Google Git
Sign in
android / platform / external / libaom / 1e651b8a0 / . / av1 / common / arm / convolve_neon.c
blob: a46cb16dd8a6ce83f29b7eadf9b070c67a7c5cb8 [file] [log] [blame]
/*
*
* Copyright (c) 2018, Alliance for Open Media. All rights reserved
*
* This source code is subject to the terms of the BSD 2 Clause License and
* the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
* was not distributed with this source code in the LICENSE file, you can
* obtain it at www.aomedia.org/license/software. If the Alliance for Open
* Media Patent License 1.0 was not distributed with this source code in the
* PATENTS file, you can obtain it at www.aomedia.org/license/patent.
*/
#include <assert.h>
#include <arm_neon.h>
#include "config/aom_config.h"
#include "config/av1_rtcd.h"
#include "aom_dsp/aom_dsp_common.h"
#include "aom_dsp/arm/mem_neon.h"
#include "aom_dsp/arm/transpose_neon.h"
#include "aom_ports/mem.h"
#include "av1/common/convolve.h"
#include "av1/common/filter.h"
#include "av1/common/arm/convolve_neon.h"
#if AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_MATMUL_INT8)
static INLINE int32x4_t convolve12_4_usdot(uint8x16_t samples,
const int8x16_t filters,
const uint8x16x3_t permute_tbl,
const int32x4_t horiz_const) {
uint8x16_t permuted_samples[3];
int32x4_t sum;
/* Permute samples ready for dot product. */
/* { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 } */
permuted_samples[0] = vqtbl1q_u8(samples, permute_tbl.val[0]);
/* { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 } */
permuted_samples[1] = vqtbl1q_u8(samples, permute_tbl.val[1]);
/* { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 } */
permuted_samples[2] = vqtbl1q_u8(samples, permute_tbl.val[2]);
/* First 4 output values. */
sum = vusdotq_laneq_s32(horiz_const, permuted_samples[0], filters, 0);
sum = vusdotq_laneq_s32(sum, permuted_samples[1], filters, 1);
sum = vusdotq_laneq_s32(sum, permuted_samples[2], filters, 2);
return sum;
}
static INLINE int16x8_t convolve12_8_usdot(uint8x16_t samples0,
uint8x16_t samples1,
const int8x16_t filters,
const uint8x16x3_t permute_tbl,
const int32x4_t horiz_const) {
uint8x16_t permuted_samples[4];
int32x4_t sum[2];
/* Permute samples ready for dot product. */
/* { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 } */
permuted_samples[0] = vqtbl1q_u8(samples0, permute_tbl.val[0]);
/* { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 } */
permuted_samples[1] = vqtbl1q_u8(samples0, permute_tbl.val[1]);
/* { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 } */
permuted_samples[2] = vqtbl1q_u8(samples0, permute_tbl.val[2]);
/* {12, 13, 14, 15, 13, 14, 15, 16, 14, 15, 16, 17, 15, 16, 17, 18 } */
permuted_samples[3] = vqtbl1q_u8(samples1, permute_tbl.val[2]);
/* First 4 output values. */
sum[0] = vusdotq_laneq_s32(horiz_const, permuted_samples[0], filters, 0);
sum[0] = vusdotq_laneq_s32(sum[0], permuted_samples[1], filters, 1);
sum[0] = vusdotq_laneq_s32(sum[0], permuted_samples[2], filters, 2);
/* Second 4 output values. */
sum[1] = vusdotq_laneq_s32(horiz_const, permuted_samples[1], filters, 0);
sum[1] = vusdotq_laneq_s32(sum[1], permuted_samples[2], filters, 1);
sum[1] = vusdotq_laneq_s32(sum[1], permuted_samples[3], filters, 2);
/* Narrow and re-pack. */
return vcombine_s16(vqrshrn_n_s32(sum[0], FILTER_BITS),
vqrshrn_n_s32(sum[1], FILTER_BITS));
}
void convolve_x_sr_12tap_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const int16_t *x_filter_ptr) {
const int16x8_t filter_0_7 = vld1q_s16(x_filter_ptr);
const int16x4_t filter_8_11 = vld1_s16(x_filter_ptr + 8);
const int16x8_t filter_8_15 = vcombine_s16(filter_8_11, vdup_n_s16(0));
const int8x16_t filter =
vcombine_s8(vmovn_s16(filter_0_7), vmovn_s16(filter_8_15));
// Special case the following no-op filter as 128 won't fit into the
// 8-bit signed dot-product instruction:
// { 0, 0, 0, 0, 0, 128, 0, 0, 0, 0, 0, 0 }
if (vgetq_lane_s16(filter_0_7, 5) == 128) {
uint8x8_t d0;
// Undo the horizontal offset in the calling function.
src += 5;
for (int i = 0; i < h; i++) {
for (int j = 0; j < w; j += 8) {
d0 = vld1_u8(src + i * src_stride + j);
if (w == 2) {
store_u8_2x1(dst + i * dst_stride, d0, 0);
} else if (w == 4) {
store_u8_4x1(dst + i * dst_stride, d0, 0);
} else {
vst1_u8(dst + i * dst_stride + j, d0);
}
}
}
} else {
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
// This shim of 1 << (ROUND0_BITS - 1) enables us to use a single rounding
// right shift by FILTER_BITS - instead of a first rounding right shift by
// ROUND0_BITS, followed by second rounding right shift by FILTER_BITS -
// ROUND0_BITS.
const int32x4_t horiz_const = vdupq_n_s32(1 << (ROUND0_BITS - 1));
if (w <= 4) {
uint8x16_t s0, s1, s2, s3;
int32x4_t d0, d1, d2, d3;
int16x8_t t01, t23;
uint8x8_t d01, d23;
do {
load_u8_16x4(src, src_stride, &s0, &s1, &s2, &s3);
d0 = convolve12_4_usdot(s0, filter, permute_tbl, horiz_const);
d1 = convolve12_4_usdot(s1, filter, permute_tbl, horiz_const);
d2 = convolve12_4_usdot(s2, filter, permute_tbl, horiz_const);
d3 = convolve12_4_usdot(s3, filter, permute_tbl, horiz_const);
t01 = vcombine_s16(vqrshrn_n_s32(d0, FILTER_BITS),
vqrshrn_n_s32(d1, FILTER_BITS));
t23 = vcombine_s16(vqrshrn_n_s32(d2, FILTER_BITS),
vqrshrn_n_s32(d3, FILTER_BITS));
d01 = vqmovun_s16(t01);
d23 = vqmovun_s16(t23);
if (w == 2) {
store_u8_2x1(dst + 0 * dst_stride, d01, 0);
store_u8_2x1(dst + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst + 2 * dst_stride, d23, 0);
store_u8_2x1(dst + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst + 0 * dst_stride, d01, 0);
store_u8_4x1(dst + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst + 2 * dst_stride, d23, 0);
store_u8_4x1(dst + 3 * dst_stride, d23, 1);
}
}
dst += 4 * dst_stride;
src += 4 * src_stride;
h -= 4;
} while (h > 0);
} else {
uint8x16_t s0, s1, s2, s3, s4, s5, s6, s7;
int16x8_t d0, d1, d2, d3;
uint8x8_t dd0, dd1, dd2, dd3;
do {
const uint8_t *s = src;
uint8_t *d = dst;
int width = w;
do {
load_u8_16x4(s, src_stride, &s0, &s1, &s2, &s3);
load_u8_16x4(s + 4, src_stride, &s4, &s5, &s6, &s7);
d0 = convolve12_8_usdot(s0, s4, filter, permute_tbl, horiz_const);
d1 = convolve12_8_usdot(s1, s5, filter, permute_tbl, horiz_const);
d2 = convolve12_8_usdot(s2, s6, filter, permute_tbl, horiz_const);
d3 = convolve12_8_usdot(s3, s7, filter, permute_tbl, horiz_const);
dd0 = vqmovun_s16(d0);
dd1 = vqmovun_s16(d1);
dd2 = vqmovun_s16(d2);
dd3 = vqmovun_s16(d3);
store_u8_8x2(d + 0 * dst_stride, dst_stride, dd0, dd1);
if (h != 2) {
store_u8_8x2(d + 2 * dst_stride, dst_stride, dd2, dd3);
}
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src += 4 * src_stride;
dst += 4 * dst_stride;
h -= 4;
} while (h > 0);
}
}
}
void av1_convolve_x_sr_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const InterpFilterParams *filter_params_x,
const int subpel_x_qn,
ConvolveParams *conv_params) {
(void)conv_params;
const uint8_t horiz_offset = filter_params_x->taps / 2 - 1;
const int16_t *x_filter_ptr = av1_get_interp_filter_subpel_kernel(
filter_params_x, subpel_x_qn & SUBPEL_MASK);
src -= horiz_offset;
if (filter_params_x->taps > 8) {
convolve_x_sr_12tap_neon(src, src_stride, dst, dst_stride, w, h,
x_filter_ptr);
return;
}
// Filter values are even, so downshift by 1 to reduce intermediate precision
// requirements.
const int8x8_t x_filter = vshrn_n_s16(vld1q_s16(x_filter_ptr), 1);
// This shim of 1 << ((ROUND0_BITS - 1) - 1) enables us to use a single
// rounding right shift by FILTER_BITS - instead of a first rounding right
// shift by ROUND0_BITS, followed by second rounding right shift by
// FILTER_BITS - ROUND0_BITS.
// The outermost -1 is needed because we halved the filter values.
const int32x4_t horiz_const = vdupq_n_s32(1 << ((ROUND0_BITS - 1) - 1));
if (w <= 4) {
const uint8x16x2_t permute_tbl = vld1q_u8_x2(dot_prod_permute_tbl);
uint8x16_t s0, s1, s2, s3;
int32x4_t t0, t1, t2, t3;
int16x8_t t01, t23;
uint8x8_t d01, d23;
do {
load_u8_16x4(src, src_stride, &s0, &s1, &s2, &s3);
t0 = convolve8_4_usdot(s0, x_filter, permute_tbl, horiz_const);
t1 = convolve8_4_usdot(s1, x_filter, permute_tbl, horiz_const);
t2 = convolve8_4_usdot(s2, x_filter, permute_tbl, horiz_const);
t3 = convolve8_4_usdot(s3, x_filter, permute_tbl, horiz_const);
t01 = vcombine_s16(vmovn_s32(t0), vmovn_s32(t1));
t23 = vcombine_s16(vmovn_s32(t2), vmovn_s32(t3));
// We halved the convolution filter values so - 1 from the right shift.
d01 = vqrshrun_n_s16(t01, FILTER_BITS - 1);
d23 = vqrshrun_n_s16(t23, FILTER_BITS - 1);
if (w == 2) {
store_u8_2x1(dst + 0 * dst_stride, d01, 0);
store_u8_2x1(dst + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst + 2 * dst_stride, d23, 0);
store_u8_2x1(dst + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst + 0 * dst_stride, d01, 0);
store_u8_4x1(dst + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst + 2 * dst_stride, d23, 0);
store_u8_4x1(dst + 3 * dst_stride, d23, 1);
}
}
h -= 4;
src += 4 * src_stride;
dst += 4 * dst_stride;
} while (h > 0);
} else {
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
uint8x16_t s0, s1, s2, s3;
int16x8_t t0, t1, t2, t3;
uint8x8_t d0, d1, d2, d3;
do {
int width = w;
const uint8_t *s = src;
uint8_t *d = dst;
do {
load_u8_16x4(s, src_stride, &s0, &s1, &s2, &s3);
t0 = convolve8_x_8_usdot(s0, x_filter, permute_tbl, horiz_const);
t1 = convolve8_x_8_usdot(s1, x_filter, permute_tbl, horiz_const);
t2 = convolve8_x_8_usdot(s2, x_filter, permute_tbl, horiz_const);
t3 = convolve8_x_8_usdot(s3, x_filter, permute_tbl, horiz_const);
// We halved the convolution filter values so - 1 from the right shift.
d0 = vqrshrun_n_s16(t0, FILTER_BITS - 1);
d1 = vqrshrun_n_s16(t1, FILTER_BITS - 1);
d2 = vqrshrun_n_s16(t2, FILTER_BITS - 1);
d3 = vqrshrun_n_s16(t3, FILTER_BITS - 1);
vst1_u8(d + 0 * dst_stride, d0);
vst1_u8(d + 1 * dst_stride, d1);
if (h != 2) {
vst1_u8(d + 2 * dst_stride, d2);
vst1_u8(d + 3 * dst_stride, d3);
}
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src += 4 * src_stride;
dst += 4 * dst_stride;
h -= 4;
} while (h > 0);
}
}
#elif AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD)
static INLINE int16x4_t convolve12_4_x(uint8x16_t samples,
const int8x16_t filter,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x3_t permute_tbl) {
int8x16_t clamped_samples, permuted_samples[3];
int32x4_t sum;
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples = vreinterpretq_s8_u8(vsubq_u8(samples, range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_s8(clamped_samples, permute_tbl.val[2]);
// Accumulate dot product into 'correction' to account for range clamp.
