #(\i*64)
|
2017-02-26 12:02:35 +00:00
|
|
|
.ifc \size,quarter
|
2016-11-22 09:07:38 +00:00
|
|
|
.if \i == 4
|
|
|
|
|
cmp r3, #9
|
|
|
|
|
ble 1f
|
2017-02-26 12:02:35 +00:00
|
|
|
.endif
|
2016-11-22 09:07:38 +00:00
|
|
|
.endif
|
|
|
|
|
add r2, r6, #(\i*2)
|
2017-02-26 12:02:35 +00:00
|
|
|
bl idct32_1d_4x32_pass1_\size\()_neon
|
2016-11-22 09:07:38 +00:00
|
|
|
.endr
|
|
|
|
|
|
2017-02-26 12:02:35 +00:00
|
|
|
.ifc \size,half
|
|
|
|
|
.irp i, 8, 12
|
2016-11-22 09:07:38 +00:00
|
|
|
add r0, sp, #(\i*64)
|
2017-02-26 12:02:35 +00:00
|
|
|
.if \i == 12
|
|
|
|
|
cmp r3, #70
|
2016-11-22 09:07:38 +00:00
|
|
|
ble 1f
|
|
|
|
|
.endif
|
|
|
|
|
add r2, r6, #(\i*2)
|
2017-02-26 12:02:35 +00:00
|
|
|
bl idct32_1d_4x32_pass1_\size\()_neon
|
2016-11-22 09:07:38 +00:00
|
|
|
.endr
|
2017-02-26 12:02:35 +00:00
|
|
|
.endif
|
2016-11-22 09:07:38 +00:00
|
|
|
b 3f
|
|
|
|
|
|
|
|
|
|
1:
|
|
|
|
|
@ Write zeros to the temp buffer for pass 2
|
|
|
|
|
vmov.i16 q14, #0
|
|
|
|
|
vmov.i16 q15, #0
|
2017-02-26 12:02:35 +00:00
|
|
|
.rept 8
|
2016-11-22 09:07:38 +00:00
|
|
|
vst1.16 {q14-q15}, [r0,:128]!
|
|
|
|
|
.endr
|
2017-02-26 12:02:35 +00:00
|
|
|
|
2016-11-22 09:07:38 +00:00
|
|
|
3:
|
|
|
|
|
.irp i, 0, 4, 8, 12, 16, 20, 24, 28
|
|
|
|
|
add r0, r4, #(\i)
|
|
|
|
|
mov r1, r5
|
|
|
|
|
add r2, sp, #(\i*2)
|
2017-02-26 12:02:35 +00:00
|
|
|
bl idct32_1d_4x32_pass2_\size\()_neon
|
2016-11-22 09:07:38 +00:00
|
|
|
.endr
|
|
|
|
|
|
|
|
|
|
add sp, sp, r7
|
2017-01-02 20:50:38 +00:00
|
|
|
vpop {q4-q6}
|
2016-11-22 09:07:38 +00:00
|
|
|
pop {r4-r8,pc}
|
|
|
|
|
endfunc
|
2017-02-26 12:02:35 +00:00
|
|
|
.endm
|
|
|
|
|
|
|
|
|
|
idct32_partial quarter
|
|
|
|
|
idct32_partial half
|