// First 4 output values.
sum = vdotq_laneq_s32(correction, permuted_samples[0], filter, 0);
sum = vdotq_laneq_s32(sum, permuted_samples[1], filter, 1);
sum = vdotq_laneq_s32(sum, permuted_samples[2], filter, 2);
return vqrshrn_n_s32(sum, FILTER_BITS);
}
static INLINE uint8x8_t convolve12_8_x(uint8x16_t samples[2],
const int8x16_t filter,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x3_t permute_tbl) {
int8x16_t clamped_samples[2], permuted_samples[4];
int32x4_t sum[2];
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples[0] = vreinterpretq_s8_u8(vsubq_u8(samples[0], range_limit));
clamped_samples[1] = vreinterpretq_s8_u8(vsubq_u8(samples[1], range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples[0], permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples[0], permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_s8(clamped_samples[0], permute_tbl.val[2]);
// {12, 13, 14, 15, 13, 14, 15, 16, 14, 15, 16, 17, 15, 16, 17, 18 }
permuted_samples[3] = vqtbl1q_s8(clamped_samples[1], permute_tbl.val[2]);
// Accumulate dot product into 'correction' to account for range clamp.
// First 4 output values.
sum[0] = vdotq_laneq_s32(correction, permuted_samples[0], filter, 0);
sum[0] = vdotq_laneq_s32(sum[0], permuted_samples[1], filter, 1);
sum[0] = vdotq_laneq_s32(sum[0], permuted_samples[2], filter, 2);
// Second 4 output values.
sum[1] = vdotq_laneq_s32(correction, permuted_samples[1], filter, 0);
sum[1] = vdotq_laneq_s32(sum[1], permuted_samples[2], filter, 1);
sum[1] = vdotq_laneq_s32(sum[1], permuted_samples[3], filter, 2);
// Narrow and re-pack.
int16x8_t sum_s16 = vcombine_s16(vqrshrn_n_s32(sum[0], FILTER_BITS),
vqrshrn_n_s32(sum[1], FILTER_BITS));
return vqmovun_s16(sum_s16);
}
void convolve_x_sr_12tap_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const int16_t *x_filter_ptr) {
const int16x8_t filter_0_7 = vld1q_s16(x_filter_ptr);
const int16x4_t filter_8_11 = vld1_s16(x_filter_ptr + 8);
const int16x8_t filter_8_15 = vcombine_s16(filter_8_11, vdup_n_s16(0));
const int8x16_t filter =
vcombine_s8(vmovn_s16(filter_0_7), vmovn_s16(filter_8_15));
const int32_t correction_s32 =
vaddvq_s32(vaddq_s32(vpaddlq_s16(vshlq_n_s16(filter_0_7, FILTER_BITS)),
vpaddlq_s16(vshlq_n_s16(filter_8_15, FILTER_BITS))));
// A shim of 1 << (ROUND0_BITS - 1) enables us to use a single rounding right
// shift by FILTER_BITS - instead of a first rounding right shift by
// ROUND0_BITS, followed by second rounding right shift by FILTER_BITS -
// ROUND0_BITS.
int32x4_t correction = vdupq_n_s32(correction_s32 + (1 << (ROUND0_BITS - 1)));
const uint8x16_t range_limit = vdupq_n_u8(128);
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
// Special case the following no-op filter as 128 won't fit into the
// 8-bit signed dot-product instruction:
// { 0, 0, 0, 0, 0, 128, 0, 0, 0, 0, 0, 0 }
if (vgetq_lane_s16(filter_0_7, 5) == 128) {
// Undo the horizontal offset in the calling function.
src += 5;
do {
const uint8_t *s = src;
uint8_t *d = dst;
int width = w;
do {
uint8x8_t d0 = vld1_u8(s);
if (w == 2) {
store_u8_2x1(d, d0, 0);
} else if (w == 4) {
store_u8_4x1(d, d0, 0);
} else {
vst1_u8(d, d0);
}
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src += src_stride;
dst += dst_stride;
} while (--h != 0);
} else {
if (w <= 4) {
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(src, src_stride, &s0, &s1, &s2, &s3);
int16x4_t d0 =
convolve12_4_x(s0, filter, correction, range_limit, permute_tbl);
int16x4_t d1 =
convolve12_4_x(s1, filter, correction, range_limit, permute_tbl);
int16x4_t d2 =
convolve12_4_x(s2, filter, correction, range_limit, permute_tbl);
int16x4_t d3 =
convolve12_4_x(s3, filter, correction, range_limit, permute_tbl);
uint8x8_t d01 = vqmovun_s16(vcombine_s16(d0, d1));
uint8x8_t d23 = vqmovun_s16(vcombine_s16(d2, d3));
if (w == 2) {
store_u8_2x1(dst + 0 * dst_stride, d01, 0);
store_u8_2x1(dst + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst + 2 * dst_stride, d23, 0);
store_u8_2x1(dst + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst + 0 * dst_stride, d01, 0);
store_u8_4x1(dst + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst + 2 * dst_stride, d23, 0);
store_u8_4x1(dst + 3 * dst_stride, d23, 1);
}
}
dst += 4 * dst_stride;
src += 4 * src_stride;
h -= 4;
} while (h > 0);
} else {
do {
const uint8_t *s = src;
uint8_t *d = dst;
int width = w;
do {
uint8x16_t s0[2], s1[2], s2[2], s3[2];
load_u8_16x4(s, src_stride, &s0[0], &s1[0], &s2[0], &s3[0]);
load_u8_16x4(s + 4, src_stride, &s0[1], &s1[1], &s2[1], &s3[1]);
uint8x8_t d0 =
convolve12_8_x(s0, filter, correction, range_limit, permute_tbl);
uint8x8_t d1 =
convolve12_8_x(s1, filter, correction, range_limit, permute_tbl);
uint8x8_t d2 =
convolve12_8_x(s2, filter, correction, range_limit, permute_tbl);
uint8x8_t d3 =
convolve12_8_x(s3, filter, correction, range_limit, permute_tbl);
store_u8_8x2(d + 0 * dst_stride, dst_stride, d0, d1);
if (h != 2) {
store_u8_8x2(d + 2 * dst_stride, dst_stride, d2, d3);
}
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src += 4 * src_stride;
dst += 4 * dst_stride;
h -= 4;
} while (h > 0);
}
}
}
static INLINE int16x4_t convolve8_4_x(uint8x16_t samples, const int8x8_t filter,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x2_t permute_tbl) {
int8x16_t clamped_samples, permuted_samples[2];
int32x4_t sum;
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples = vreinterpretq_s8_u8(vsubq_u8(samples, range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples, permute_tbl.val[1]);
// Accumulate dot product into 'correction' to account for range clamp.
sum = vdotq_lane_s32(correction, permuted_samples[0], filter, 0);
sum = vdotq_lane_s32(sum, permuted_samples[1], filter, 1);
// Packing is performed by the caller.
return vmovn_s32(sum);
}
static INLINE uint8x8_t convolve8_8_x(uint8x16_t samples, const int8x8_t filter,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x3_t permute_tbl) {
int8x16_t clamped_samples, permuted_samples[3];
int32x4_t sum[2];
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples = vreinterpretq_s8_u8(vsubq_u8(samples, range_limit));
// Permute samples ready for dot product. */
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_s8(clamped_samples, permute_tbl.val[2]);
// Accumulate dot product into 'correction' to account for range clamp.
// First 4 output values.
sum[0] = vdotq_lane_s32(correction, permuted_samples[0], filter, 0);
sum[0] = vdotq_lane_s32(sum[0], permuted_samples[1], filter, 1);
// Second 4 output values.
sum[1] = vdotq_lane_s32(correction, permuted_samples[1], filter, 0);
sum[1] = vdotq_lane_s32(sum[1], permuted_samples[2], filter, 1);
// Narrow and re-pack.
int16x8_t sum_s16 = vcombine_s16(vmovn_s32(sum[0]), vmovn_s32(sum[1]));
// We halved the convolution filter values so - 1 from the right shift.
return vqrshrun_n_s16(sum_s16, FILTER_BITS - 1);
}
void av1_convolve_x_sr_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const InterpFilterParams *filter_params_x,
const int subpel_x_qn,
ConvolveParams *conv_params) {
(void)conv_params;
const uint8_t horiz_offset = filter_params_x->taps / 2 - 1;
src -= horiz_offset;
const int16_t *x_filter_ptr = av1_get_interp_filter_subpel_kernel(
filter_params_x, subpel_x_qn & SUBPEL_MASK);
if (filter_params_x->taps > 8) {
convolve_x_sr_12tap_neon(src, src_stride, dst, dst_stride, w, h,
x_filter_ptr);
return;
}
// Filter values are even, so halve to reduce intermediate precision
// requirements.
const int8x8_t x_filter = vshrn_n_s16(vld1q_s16(x_filter_ptr), 1);
// Dot product constants.
const int32_t correction_s32 = vaddlvq_s16(vshll_n_s8(x_filter, FILTER_BITS));
// This shim of (1 << ((ROUND0_BITS - 1) - 1) enables us to use a single
// rounding right shift by FILTER_BITS - instead of a first rounding right
// shift by ROUND0_BITS, followed by second rounding right shift by
// FILTER_BITS - ROUND0_BITS.
// The outermost -1 is needed because we halved the filter values.
const int32x4_t correction =
vdupq_n_s32(correction_s32 + (1 << ((ROUND0_BITS - 1) - 1)));
const uint8x16_t range_limit = vdupq_n_u8(128);
if (w <= 4) {
const uint8x16x2_t permute_tbl = vld1q_u8_x2(dot_prod_permute_tbl);
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(src, src_stride, &s0, &s1, &s2, &s3);
int16x4_t d0 =
convolve8_4_x(s0, x_filter, correction, range_limit, permute_tbl);
int16x4_t d1 =
convolve8_4_x(s1, x_filter, correction, range_limit, permute_tbl);
int16x4_t d2 =
convolve8_4_x(s2, x_filter, correction, range_limit, permute_tbl);
int16x4_t d3 =
convolve8_4_x(s3, x_filter, correction, range_limit, permute_tbl);
// We halved the convolution filter values so - 1 from the right shift.
uint8x8_t d01 = vqrshrun_n_s16(vcombine_s16(d0, d1), FILTER_BITS - 1);
uint8x8_t d23 = vqrshrun_n_s16(vcombine_s16(d2, d3), FILTER_BITS - 1);
if (w == 2) {
store_u8_2x1(dst + 0 * dst_stride, d01, 0);
store_u8_2x1(dst + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst + 2 * dst_stride, d23, 0);
store_u8_2x1(dst + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst + 0 * dst_stride, d01, 0);
store_u8_4x1(dst + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst + 2 * dst_stride, d23, 0);
store_u8_4x1(dst + 3 * dst_stride, d23, 1);
}
}
src += 4 * src_stride;
dst += 4 * dst_stride;
h -= 4;
} while (h > 0);
} else {
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
do {
int width = w;
const uint8_t *s = src;
uint8_t *d = dst;
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(s, src_stride, &s0, &s1, &s2, &s3);
uint8x8_t d0 =
convolve8_8_x(s0, x_filter, correction, range_limit, permute_tbl);
uint8x8_t d1 =
convolve8_8_x(s1, x_filter, correction, range_limit, permute_tbl);
uint8x8_t d2 =
convolve8_8_x(s2, x_filter, correction, range_limit, permute_tbl);
uint8x8_t d3 =
convolve8_8_x(s3, x_filter, correction, range_limit, permute_tbl);
vst1_u8(d + 0 * dst_stride, d0);
vst1_u8(d + 1 * dst_stride, d1);
if (h != 2) {
vst1_u8(d + 2 * dst_stride, d2);
vst1_u8(d + 3 * dst_stride, d3);
}
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src += 4 * src_stride;
dst += 4 * dst_stride;
h -= 4;
} while (h > 0);
}
}
#else // !(AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD))
static INLINE int16x4_t convolve12_4_x(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7,
const int16x4_t s8, const int16x4_t s9,
const int16x4_t s10, const int16x4_t s11,
const int16x8_t x_filter_0_7,
const int16x4_t x_filter_8_11,
const int32x4_t horiz_const) {
const int16x4_t x_filter_0_3 = vget_low_s16(x_filter_0_7);
const int16x4_t x_filter_4_7 = vget_high_s16(x_filter_0_7);
int32x4_t sum = horiz_const;
sum = vmlal_lane_s16(sum, s0, x_filter_0_3, 0);
sum = vmlal_lane_s16(sum, s1, x_filter_0_3, 1);
sum = vmlal_lane_s16(sum, s2, x_filter_0_3, 2);
sum = vmlal_lane_s16(sum, s3, x_filter_0_3, 3);
sum = vmlal_lane_s16(sum, s4, x_filter_4_7, 0);
sum = vmlal_lane_s16(sum, s5, x_filter_4_7, 1);
sum = vmlal_lane_s16(sum, s6, x_filter_4_7, 2);
sum = vmlal_lane_s16(sum, s7, x_filter_4_7, 3);
sum = vmlal_lane_s16(sum, s8, x_filter_8_11, 0);
sum = vmlal_lane_s16(sum, s9, x_filter_8_11, 1);
sum = vmlal_lane_s16(sum, s10, x_filter_8_11, 2);
sum = vmlal_lane_s16(sum, s11, x_filter_8_11, 3);
return vqrshrn_n_s32(sum, FILTER_BITS);
}
static INLINE void convolve_x_sr_12tap_neon(const uint8_t *src_ptr,
int src_stride, uint8_t *dst_ptr,
const int dst_stride, int w, int h,
const int16_t *x_filter_ptr) {
const int16x8_t x_filter_0_7 = vld1q_s16(x_filter_ptr);
const int16x4_t x_filter_8_11 = vld1_s16(x_filter_ptr + 8);
// A shim of 1 << (ROUND0_BITS - 1) enables us to use a single rounding right
// shift by FILTER_BITS - instead of a first rounding right shift by
// ROUND0_BITS, followed by second rounding right shift by FILTER_BITS -
// ROUND0_BITS.
const int32x4_t horiz_const = vdupq_n_s32(1 << (ROUND0_BITS - 1));
#if AOM_ARCH_AARCH64
do {
const uint8_t *s = src_ptr;
uint8_t *d = dst_ptr;
int width = w;
uint8x8_t t0, t1, t2, t3;
load_u8_8x4(s, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s1 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s2 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s3 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t s4 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s5 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s6 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s7 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
load_u8_8x4(s + 8, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s8 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s9 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s10 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
s += 11;
do {
load_u8_8x4(s, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s11 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s12 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s13 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s14 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t d0 =
convolve12_4_x(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
x_filter_0_7, x_filter_8_11, horiz_const);
int16x4_t d1 =
convolve12_4_x(s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12,
x_filter_0_7, x_filter_8_11, horiz_const);
int16x4_t d2 =
convolve12_4_x(s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13,
x_filter_0_7, x_filter_8_11, horiz_const);
int16x4_t d3 =
convolve12_4_x(s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13, s14,
x_filter_0_7, x_filter_8_11, horiz_const);
transpose_s16_4x4d(&d0, &d1, &d2, &d3);
uint8x8_t d01 = vqmovun_s16(vcombine_s16(d0, d1));
uint8x8_t d23 = vqmovun_s16(vcombine_s16(d2, d3));
if (w == 2) {
store_u8_2x1(d + 0 * dst_stride, d01, 0);
store_u8_2x1(d + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(d + 2 * dst_stride, d23, 0);
store_u8_2x1(d + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(d + 0 * dst_stride, d01, 0);
store_u8_4x1(d + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(d + 2 * dst_stride, d23, 0);
store_u8_4x1(d + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s7 = s11;
s8 = s12;
s9 = s13;
s10 = s14;
s += 4;
d += 4;
width -= 4;
} while (width > 0);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 0);
#else // !AOM_ARCH_AARCH64
do {
const uint8_t *s = src_ptr;
uint8_t *d = dst_ptr;
int width = w;
do {
uint8x16_t t0 = vld1q_u8(s);
int16x8_t tt0 = vreinterpretq_s16_u16(vmovl_u8(vget_low_u8(t0)));
int16x8_t tt8 = vreinterpretq_s16_u16(vmovl_u8(vget_high_u8(t0)));
int16x4_t s0 = vget_low_s16(tt0);
int16x4_t s4 = vget_high_s16(tt0);
int16x4_t s8 = vget_low_s16(tt8);
int16x4_t s12 = vget_high_s16(tt8);
int16x4_t s1 = vext_s16(s0, s4, 1); // a1 a2 a3 a4
int16x4_t s2 = vext_s16(s0, s4, 2); // a2 a3 a4 a5
int16x4_t s3 = vext_s16(s0, s4, 3); // a3 a4 a5 a6
int16x4_t s5 = vext_s16(s4, s8, 1); // a5 a6 a7 a8
int16x4_t s6 = vext_s16(s4, s8, 2); // a6 a7 a8 a9
int16x4_t s7 = vext_s16(s4, s8, 3); // a7 a8 a9 a10
int16x4_t s9 = vext_s16(s8, s12, 1); // a9 a10 a11 a12
int16x4_t s10 = vext_s16(s8, s12, 2); // a10 a11 a12 a13
int16x4_t s11 = vext_s16(s8, s12, 3); // a11 a12 a13 a14
int16x4_t d0 =
convolve12_4_x(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
x_filter_0_7, x_filter_8_11, horiz_const);
uint8x8_t dd0 = vqmovun_s16(vcombine_s16(d0, vdup_n_s16(0)));
if (w == 2) {
store_u8_2x1(d, dd0, 0);
} else {
store_u8_4x1(d, dd0, 0);
}
s += 4;
d += 4;
width -= 4;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
#endif // AOM_ARCH_AARCH64
}
static INLINE uint8x8_t convolve8_4_x(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7,
const int16x8_t filter,
const int16x4_t horiz_const) {
const int16x4_t filter_lo = vget_low_s16(filter);
const int16x4_t filter_hi = vget_high_s16(filter);
int16x4_t sum = horiz_const;
sum = vmla_lane_s16(sum, s0, filter_lo, 0);
sum = vmla_lane_s16(sum, s1, filter_lo, 1);
sum = vmla_lane_s16(sum, s2, filter_lo, 2);
sum = vmla_lane_s16(sum, s3, filter_lo, 3);
sum = vmla_lane_s16(sum, s4, filter_hi, 0);
sum = vmla_lane_s16(sum, s5, filter_hi, 1);
sum = vmla_lane_s16(sum, s6, filter_hi, 2);
sum = vmla_lane_s16(sum, s7, filter_hi, 3);
// We halved the convolution filter values so - 1 from the right shift.
return vqrshrun_n_s16(vcombine_s16(sum, vdup_n_s16(0)), FILTER_BITS - 1);
}
static INLINE uint8x8_t convolve8_8_x(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t s6, const int16x8_t s7,
const int16x8_t filter,
const int16x8_t horiz_const) {
const int16x4_t filter_lo = vget_low_s16(filter);
const int16x4_t filter_hi = vget_high_s16(filter);
int16x8_t sum = horiz_const;
sum = vmlaq_lane_s16(sum, s0, filter_lo, 0);
sum = vmlaq_lane_s16(sum, s1, filter_lo, 1);
sum = vmlaq_lane_s16(sum, s2, filter_lo, 2);
sum = vmlaq_lane_s16(sum, s3, filter_lo, 3);
sum = vmlaq_lane_s16(sum, s4, filter_hi, 0);
sum = vmlaq_lane_s16(sum, s5, filter_hi, 1);
sum = vmlaq_lane_s16(sum, s6, filter_hi, 2);
sum = vmlaq_lane_s16(sum, s7, filter_hi, 3);
// We halved the convolution filter values so - 1 from the right shift.
return vqrshrun_n_s16(sum, FILTER_BITS - 1);
}
void av1_convolve_x_sr_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const InterpFilterParams *filter_params_x,
const int subpel_x_qn,
ConvolveParams *conv_params) {
(void)conv_params;
const uint8_t horiz_offset = filter_params_x->taps / 2 - 1;
src -= horiz_offset;
const int16_t *x_filter_ptr = av1_get_interp_filter_subpel_kernel(
filter_params_x, subpel_x_qn & SUBPEL_MASK);
if (filter_params_x->taps > 8) {
convolve_x_sr_12tap_neon(src, src_stride, dst, dst_stride, w, h,
x_filter_ptr);
return;
}
// Filter values are even so halve to reduce precision requirements.
const int16x8_t x_filter = vshrq_n_s16(vld1q_s16(x_filter_ptr), 1);
// This shim of 1 << ((ROUND0_BITS - 1) - 1) enables us to use a single
// rounding right shift by FILTER_BITS - instead of a first rounding right
// shift by ROUND0_BITS, followed by second rounding right shift by
// FILTER_BITS - ROUND0_BITS.
// The outermost -1 is needed because we halved the filter values.
const int16x8_t horiz_const = vdupq_n_s16(1 << ((ROUND0_BITS - 1) - 1));
if (w <= 4) {
do {
uint8x8_t t0 = vld1_u8(src); // a0 a1 a2 a3 a4 a5 a6 a7
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s4 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
uint8x8_t t8 = vld1_u8(src + 8); // a8 a9 a10 a11 a12 a13 a14 a15
int16x4_t s8 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t8)));
int16x4_t s1 = vext_s16(s0, s4, 1); // a1 a2 a3 a4
int16x4_t s2 = vext_s16(s0, s4, 2); // a2 a3 a4 a5
int16x4_t s3 = vext_s16(s0, s4, 3); // a3 a4 a5 a6
int16x4_t s5 = vext_s16(s4, s8, 1); // a5 a6 a7 a8
int16x4_t s6 = vext_s16(s4, s8, 2); // a6 a7 a8 a9
int16x4_t s7 = vext_s16(s4, s8, 3); // a7 a8 a9 a10
uint8x8_t d0 = convolve8_4_x(s0, s1, s2, s3, s4, s5, s6, s7, x_filter,
vget_low_s16(horiz_const));
if (w == 4) {
store_u8_4x1(dst, d0, 0);
} else if (w == 2) {
store_u8_2x1(dst, d0, 0);
}
src += src_stride;
dst += dst_stride;
} while (--h != 0);
} else {
#if AOM_ARCH_AARCH64
while (h >= 8) {
uint8x8_t t0, t1, t2, t3, t4, t5, t6, t7;
load_u8_8x8(src, src_stride, &t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7);
transpose_u8_8x8(&t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7);
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
int16x8_t s1 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s2 = vreinterpretq_s16_u16(vmovl_u8(t2));
int16x8_t s3 = vreinterpretq_s16_u16(vmovl_u8(t3));
int16x8_t s4 = vreinterpretq_s16_u16(vmovl_u8(t4));
int16x8_t s5 = vreinterpretq_s16_u16(vmovl_u8(t5));
int16x8_t s6 = vreinterpretq_s16_u16(vmovl_u8(t6));
int width = w;
const uint8_t *s = src + 7;
uint8_t *d = dst;
__builtin_prefetch(d + 0 * dst_stride);
__builtin_prefetch(d + 1 * dst_stride);
__builtin_prefetch(d + 2 * dst_stride);
__builtin_prefetch(d + 3 * dst_stride);
__builtin_prefetch(d + 4 * dst_stride);
__builtin_prefetch(d + 5 * dst_stride);
__builtin_prefetch(d + 6 * dst_stride);
__builtin_prefetch(d + 7 * dst_stride);
do {
uint8x8_t t8, t9, t10, t11, t12, t13, t14;
load_u8_8x8(s, src_stride, &t7, &t8, &t9, &t10, &t11, &t12, &t13, &t14);
transpose_u8_8x8(&t7, &t8, &t9, &t10, &t11, &t12, &t13, &t14);
int16x8_t s7 = vreinterpretq_s16_u16(vmovl_u8(t7));
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t8));
int16x8_t s9 = vreinterpretq_s16_u16(vmovl_u8(t9));
int16x8_t s10 = vreinterpretq_s16_u16(vmovl_u8(t10));
int16x8_t s11 = vreinterpretq_s16_u16(vmovl_u8(t11));
int16x8_t s12 = vreinterpretq_s16_u16(vmovl_u8(t12));
int16x8_t s13 = vreinterpretq_s16_u16(vmovl_u8(t13));
int16x8_t s14 = vreinterpretq_s16_u16(vmovl_u8(t14));
uint8x8_t d0 = convolve8_8_x(s0, s1, s2, s3, s4, s5, s6, s7, x_filter,
horiz_const);
uint8x8_t d1 = convolve8_8_x(s1, s2, s3, s4, s5, s6, s7, s8, x_filter,
horiz_const);
uint8x8_t d2 = convolve8_8_x(s2, s3, s4, s5, s6, s7, s8, s9, x_filter,
horiz_const);
uint8x8_t d3 = convolve8_8_x(s3, s4, s5, s6, s7, s8, s9, s10, x_filter,
horiz_const);
uint8x8_t d4 = convolve8_8_x(s4, s5, s6, s7, s8, s9, s10, s11, x_filter,
horiz_const);
uint8x8_t d5 = convolve8_8_x(s5, s6, s7, s8, s9, s10, s11, s12,
x_filter, horiz_const);
uint8x8_t d6 = convolve8_8_x(s6, s7, s8, s9, s10, s11, s12, s13,
x_filter, horiz_const);
uint8x8_t d7 = convolve8_8_x(s7, s8, s9, s10, s11, s12, s13, s14,
x_filter, horiz_const);
transpose_u8_8x8(&d0, &d1, &d2, &d3, &d4, &d5, &d6, &d7);
store_u8_8x8(d, dst_stride, d0, d1, d2, d3, d4, d5, d6, d7);
s0 = s8;
s1 = s9;
s2 = s10;
s3 = s11;
s4 = s12;
s5 = s13;
s6 = s14;
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src += 8 * src_stride;
dst += 8 * dst_stride;
h -= 8;
}
#endif // !AOM_ARCH_AARCH64
while (h-- != 0) {
uint8x8_t t0 = vld1_u8(src); // a0 a1 a2 a3 a4 a5 a6 a7
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
int width = w;
const uint8_t *s = src + 8;
uint8_t *d = dst;
__builtin_prefetch(d);
do {
uint8x8_t t8 = vld1_u8(s); // a8 a9 a10 a11 a12 a13 a14 a15
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t8));
int16x8_t s1 = vextq_s16(s0, s8, 1); // a1 a2 a3 a4 a5 a6 a7 a8
int16x8_t s2 = vextq_s16(s0, s8, 2); // a2 a3 a4 a5 a6 a7 a8 a9
int16x8_t s3 = vextq_s16(s0, s8, 3); // a3 a4 a5 a6 a7 a8 a9 a10
int16x8_t s4 = vextq_s16(s0, s8, 4); // a4 a5 a6 a7 a8 a9 a10 a11
int16x8_t s5 = vextq_s16(s0, s8, 5); // a5 a6 a7 a8 a9 a10 a11 a12
int16x8_t s6 = vextq_s16(s0, s8, 6); // a6 a7 a8 a9 a10 a11 a12 a13
int16x8_t s7 = vextq_s16(s0, s8, 7); // a7 a8 a9 a10 a11 a12 a13 a14
uint8x8_t d0 = convolve8_8_x(s0, s1, s2, s3, s4, s5, s6, s7, x_filter,
horiz_const);
vst1_u8(d, d0);
s0 = s8;
s += 8;
d += 8;
width -= 8;
} while (width != 0);
src += src_stride;
dst += dst_stride;
}
}
}
#endif // AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_MATMUL_INT8)
static INLINE int16x4_t convolve6_4_y(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x8_t y_filter_0_7) {
const int16x4_t y_filter_0_3 = vget_low_s16(y_filter_0_7);
const int16x4_t y_filter_4_7 = vget_high_s16(y_filter_0_7);
// Filter values at indices 0 and 7 are 0.
int16x4_t sum = vmul_lane_s16(s0, y_filter_0_3, 1);
sum = vmla_lane_s16(sum, s1, y_filter_0_3, 2);
sum = vmla_lane_s16(sum, s2, y_filter_0_3, 3);
sum = vmla_lane_s16(sum, s3, y_filter_4_7, 0);
sum = vmla_lane_s16(sum, s4, y_filter_4_7, 1);
sum = vmla_lane_s16(sum, s5, y_filter_4_7, 2);
return sum;
}
static INLINE uint8x8_t convolve6_8_y(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t y_filters) {
const int16x4_t y_filter_lo = vget_low_s16(y_filters);
const int16x4_t y_filter_hi = vget_high_s16(y_filters);
// Filter values at indices 0 and 7 are 0.
int16x8_t sum = vmulq_lane_s16(s0, y_filter_lo, 1);
sum = vmlaq_lane_s16(sum, s1, y_filter_lo, 2);
sum = vmlaq_lane_s16(sum, s2, y_filter_lo, 3);
sum = vmlaq_lane_s16(sum, s3, y_filter_hi, 0);
sum = vmlaq_lane_s16(sum, s4, y_filter_hi, 1);
sum = vmlaq_lane_s16(sum, s5, y_filter_hi, 2);
// We halved the convolution filter values so -1 from the right shift.
return vqrshrun_n_s16(sum, FILTER_BITS - 1);
}
static INLINE void convolve_y_sr_6tap_neon(const uint8_t *src_ptr,
int src_stride, uint8_t *dst_ptr,
const int dst_stride, int w, int h,
const int16x8_t y_filter) {
if (w <= 4) {
uint8x8_t t0 = load_unaligned_u8_4x1(src_ptr + 0 * src_stride);
uint8x8_t t1 = load_unaligned_u8_4x1(src_ptr + 1 * src_stride);
uint8x8_t t2 = load_unaligned_u8_4x1(src_ptr + 2 * src_stride);
uint8x8_t t3 = load_unaligned_u8_4x1(src_ptr + 3 * src_stride);
uint8x8_t t4 = load_unaligned_u8_4x1(src_ptr + 4 * src_stride);
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s1 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s2 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s3 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t s4 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t4)));
src_ptr += 5 * src_stride;
do {
#if AOM_ARCH_AARCH64
uint8x8_t t5 = load_unaligned_u8_4x1(src_ptr + 0 * src_stride);
uint8x8_t t6 = load_unaligned_u8_4x1(src_ptr + 1 * src_stride);
uint8x8_t t7 = load_unaligned_u8_4x1(src_ptr + 2 * src_stride);
uint8x8_t t8 = load_unaligned_u8_4x1(src_ptr + 3 * src_stride);
int16x4_t s5 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t5)));
int16x4_t s6 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t6)));
int16x4_t s7 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t7)));
int16x4_t s8 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t8)));
int16x4_t d0 = convolve6_4_y(s0, s1, s2, s3, s4, s5, y_filter);
int16x4_t d1 = convolve6_4_y(s1, s2, s3, s4, s5, s6, y_filter);
int16x4_t d2 = convolve6_4_y(s2, s3, s4, s5, s6, s7, y_filter);
int16x4_t d3 = convolve6_4_y(s3, s4, s5, s6, s7, s8, y_filter);
// We halved the convolution filter values so -1 from the right shift.
uint8x8_t d01 = vqrshrun_n_s16(vcombine_s16(d0, d1), FILTER_BITS - 1);
uint8x8_t d23 = vqrshrun_n_s16(vcombine_s16(d2, d3), FILTER_BITS - 1);
if (w == 2) {
store_u8_2x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_2x1(dst_ptr + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_2x1(dst_ptr + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_4x1(dst_ptr + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_4x1(dst_ptr + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
#else // !AOM_ARCH_AARCH64
uint8x8_t t5 = load_unaligned_u8_4x1(src_ptr);
int16x4_t s5 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t5)));
int16x4_t d0 = convolve6_4_y(s0, s1, s2, s3, s4, s5, y_filter);
// We halved the convolution filter values so -1 from the right shift.
uint8x8_t d01 =
vqrshrun_n_s16(vcombine_s16(d0, vdup_n_s16(0)), FILTER_BITS - 1);
if (w == 2) {
store_u8_2x1(dst_ptr, d01, 0);
} else {
store_u8_4x1(dst_ptr, d01, 0);
}
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
src_ptr += src_stride;
dst_ptr += dst_stride;
h--;
#endif // AOM_ARCH_AARCH64
} while (h > 0);
} else {
do {
const uint8_t *s = src_ptr;
uint8_t *d = dst_ptr;
int height = h;
uint8x8_t t0, t1, t2, t3, t4;
load_u8_8x5(s, src_stride, &t0, &t1, &t2, &t3, &t4);
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
int16x8_t s1 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s2 = vreinterpretq_s16_u16(vmovl_u8(t2));
int16x8_t s3 = vreinterpretq_s16_u16(vmovl_u8(t3));
int16x8_t s4 = vreinterpretq_s16_u16(vmovl_u8(t4));
s += 5 * src_stride;
do {
#if AOM_ARCH_AARCH64
uint8x8_t t5, t6, t7, t8;
load_u8_8x4(s, src_stride, &t5, &t6, &t7, &t8);
int16x8_t s5 = vreinterpretq_s16_u16(vmovl_u8(t5));
int16x8_t s6 = vreinterpretq_s16_u16(vmovl_u8(t6));
int16x8_t s7 = vreinterpretq_s16_u16(vmovl_u8(t7));
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t8));
uint8x8_t d0 = convolve6_8_y(s0, s1, s2, s3, s4, s5, y_filter);
uint8x8_t d1 = convolve6_8_y(s1, s2, s3, s4, s5, s6, y_filter);
uint8x8_t d2 = convolve6_8_y(s2, s3, s4, s5, s6, s7, y_filter);
uint8x8_t d3 = convolve6_8_y(s3, s4, s5, s6, s7, s8, y_filter);
if (h != 2) {
store_u8_8x4(d, dst_stride, d0, d1, d2, d3);
} else {
store_u8_8x2(d, dst_stride, d0, d1);
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s += 4 * src_stride;
d += 4 * dst_stride;
height -= 4;
#else // !AOM_ARCH_AARCH64
int16x8_t s5 = vreinterpretq_s16_u16(vmovl_u8(vld1_u8(s)));
uint8x8_t d0 = convolve6_8_y(s0, s1, s2, s3, s4, s5, y_filter);
vst1_u8(d, d0);
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
s += src_stride;
d += dst_stride;
height--;
#endif // AOM_ARCH_AARCH64
} while (height > 0);
src_ptr += 8;
dst_ptr += 8;
w -= 8;
} while (w > 0);
}
}
static INLINE int16x4_t convolve8_4_y(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7,
const int16x8_t filter) {
const int16x4_t filter_lo = vget_low_s16(filter);
const int16x4_t filter_hi = vget_high_s16(filter);
int16x4_t sum = vmul_lane_s16(s0, filter_lo, 0);
sum = vmla_lane_s16(sum, s1, filter_lo, 1);
sum = vmla_lane_s16(sum, s2, filter_lo, 2);
sum = vmla_lane_s16(sum, s3, filter_lo, 3);
sum = vmla_lane_s16(sum, s4, filter_hi, 0);
sum = vmla_lane_s16(sum, s5, filter_hi, 1);
sum = vmla_lane_s16(sum, s6, filter_hi, 2);
sum = vmla_lane_s16(sum, s7, filter_hi, 3);
return sum;
}
static INLINE uint8x8_t convolve8_8_y(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t s6, const int16x8_t s7,
const int16x8_t filter) {
const int16x4_t filter_lo = vget_low_s16(filter);
const int16x4_t filter_hi = vget_high_s16(filter);
int16x8_t sum = vmulq_lane_s16(s0, filter_lo, 0);
sum = vmlaq_lane_s16(sum, s1, filter_lo, 1);
sum = vmlaq_lane_s16(sum, s2, filter_lo, 2);
sum = vmlaq_lane_s16(sum, s3, filter_lo, 3);
sum = vmlaq_lane_s16(sum, s4, filter_hi, 0);
sum = vmlaq_lane_s16(sum, s5, filter_hi, 1);
sum = vmlaq_lane_s16(sum, s6, filter_hi, 2);
sum = vmlaq_lane_s16(sum, s7, filter_hi, 3);
// We halved the convolution filter values so -1 from the right shift.
return vqrshrun_n_s16(sum, FILTER_BITS - 1);
}
static INLINE void convolve_y_sr_8tap_neon(const uint8_t *src_ptr,
int src_stride, uint8_t *dst_ptr,
const int dst_stride, int w, int h,
const int16x8_t y_filter) {
if (w <= 4) {
uint8x8_t t0 = load_unaligned_u8_4x1(src_ptr + 0 * src_stride);
uint8x8_t t1 = load_unaligned_u8_4x1(src_ptr + 1 * src_stride);
uint8x8_t t2 = load_unaligned_u8_4x1(src_ptr + 2 * src_stride);
uint8x8_t t3 = load_unaligned_u8_4x1(src_ptr + 3 * src_stride);
uint8x8_t t4 = load_unaligned_u8_4x1(src_ptr + 4 * src_stride);
uint8x8_t t5 = load_unaligned_u8_4x1(src_ptr + 5 * src_stride);
uint8x8_t t6 = load_unaligned_u8_4x1(src_ptr + 6 * src_stride);
int16x4_t s0 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t0)));
int16x4_t s1 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t1)));
int16x4_t s2 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t2)));
int16x4_t s3 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t3)));
int16x4_t s4 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t4)));
int16x4_t s5 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t5)));
int16x4_t s6 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t6)));
src_ptr += 7 * src_stride;
do {
#if AOM_ARCH_AARCH64
uint8x8_t t7 = load_unaligned_u8_4x1(src_ptr + 0 * src_stride);
uint8x8_t t8 = load_unaligned_u8_4x1(src_ptr + 1 * src_stride);
uint8x8_t t9 = load_unaligned_u8_4x1(src_ptr + 2 * src_stride);
uint8x8_t t10 = load_unaligned_u8_4x1(src_ptr + 3 * src_stride);
int16x4_t s7 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t7)));
int16x4_t s8 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t8)));
int16x4_t s9 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t9)));
int16x4_t s10 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t10)));
int16x4_t d0 = convolve8_4_y(s0, s1, s2, s3, s4, s5, s6, s7, y_filter);
int16x4_t d1 = convolve8_4_y(s1, s2, s3, s4, s5, s6, s7, s8, y_filter);
int16x4_t d2 = convolve8_4_y(s2, s3, s4, s5, s6, s7, s8, s9, y_filter);
int16x4_t d3 = convolve8_4_y(s3, s4, s5, s6, s7, s8, s9, s10, y_filter);
// We halved the convolution filter values so -1 from the right shift.
uint8x8_t d01 = vqrshrun_n_s16(vcombine_s16(d0, d1), FILTER_BITS - 1);
uint8x8_t d23 = vqrshrun_n_s16(vcombine_s16(d2, d3), FILTER_BITS - 1);
if (w == 2) {
store_u8_2x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_2x1(dst_ptr + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_2x1(dst_ptr + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_4x1(dst_ptr + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_4x1(dst_ptr + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
#else // !AOM_ARCH_AARCH64
uint8x8_t t7 = load_unaligned_u8_4x1(src_ptr);
int16x4_t s7 = vreinterpret_s16_u16(vget_low_u16(vmovl_u8(t7)));
int16x4_t d0 = convolve8_4_y(s0, s1, s2, s3, s4, s5, s6, s7, y_filter);
// We halved the convolution filter values so -1 from the right shift.
uint8x8_t d01 =
vqrshrun_n_s16(vcombine_s16(d0, vdup_n_s16(0)), FILTER_BITS - 1);
if (w == 4) {
store_u8_4x1(dst_ptr, d01, 0);
} else if (w == 2) {
store_u8_2x1(dst_ptr, d01, 0);
}
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
s5 = s6;
s6 = s7;
src_ptr += src_stride;
dst_ptr += dst_stride;
h--;
#endif // AOM_ARCH_AARCH64
} while (h > 0);
} else {
do {
const uint8_t *s = src_ptr;
uint8_t *d = dst_ptr;
int height = h;
uint8x8_t t0, t1, t2, t3, t4, t5, t6;
load_u8_8x7(s, src_stride, &t0, &t1, &t2, &t3, &t4, &t5, &t6);
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
int16x8_t s1 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s2 = vreinterpretq_s16_u16(vmovl_u8(t2));
int16x8_t s3 = vreinterpretq_s16_u16(vmovl_u8(t3));
int16x8_t s4 = vreinterpretq_s16_u16(vmovl_u8(t4));
int16x8_t s5 = vreinterpretq_s16_u16(vmovl_u8(t5));
int16x8_t s6 = vreinterpretq_s16_u16(vmovl_u8(t6));
s += 7 * src_stride;
do {
#if AOM_ARCH_AARCH64
uint8x8_t t7, t8, t9, t10;
load_u8_8x4(s, src_stride, &t7, &t8, &t9, &t10);
int16x8_t s7 = vreinterpretq_s16_u16(vmovl_u8(t7));
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t8));
int16x8_t s9 = vreinterpretq_s16_u16(vmovl_u8(t9));
int16x8_t s10 = vreinterpretq_s16_u16(vmovl_u8(t10));
uint8x8_t d0 = convolve8_8_y(s0, s1, s2, s3, s4, s5, s6, s7, y_filter);
uint8x8_t d1 = convolve8_8_y(s1, s2, s3, s4, s5, s6, s7, s8, y_filter);
uint8x8_t d2 = convolve8_8_y(s2, s3, s4, s5, s6, s7, s8, s9, y_filter);
uint8x8_t d3 = convolve8_8_y(s3, s4, s5, s6, s7, s8, s9, s10, y_filter);
if (h != 2) {
store_u8_8x4(d, dst_stride, d0, d1, d2, d3);
} else {
store_u8_8x2(d, dst_stride, d0, d1);
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s += 4 * src_stride;
d += 4 * dst_stride;
height -= 4;
#else // !AOM_ARCH_AARCH64
int16x8_t s7 = vreinterpretq_s16_u16(vmovl_u8(vld1_u8(s)));
uint8x8_t d0 = convolve8_8_y(s0, s1, s2, s3, s4, s5, s6, s7, y_filter);
vst1_u8(d, d0);
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
s5 = s6;
s6 = s7;
s += src_stride;
d += dst_stride;
height--;
#endif // AOM_ARCH_AARCH64
} while (height > 0);
src_ptr += 8;
dst_ptr += 8;
w -= 8;
} while (w > 0);
}
}
static INLINE int16x4_t convolve12_4_y(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7,
const int16x4_t s8, const int16x4_t s9,
const int16x4_t s10, const int16x4_t s11,
const int16x8_t y_filter_0_7,
const int16x4_t y_filter_8_11) {
const int16x4_t y_filter_0_3 = vget_low_s16(y_filter_0_7);
const int16x4_t y_filter_4_7 = vget_high_s16(y_filter_0_7);
int16x4_t sum;
sum = vmul_lane_s16(s0, y_filter_0_3, 0);
sum = vmla_lane_s16(sum, s1, y_filter_0_3, 1);
sum = vmla_lane_s16(sum, s2, y_filter_0_3, 2);
sum = vmla_lane_s16(sum, s3, y_filter_0_3, 3);
sum = vmla_lane_s16(sum, s4, y_filter_4_7, 0);
sum = vmla_lane_s16(sum, s7, y_filter_4_7, 3);
sum = vmla_lane_s16(sum, s8, y_filter_8_11, 0);
sum = vmla_lane_s16(sum, s9, y_filter_8_11, 1);
sum = vmla_lane_s16(sum, s10, y_filter_8_11, 2);
sum = vmla_lane_s16(sum, s11, y_filter_8_11, 3);
// Saturating addition is required for the largest filter taps to avoid
// overflow (while staying in 16-bit elements.)
sum = vqadd_s16(sum, vmul_lane_s16(s5, y_filter_4_7, 1));
sum = vqadd_s16(sum, vmul_lane_s16(s6, y_filter_4_7, 2));
return sum;
}
static INLINE uint8x8_t convolve12_8_y(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t s6, const int16x8_t s7,
const int16x8_t s8, const int16x8_t s9,
const int16x8_t s10, const int16x8_t s11,
const int16x8_t y_filter_0_7,
const int16x4_t y_filter_8_11) {
const int16x4_t y_filter_0_3 = vget_low_s16(y_filter_0_7);
const int16x4_t y_filter_4_7 = vget_high_s16(y_filter_0_7);
int16x8_t sum;
sum = vmulq_lane_s16(s0, y_filter_0_3, 0);
sum = vmlaq_lane_s16(sum, s1, y_filter_0_3, 1);
sum = vmlaq_lane_s16(sum, s2, y_filter_0_3, 2);
sum = vmlaq_lane_s16(sum, s3, y_filter_0_3, 3);
sum = vmlaq_lane_s16(sum, s4, y_filter_4_7, 0);
sum = vmlaq_lane_s16(sum, s7, y_filter_4_7, 3);
sum = vmlaq_lane_s16(sum, s8, y_filter_8_11, 0);
sum = vmlaq_lane_s16(sum, s9, y_filter_8_11, 1);
sum = vmlaq_lane_s16(sum, s10, y_filter_8_11, 2);
sum = vmlaq_lane_s16(sum, s11, y_filter_8_11, 3);
// Saturating addition is required for the largest filter taps to avoid
// overflow (while staying in 16-bit elements.)
sum = vqaddq_s16(sum, vmulq_lane_s16(s5, y_filter_4_7, 1));
sum = vqaddq_s16(sum, vmulq_lane_s16(s6, y_filter_4_7, 2));
return vqrshrun_n_s16(sum, FILTER_BITS);
}
static INLINE void convolve_y_sr_12tap_neon(const uint8_t *src_ptr,
int src_stride, uint8_t *dst_ptr,
int dst_stride, int w, int h,
const int16_t *y_filter_ptr) {
const int16x8_t y_filter_0_7 = vld1q_s16(y_filter_ptr);
const int16x4_t y_filter_8_11 = vld1_s16(y_filter_ptr + 8);
if (w <= 4) {
uint8x8_t t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10;
load_u8_8x11(src_ptr, src_stride, &t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7,
&t8, &t9, &t10);
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s1 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s2 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s3 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t s4 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t4)));
int16x4_t s5 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t5)));
int16x4_t s6 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t6)));
int16x4_t s7 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t7)));
int16x4_t s8 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t8)));
int16x4_t s9 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t9)));
int16x4_t s10 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t10)));
src_ptr += 11 * src_stride;
do {
uint8x8_t t11, t12, t13, t14;
load_u8_8x4(src_ptr, src_stride, &t11, &t12, &t13, &t14);
int16x4_t s11 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t11)));
int16x4_t s12 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t12)));
int16x4_t s13 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t13)));
int16x4_t s14 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t14)));
int16x4_t d0 = convolve12_4_y(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10,
s11, y_filter_0_7, y_filter_8_11);
int16x4_t d1 = convolve12_4_y(s1, s2, s3, s4, s5, s6, s7, s8, s9, s10,
s11, s12, y_filter_0_7, y_filter_8_11);
int16x4_t d2 = convolve12_4_y(s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
s12, s13, y_filter_0_7, y_filter_8_11);
int16x4_t d3 = convolve12_4_y(s3, s4, s5, s6, s7, s8, s9, s10, s11, s12,
s13, s14, y_filter_0_7, y_filter_8_11);
uint8x8_t d01 = vqrshrun_n_s16(vcombine_s16(d0, d1), FILTER_BITS);
uint8x8_t d23 = vqrshrun_n_s16(vcombine_s16(d2, d3), FILTER_BITS);
if (w == 2) {
store_u8_2x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_2x1(dst_ptr + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_2x1(dst_ptr + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_4x1(dst_ptr + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_4x1(dst_ptr + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s7 = s11;
s8 = s12;
s9 = s13;
s10 = s14;
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 0);
} else {
do {
const uint8_t *s = src_ptr;
uint8_t *d = dst_ptr;
int height = h;
uint8x8_t t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10;
load_u8_8x11(s, src_stride, &t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7, &t8,
&t9, &t10);
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
int16x8_t s1 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s2 = vreinterpretq_s16_u16(vmovl_u8(t2));
int16x8_t s3 = vreinterpretq_s16_u16(vmovl_u8(t3));
int16x8_t s4 = vreinterpretq_s16_u16(vmovl_u8(t4));
int16x8_t s5 = vreinterpretq_s16_u16(vmovl_u8(t5));
int16x8_t s6 = vreinterpretq_s16_u16(vmovl_u8(t6));
int16x8_t s7 = vreinterpretq_s16_u16(vmovl_u8(t7));
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t8));
int16x8_t s9 = vreinterpretq_s16_u16(vmovl_u8(t9));
int16x8_t s10 = vreinterpretq_s16_u16(vmovl_u8(t10));
s += 11 * src_stride;
do {
uint8x8_t t11, t12, t13, t14;
load_u8_8x4(s, src_stride, &t11, &t12, &t13, &t14);
int16x8_t s11 = vreinterpretq_s16_u16(vmovl_u8(t11));
int16x8_t s12 = vreinterpretq_s16_u16(vmovl_u8(t12));
int16x8_t s13 = vreinterpretq_s16_u16(vmovl_u8(t13));
int16x8_t s14 = vreinterpretq_s16_u16(vmovl_u8(t14));
uint8x8_t d0 = convolve12_8_y(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9,
s10, s11, y_filter_0_7, y_filter_8_11);
uint8x8_t d1 = convolve12_8_y(s1, s2, s3, s4, s5, s6, s7, s8, s9, s10,
s11, s12, y_filter_0_7, y_filter_8_11);
uint8x8_t d2 = convolve12_8_y(s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
s12, s13, y_filter_0_7, y_filter_8_11);
uint8x8_t d3 = convolve12_8_y(s3, s4, s5, s6, s7, s8, s9, s10, s11, s12,
s13, s14, y_filter_0_7, y_filter_8_11);
if (h != 2) {
store_u8_8x4(d, dst_stride, d0, d1, d2, d3);
} else {
store_u8_8x2(d, dst_stride, d0, d1);
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s7 = s11;
s8 = s12;
s9 = s13;
s10 = s14;
s += 4 * src_stride;
d += 4 * dst_stride;
height -= 4;
} while (height > 0);
src_ptr += 8;
dst_ptr += 8;
w -= 8;
} while (w > 0);
}
}
void av1_convolve_y_sr_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const InterpFilterParams *filter_params_y,
const int subpel_y_qn) {
const int y_filter_taps = get_filter_tap(filter_params_y, subpel_y_qn);
const int clamped_y_taps = y_filter_taps < 6 ? 6 : y_filter_taps;
const int vert_offset = clamped_y_taps / 2 - 1;
src -= vert_offset * src_stride;
const int16_t *y_filter_ptr = av1_get_interp_filter_subpel_kernel(
filter_params_y, subpel_y_qn & SUBPEL_MASK);
if (y_filter_taps > 8) {
convolve_y_sr_12tap_neon(src, src_stride, dst, dst_stride, w, h,
y_filter_ptr);
return;
}
// Filter values are even so halve to reduce precision requirements.
const int16x8_t y_filter = vshrq_n_s16(vld1q_s16(y_filter_ptr), 1);
if (y_filter_taps < 8) {
convolve_y_sr_6tap_neon(src, src_stride, dst, dst_stride, w, h, y_filter);
} else {
convolve_y_sr_8tap_neon(src, src_stride, dst, dst_stride, w, h, y_filter);
}
}
#if AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_MATMUL_INT8)
static INLINE int16x4_t convolve12_4_2d_h(uint8x16_t samples,
const int8x16_t filters,
const uint8x16x3_t permute_tbl,
int32x4_t horiz_const) {
uint8x16_t permuted_samples[3];
int32x4_t sum;
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_u8(samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_u8(samples, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_u8(samples, permute_tbl.val[2]);
// First 4 output values.
sum = vusdotq_laneq_s32(horiz_const, permuted_samples[0], filters, 0);
sum = vusdotq_laneq_s32(sum, permuted_samples[1], filters, 1);
sum = vusdotq_laneq_s32(sum, permuted_samples[2], filters, 2);
// Narrow and re-pack.
return vshrn_n_s32(sum, ROUND0_BITS);
}
static INLINE int16x8_t convolve12_8_2d_h(uint8x16_t samples0,
uint8x16_t samples1,
const int8x16_t filters,
const uint8x16x3_t permute_tbl,
const int32x4_t horiz_const) {
uint8x16_t permuted_samples[4];
int32x4_t sum[2];
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_u8(samples0, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_u8(samples0, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_u8(samples0, permute_tbl.val[2]);
// {12, 13, 14, 15, 13, 14, 15, 16, 14, 15, 16, 17, 15, 16, 17, 18 }
permuted_samples[3] = vqtbl1q_u8(samples1, permute_tbl.val[2]);
// First 4 output values.
sum[0] = vusdotq_laneq_s32(horiz_const, permuted_samples[0], filters, 0);
sum[0] = vusdotq_laneq_s32(sum[0], permuted_samples[1], filters, 1);
sum[0] = vusdotq_laneq_s32(sum[0], permuted_samples[2], filters, 2);
// Second 4 output values.
sum[1] = vusdotq_laneq_s32(horiz_const, permuted_samples[1], filters, 0);
sum[1] = vusdotq_laneq_s32(sum[1], permuted_samples[2], filters, 1);
sum[1] = vusdotq_laneq_s32(sum[1], permuted_samples[3], filters, 2);
// Narrow and re-pack.
return vcombine_s16(vshrn_n_s32(sum[0], ROUND0_BITS),
vshrn_n_s32(sum[1], ROUND0_BITS));
}
static INLINE void convolve_2d_sr_horiz_12tap_neon(
const uint8_t *src_ptr, int src_stride, int16_t *dst_ptr,
const int dst_stride, int w, int h, const int16x8_t x_filter_0_7,
const int16x4_t x_filter_8_11) {
const int bd = 8;
// Special case the following no-op filter as 128 won't fit into the
// 8-bit signed dot-product instruction:
// { 0, 0, 0, 0, 0, 128, 0, 0, 0, 0, 0, 0 }
if (vgetq_lane_s16(x_filter_0_7, 5) == 128) {
const uint16x8_t horiz_const = vdupq_n_u16((1 << (bd - 1)));
// Undo the horizontal offset in the calling function.
src_ptr += 5;
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x8_t s0 = vld1_u8(s);
uint16x8_t d0 = vaddw_u8(horiz_const, s0);
d0 = vshlq_n_u16(d0, FILTER_BITS - ROUND0_BITS);
// Store 8 elements to avoid additional branches. This is safe if the
// actual block width is < 8 because the intermediate buffer is large
// enough to accommodate 128x128 blocks.
vst1q_s16(d, vreinterpretq_s16_u16(d0));
d += 8;
s += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
} else {
// Narrow filter values to 8-bit.
const int16x8x2_t x_filter_s16 = {
{ x_filter_0_7, vcombine_s16(x_filter_8_11, vdup_n_s16(0)) }
};
const int8x16_t x_filter = vcombine_s8(vmovn_s16(x_filter_s16.val[0]),
vmovn_s16(x_filter_s16.val[1]));
// This shim of 1 << (ROUND0_BITS - 1) enables us to use non-rounding shifts
// - which are generally faster than rounding shifts on modern CPUs.
const int32x4_t horiz_const =
vdupq_n_s32((1 << (bd + FILTER_BITS - 1)) + (1 << (ROUND0_BITS - 1)));
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
if (w <= 4) {
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(src_ptr, src_stride, &s0, &s1, &s2, &s3);
int16x4_t d0 =
convolve12_4_2d_h(s0, x_filter, permute_tbl, horiz_const);
int16x4_t d1 =
convolve12_4_2d_h(s1, x_filter, permute_tbl, horiz_const);
int16x4_t d2 =
convolve12_4_2d_h(s2, x_filter, permute_tbl, horiz_const);
int16x4_t d3 =
convolve12_4_2d_h(s3, x_filter, permute_tbl, horiz_const);
// Store 4 elements per row to avoid additional branches. (Safe.)
store_s16_4x4(dst_ptr, dst_stride, d0, d1, d2, d3);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 4);
do {
uint8x16_t s0 = vld1q_u8(src_ptr);
int16x4_t d0 =
convolve12_4_2d_h(s0, x_filter, permute_tbl, horiz_const);
// Store 4 elements to avoid additional branches. Safe as noted above.
vst1_s16(dst_ptr, d0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
} else {
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0[2], s1[2], s2[2], s3[2];
load_u8_16x4(s, src_stride, &s0[0], &s1[0], &s2[0], &s3[0]);
load_u8_16x4(s + 4, src_stride, &s0[1], &s1[1], &s2[1], &s3[1]);
int16x8_t d0 = convolve12_8_2d_h(s0[0], s0[1], x_filter, permute_tbl,
horiz_const);
int16x8_t d1 = convolve12_8_2d_h(s1[0], s1[1], x_filter, permute_tbl,
horiz_const);
int16x8_t d2 = convolve12_8_2d_h(s2[0], s2[1], x_filter, permute_tbl,
horiz_const);
int16x8_t d3 = convolve12_8_2d_h(s3[0], s3[1], x_filter, permute_tbl,
horiz_const);
store_s16_8x4(d, dst_stride, d0, d1, d2, d3);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 4);
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0[2];
s0[0] = vld1q_u8(s);
s0[1] = vld1q_u8(s + 4);
int16x8_t d0 = convolve12_8_2d_h(s0[0], s0[1], x_filter, permute_tbl,
horiz_const);
vst1q_s16(d, d0);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
}
}
}
#elif AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD)
static INLINE int16x4_t convolve12_4_2d_h(uint8x16_t samples,
const int8x16_t filters,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x3_t permute_tbl) {
int8x16_t clamped_samples, permuted_samples[3];
int32x4_t sum;
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples = vreinterpretq_s8_u8(vsubq_u8(samples, range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_s8(clamped_samples, permute_tbl.val[2]);
// Accumulate dot product into 'correction' to account for range clamp.
// First 4 output values.
sum = vdotq_laneq_s32(correction, permuted_samples[0], filters, 0);
sum = vdotq_laneq_s32(sum, permuted_samples[1], filters, 1);
sum = vdotq_laneq_s32(sum, permuted_samples[2], filters, 2);
// Narrow and re-pack.
return vshrn_n_s32(sum, ROUND0_BITS);
}
static INLINE int16x8_t convolve12_8_2d_h(uint8x16_t samples0,
uint8x16_t samples1,
const int8x16_t filters,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x3_t permute_tbl) {
int8x16_t clamped_samples[2], permuted_samples[4];
int32x4_t sum[2];
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples[0] = vreinterpretq_s8_u8(vsubq_u8(samples0, range_limit));
clamped_samples[1] = vreinterpretq_s8_u8(vsubq_u8(samples1, range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples[0], permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples[0], permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_s8(clamped_samples[0], permute_tbl.val[2]);
// {12, 13, 14, 15, 13, 14, 15, 16, 14, 15, 16, 17, 15, 16, 17, 18 }
permuted_samples[3] = vqtbl1q_s8(clamped_samples[1], permute_tbl.val[2]);
// Accumulate dot product into 'correction' to account for range clamp.
// First 4 output values.
sum[0] = vdotq_laneq_s32(correction, permuted_samples[0], filters, 0);
sum[0] = vdotq_laneq_s32(sum[0], permuted_samples[1], filters, 1);
sum[0] = vdotq_laneq_s32(sum[0], permuted_samples[2], filters, 2);
// Second 4 output values.
sum[1] = vdotq_laneq_s32(correction, permuted_samples[1], filters, 0);
sum[1] = vdotq_laneq_s32(sum[1], permuted_samples[2], filters, 1);
sum[1] = vdotq_laneq_s32(sum[1], permuted_samples[3], filters, 2);
// Narrow and re-pack.
return vcombine_s16(vshrn_n_s32(sum[0], ROUND0_BITS),
vshrn_n_s32(sum[1], ROUND0_BITS));
}
static INLINE void convolve_2d_sr_horiz_12tap_neon(
const uint8_t *src_ptr, int src_stride, int16_t *dst_ptr,
const int dst_stride, int w, int h, const int16x8_t x_filter_0_7,
const int16x4_t x_filter_8_11) {
const int bd = 8;
// Special case the following no-op filter as 128 won't fit into the 8-bit
// signed dot-product instruction:
// { 0, 0, 0, 0, 0, 128, 0, 0, 0, 0, 0, 0 }
if (vgetq_lane_s16(x_filter_0_7, 5) == 128) {
const uint16x8_t horiz_const = vdupq_n_u16((1 << (bd - 1)));
// Undo the horizontal offset in the calling function.
src_ptr += 5;
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x8_t s0 = vld1_u8(s);
uint16x8_t d0 = vaddw_u8(horiz_const, s0);
d0 = vshlq_n_u16(d0, FILTER_BITS - ROUND0_BITS);
// Store 8 elements to avoid additional branches. This is safe if the
// actual block width is < 8 because the intermediate buffer is large
// enough to accommodate 128x128 blocks.
vst1q_s16(d, vreinterpretq_s16_u16(d0));
d += 8;
s += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
} else {
// Narrow filter values to 8-bit.
const int16x8x2_t x_filter_s16 = {
{ x_filter_0_7, vcombine_s16(x_filter_8_11, vdup_n_s16(0)) }
};
const int8x16_t x_filter = vcombine_s8(vmovn_s16(x_filter_s16.val[0]),
vmovn_s16(x_filter_s16.val[1]));
// This shim of 1 << (ROUND0_BITS - 1) enables us to use non-rounding shifts
// - which are generally faster than rounding shifts on modern CPUs.
const int32_t horiz_const =
((1 << (bd + FILTER_BITS - 1)) + (1 << (ROUND0_BITS - 1)));
// Dot product constants.
const int32x4_t correct_tmp =
vaddq_s32(vpaddlq_s16(vshlq_n_s16(x_filter_s16.val[0], 7)),
vpaddlq_s16(vshlq_n_s16(x_filter_s16.val[1], 7)));
const int32x4_t correction =
vdupq_n_s32(vaddvq_s32(correct_tmp) + horiz_const);
const uint8x16_t range_limit = vdupq_n_u8(128);
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
if (w <= 4) {
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(src_ptr, src_stride, &s0, &s1, &s2, &s3);
int16x4_t d0 = convolve12_4_2d_h(s0, x_filter, correction, range_limit,
permute_tbl);
int16x4_t d1 = convolve12_4_2d_h(s1, x_filter, correction, range_limit,
permute_tbl);
int16x4_t d2 = convolve12_4_2d_h(s2, x_filter, correction, range_limit,
permute_tbl);
int16x4_t d3 = convolve12_4_2d_h(s3, x_filter, correction, range_limit,
permute_tbl);
// Store 4 elements per row to avoid additional branches. (Safe.)
store_s16_4x4(dst_ptr, dst_stride, d0, d1, d2, d3);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 4);
do {
uint8x16_t s0 = vld1q_u8(src_ptr);
int16x4_t d0 = convolve12_4_2d_h(s0, x_filter, correction, range_limit,
permute_tbl);
// Store 4 elements to avoid additional branches. (Safe if w == 2.)
vst1_s16(dst_ptr, d0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
} else {
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0[2], s1[2], s2[2], s3[2];
load_u8_16x4(s, src_stride, &s0[0], &s1[0], &s2[0], &s3[0]);
load_u8_16x4(s + 4, src_stride, &s0[1], &s1[1], &s2[1], &s3[1]);
int16x8_t d0 = convolve12_8_2d_h(s0[0], s0[1], x_filter, correction,
range_limit, permute_tbl);
int16x8_t d1 = convolve12_8_2d_h(s1[0], s1[1], x_filter, correction,
range_limit, permute_tbl);
int16x8_t d2 = convolve12_8_2d_h(s2[0], s2[1], x_filter, correction,
range_limit, permute_tbl);
int16x8_t d3 = convolve12_8_2d_h(s3[0], s3[1], x_filter, correction,
range_limit, permute_tbl);
store_s16_8x4(d, dst_stride, d0, d1, d2, d3);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 4);
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0[2];
s0[0] = vld1q_u8(s);
s0[1] = vld1q_u8(s + 4);
int16x8_t d0 = convolve12_8_2d_h(s0[0], s0[1], x_filter, correction,
range_limit, permute_tbl);
vst1q_s16(d, d0);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
}
}
}
#else // !(AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD))
static INLINE int16x4_t
convolve12_4_2d_h(const int16x4_t s0, const int16x4_t s1, const int16x4_t s2,
const int16x4_t s3, const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7, const int16x4_t s8,
const int16x4_t s9, const int16x4_t s10, const int16x4_t s11,
const int16x8_t x_filter_0_7, const int16x4_t x_filter_8_11,
const int32x4_t horiz_const) {
const int16x4_t x_filter_0_3 = vget_low_s16(x_filter_0_7);
const int16x4_t x_filter_4_7 = vget_high_s16(x_filter_0_7);
int32x4_t sum = horiz_const;
sum = vmlal_lane_s16(sum, s0, x_filter_0_3, 0);
sum = vmlal_lane_s16(sum, s1, x_filter_0_3, 1);
sum = vmlal_lane_s16(sum, s2, x_filter_0_3, 2);
sum = vmlal_lane_s16(sum, s3, x_filter_0_3, 3);
sum = vmlal_lane_s16(sum, s4, x_filter_4_7, 0);
sum = vmlal_lane_s16(sum, s5, x_filter_4_7, 1);
sum = vmlal_lane_s16(sum, s6, x_filter_4_7, 2);
sum = vmlal_lane_s16(sum, s7, x_filter_4_7, 3);
sum = vmlal_lane_s16(sum, s8, x_filter_8_11, 0);
sum = vmlal_lane_s16(sum, s9, x_filter_8_11, 1);
sum = vmlal_lane_s16(sum, s10, x_filter_8_11, 2);
sum = vmlal_lane_s16(sum, s11, x_filter_8_11, 3);
return vshrn_n_s32(sum, ROUND0_BITS);
}
static INLINE void convolve_2d_sr_horiz_12tap_neon(
const uint8_t *src_ptr, int src_stride, int16_t *dst_ptr,
const int dst_stride, int w, int h, const int16x8_t x_filter_0_7,
const int16x4_t x_filter_8_11) {
const int bd = 8;
// A shim of 1 << (ROUND0_BITS - 1) enables us to use non-rounding shifts -
// which are generally faster than rounding shifts on modern CPUs.
const int32x4_t horiz_const =
vdupq_n_s32((1 << (bd + FILTER_BITS - 1)) + (1 << (ROUND0_BITS - 1)));
#if AOM_ARCH_AARCH64
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
uint8x8_t t0, t1, t2, t3;
load_u8_8x4(s, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s1 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s2 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s3 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t s4 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s5 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s6 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s7 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
load_u8_8x4(s + 8, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s8 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s9 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s10 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
s += 11;
do {
load_u8_8x4(s, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s11 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s12 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s13 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s14 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t d0 =
convolve12_4_2d_h(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
x_filter_0_7, x_filter_8_11, horiz_const);
int16x4_t d1 =
convolve12_4_2d_h(s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12,
x_filter_0_7, x_filter_8_11, horiz_const);
int16x4_t d2 =
convolve12_4_2d_h(s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13,
x_filter_0_7, x_filter_8_11, horiz_const);
int16x4_t d3 =
convolve12_4_2d_h(s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13, s14,
x_filter_0_7, x_filter_8_11, horiz_const);
transpose_s16_4x4d(&d0, &d1, &d2, &d3);
// Store 4 elements per row to avoid additional branches. This is safe if
// the actual block width is < 4 because the intermediate buffer is large
// enough to accommodate 128x128 blocks.
store_s16_4x4(d, dst_stride, d0, d1, d2, d3);
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s7 = s11;
s8 = s12;
s9 = s13;
s10 = s14;
s += 4;
d += 4;
width -= 4;
} while (width > 0);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 4);
#endif // AOM_ARCH_AARCH64
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t t0 = vld1q_u8(s);
int16x8_t tt0 = vreinterpretq_s16_u16(vmovl_u8(vget_low_u8(t0)));
int16x8_t tt1 = vreinterpretq_s16_u16(vmovl_u8(vget_high_u8(t0)));
int16x4_t s0 = vget_low_s16(tt0);
int16x4_t s4 = vget_high_s16(tt0);
int16x4_t s8 = vget_low_s16(tt1);
int16x4_t s12 = vget_high_s16(tt1);
int16x4_t s1 = vext_s16(s0, s4, 1); // a1 a2 a3 a4
int16x4_t s2 = vext_s16(s0, s4, 2); // a2 a3 a4 a5
int16x4_t s3 = vext_s16(s0, s4, 3); // a3 a4 a5 a6
int16x4_t s5 = vext_s16(s4, s8, 1); // a5 a6 a7 a8
int16x4_t s6 = vext_s16(s4, s8, 2); // a6 a7 a8 a9
int16x4_t s7 = vext_s16(s4, s8, 3); // a7 a8 a9 a10
int16x4_t s9 = vext_s16(s8, s12, 1); // a9 a10 a11 a12
int16x4_t s10 = vext_s16(s8, s12, 2); // a10 a11 a12 a13
int16x4_t s11 = vext_s16(s8, s12, 3); // a11 a12 a13 a14
int16x4_t d0 =
convolve12_4_2d_h(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
x_filter_0_7, x_filter_8_11, horiz_const);
// Store 4 elements to avoid additional branches. (Safe as noted above.)
vst1_s16(d, d0);
s += 4;
d += 4;
width -= 4;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--h != 0);
}
#endif // AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD)
#if AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_MATMUL_INT8)
static INLINE int16x4_t convolve4_4_2d_h(uint8x16_t samples,
const int8x8_t filters,
const uint8x16_t permute_tbl,
const int32x4_t horiz_const) {
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
uint8x16_t permuted_samples = vqtbl1q_u8(samples, permute_tbl);
// First 4 output values.
int32x4_t sum = vusdotq_lane_s32(horiz_const, permuted_samples, filters, 0);
// We halved the convolution filter values so -1 from the right shift.
return vshrn_n_s32(sum, ROUND0_BITS - 1);
}
static INLINE int16x8_t convolve8_8_2d_h(uint8x16_t samples,
const int8x8_t filters,
const uint8x16x3_t permute_tbl,
const int32x4_t horiz_const) {
uint8x16_t permuted_samples[3];
int32x4_t sum[2];
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_u8(samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_u8(samples, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_u8(samples, permute_tbl.val[2]);
// First 4 output values.
sum[0] = vusdotq_lane_s32(horiz_const, permuted_samples[0], filters, 0);
sum[0] = vusdotq_lane_s32(sum[0], permuted_samples[1], filters, 1);
// Second 4 output values.
sum[1] = vusdotq_lane_s32(horiz_const, permuted_samples[1], filters, 0);
sum[1] = vusdotq_lane_s32(sum[1], permuted_samples[2], filters, 1);
// Narrow and re-pack.
// We halved the convolution filter values so -1 from the right shift.
return vcombine_s16(vshrn_n_s32(sum[0], ROUND0_BITS - 1),
vshrn_n_s32(sum[1], ROUND0_BITS - 1));
}
static INLINE void convolve_2d_sr_horiz_neon(const uint8_t *src, int src_stride,
int16_t *im_block, int im_stride,
int w, int im_h,
const int16_t *x_filter_ptr) {
const int bd = 8;
// This shim of 1 << ((ROUND0_BITS - 1) - 1) enables us to use non-rounding
// shifts - which are generally faster than rounding shifts on modern CPUs.
// The outermost -1 is needed because we halved the filter values.
const int32x4_t horiz_const = vdupq_n_s32((1 << (bd + FILTER_BITS - 2)) +
(1 << ((ROUND0_BITS - 1) - 1)));
const uint8_t *src_ptr = src;
int16_t *dst_ptr = im_block;
int dst_stride = im_stride;
int height = im_h;
if (w <= 4) {
const uint8x16_t permute_tbl = vld1q_u8(dot_prod_permute_tbl);
// 4-tap filters are used for blocks having width <= 4.
// Filter values are even, so halve to reduce intermediate precision reqs.
const int8x8_t x_filter =
vshrn_n_s16(vcombine_s16(vld1_s16(x_filter_ptr + 2), vdup_n_s16(0)), 1);
src_ptr += 2;
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(src_ptr, src_stride, &s0, &s1, &s2, &s3);
int16x4_t d0 = convolve4_4_2d_h(s0, x_filter, permute_tbl, horiz_const);
int16x4_t d1 = convolve4_4_2d_h(s1, x_filter, permute_tbl, horiz_const);
int16x4_t d2 = convolve4_4_2d_h(s2, x_filter, permute_tbl, horiz_const);
int16x4_t d3 = convolve4_4_2d_h(s3, x_filter, permute_tbl, horiz_const);
// Store 4 elements per row to avoid additional branches. This is safe if
// the actual block width is < 4 because the intermediate buffer is large
// enough to accommodate 128x128 blocks.
store_s16_4x4(dst_ptr, dst_stride, d0, d1, d2, d3);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
height -= 4;
} while (height > 4);
do {
uint8x16_t s0 = vld1q_u8(src_ptr);
int16x4_t d0 = convolve4_4_2d_h(s0, x_filter, permute_tbl, horiz_const);
// Store 4 elements to avoid additional branches. (Safe if w == 2.)
vst1_s16(dst_ptr, d0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--height != 0);
} else {
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
// Filter values are even, so halve to reduce intermediate precision reqs.
const int8x8_t x_filter = vshrn_n_s16(vld1q_s16(x_filter_ptr), 1);
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(s, src_stride, &s0, &s1, &s2, &s3);
int16x8_t d0 = convolve8_8_2d_h(s0, x_filter, permute_tbl, horiz_const);
int16x8_t d1 = convolve8_8_2d_h(s1, x_filter, permute_tbl, horiz_const);
int16x8_t d2 = convolve8_8_2d_h(s2, x_filter, permute_tbl, horiz_const);
int16x8_t d3 = convolve8_8_2d_h(s3, x_filter, permute_tbl, horiz_const);
store_s16_8x4(d, dst_stride, d0, d1, d2, d3);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
height -= 4;
} while (height > 4);
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0 = vld1q_u8(s);
int16x8_t d0 = convolve8_8_2d_h(s0, x_filter, permute_tbl, horiz_const);
vst1q_s16(d, d0);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--height != 0);
}
}
#elif AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD)
static INLINE int16x4_t convolve4_4_2d_h(uint8x16_t samples,
const int8x8_t filters,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16_t permute_tbl) {
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
int8x16_t clamped_samples =
vreinterpretq_s8_u8(vsubq_u8(samples, range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
int8x16_t permuted_samples = vqtbl1q_s8(clamped_samples, permute_tbl);
// Accumulate dot product into 'correction' to account for range clamp.
int32x4_t sum = vdotq_lane_s32(correction, permuted_samples, filters, 0);
// We halved the convolution filter values so -1 from the right shift.
return vshrn_n_s32(sum, ROUND0_BITS - 1);
}
static INLINE int16x8_t convolve8_8_2d_h(uint8x16_t samples,
const int8x8_t filters,
const int32x4_t correction,
const uint8x16_t range_limit,
const uint8x16x3_t permute_tbl) {
int8x16_t clamped_samples, permuted_samples[3];
int32x4_t sum[2];
// Clamp sample range to [-128, 127] for 8-bit signed dot product.
clamped_samples = vreinterpretq_s8_u8(vsubq_u8(samples, range_limit));
// Permute samples ready for dot product.
// { 0, 1, 2, 3, 1, 2, 3, 4, 2, 3, 4, 5, 3, 4, 5, 6 }
permuted_samples[0] = vqtbl1q_s8(clamped_samples, permute_tbl.val[0]);
// { 4, 5, 6, 7, 5, 6, 7, 8, 6, 7, 8, 9, 7, 8, 9, 10 }
permuted_samples[1] = vqtbl1q_s8(clamped_samples, permute_tbl.val[1]);
// { 8, 9, 10, 11, 9, 10, 11, 12, 10, 11, 12, 13, 11, 12, 13, 14 }
permuted_samples[2] = vqtbl1q_s8(clamped_samples, permute_tbl.val[2]);
// Accumulate dot product into 'correction' to account for range clamp.
// First 4 output values.
sum[0] = vdotq_lane_s32(correction, permuted_samples[0], filters, 0);
sum[0] = vdotq_lane_s32(sum[0], permuted_samples[1], filters, 1);
// Second 4 output values.
sum[1] = vdotq_lane_s32(correction, permuted_samples[1], filters, 0);
sum[1] = vdotq_lane_s32(sum[1], permuted_samples[2], filters, 1);
// Narrow and re-pack.
// We halved the convolution filter values so -1 from the right shift.
return vcombine_s16(vshrn_n_s32(sum[0], ROUND0_BITS - 1),
vshrn_n_s32(sum[1], ROUND0_BITS - 1));
}
static INLINE void convolve_2d_sr_horiz_neon(const uint8_t *src, int src_stride,
int16_t *im_block, int im_stride,
int w, int im_h,
const int16_t *x_filter_ptr) {
const int bd = 8;
// This shim of 1 << ((ROUND0_BITS - 1) - 1) enables us to use non-rounding
// shifts - which are generally faster than rounding shifts on modern CPUs.
// The outermost -1 is needed because we halved the filter values.
const int32_t horiz_const =
((1 << (bd + FILTER_BITS - 2)) + (1 << ((ROUND0_BITS - 1) - 1)));
// Dot product constants.
const int16x8_t x_filter_s16 = vld1q_s16(x_filter_ptr);
const int32_t correction_s32 =
vaddlvq_s16(vshlq_n_s16(x_filter_s16, FILTER_BITS - 1));
const int32x4_t correction = vdupq_n_s32(correction_s32 + horiz_const);
const uint8x16_t range_limit = vdupq_n_u8(128);
const uint8_t *src_ptr = src;
int16_t *dst_ptr = im_block;
int dst_stride = im_stride;
int height = im_h;
if (w <= 4) {
const uint8x16_t permute_tbl = vld1q_u8(dot_prod_permute_tbl);
// 4-tap filters are used for blocks having width <= 4.
// Filter values are even, so halve to reduce intermediate precision reqs.
const int8x8_t x_filter =
vshrn_n_s16(vcombine_s16(vld1_s16(x_filter_ptr + 2), vdup_n_s16(0)), 1);
src_ptr += 2;
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(src_ptr, src_stride, &s0, &s1, &s2, &s3);
int16x4_t d0 =
convolve4_4_2d_h(s0, x_filter, correction, range_limit, permute_tbl);
int16x4_t d1 =
convolve4_4_2d_h(s1, x_filter, correction, range_limit, permute_tbl);
int16x4_t d2 =
convolve4_4_2d_h(s2, x_filter, correction, range_limit, permute_tbl);
int16x4_t d3 =
convolve4_4_2d_h(s3, x_filter, correction, range_limit, permute_tbl);
// Store 4 elements per row to avoid additional branches. This is safe if
// the actual block width is < 4 because the intermediate buffer is large
// enough to accommodate 128x128 blocks.
store_s16_4x4(dst_ptr, dst_stride, d0, d1, d2, d3);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
height -= 4;
} while (height > 4);
do {
uint8x16_t s0 = vld1q_u8(src_ptr);
int16x4_t d0 =
convolve4_4_2d_h(s0, x_filter, correction, range_limit, permute_tbl);
// Store 4 elements to avoid additional branches. (Safe if w == 2.)
vst1_s16(dst_ptr, d0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--height != 0);
} else {
const uint8x16x3_t permute_tbl = vld1q_u8_x3(dot_prod_permute_tbl);
// Filter values are even, so halve to reduce intermediate precision reqs.
const int8x8_t x_filter = vshrn_n_s16(x_filter_s16, 1);
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0, s1, s2, s3;
load_u8_16x4(s, src_stride, &s0, &s1, &s2, &s3);
int16x8_t d0 = convolve8_8_2d_h(s0, x_filter, correction, range_limit,
permute_tbl);
int16x8_t d1 = convolve8_8_2d_h(s1, x_filter, correction, range_limit,
permute_tbl);
int16x8_t d2 = convolve8_8_2d_h(s2, x_filter, correction, range_limit,
permute_tbl);
int16x8_t d3 = convolve8_8_2d_h(s3, x_filter, correction, range_limit,
permute_tbl);
store_s16_8x4(d, dst_stride, d0, d1, d2, d3);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
height -= 4;
} while (height >= 4);
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
do {
uint8x16_t s0 = vld1q_u8(s);
int16x8_t d0 = convolve8_8_2d_h(s0, x_filter, correction, range_limit,
permute_tbl);
vst1q_s16(d, d0);
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--height != 0);
}
}
#else // !(AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD))
static INLINE int16x4_t convolve4_4_2d_h(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t filter,
const int16x4_t horiz_const) {
int16x4_t sum = horiz_const;
sum = vmla_lane_s16(sum, s0, filter, 0);
sum = vmla_lane_s16(sum, s1, filter, 1);
sum = vmla_lane_s16(sum, s2, filter, 2);
sum = vmla_lane_s16(sum, s3, filter, 3);
// We halved the convolution filter values so -1 from the right shift.
return vshr_n_s16(sum, ROUND0_BITS - 1);
}
static INLINE int16x8_t convolve8_8_2d_h(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t s6, const int16x8_t s7,
const int16x8_t filter,
const int16x8_t horiz_const) {
const int16x4_t filter_lo = vget_low_s16(filter);
const int16x4_t filter_hi = vget_high_s16(filter);
int16x8_t sum = horiz_const;
sum = vmlaq_lane_s16(sum, s0, filter_lo, 0);
sum = vmlaq_lane_s16(sum, s1, filter_lo, 1);
sum = vmlaq_lane_s16(sum, s2, filter_lo, 2);
sum = vmlaq_lane_s16(sum, s3, filter_lo, 3);
sum = vmlaq_lane_s16(sum, s4, filter_hi, 0);
sum = vmlaq_lane_s16(sum, s5, filter_hi, 1);
sum = vmlaq_lane_s16(sum, s6, filter_hi, 2);
sum = vmlaq_lane_s16(sum, s7, filter_hi, 3);
// We halved the convolution filter values so -1 from the right shift.
return vshrq_n_s16(sum, ROUND0_BITS - 1);
}
static INLINE void convolve_2d_sr_horiz_neon(const uint8_t *src, int src_stride,
int16_t *im_block, int im_stride,
int w, int im_h,
const int16_t *x_filter_ptr) {
const int bd = 8;
const uint8_t *src_ptr = src;
int16_t *dst_ptr = im_block;
int dst_stride = im_stride;
int height = im_h;
if (w <= 4) {
// A shim of 1 << ((ROUND0_BITS - 1) - 1) enables us to use non-rounding
// shifts - which are generally faster than rounding shifts on modern CPUs.
// (The extra -1 is needed because we halved the filter values.)
const int16x4_t horiz_const = vdup_n_s16((1 << (bd + FILTER_BITS - 2)) +
(1 << ((ROUND0_BITS - 1) - 1)));
// 4-tap filters are used for blocks having width <= 4.
// Filter values are even, so halve to reduce intermediate precision reqs.
const int16x4_t x_filter = vshr_n_s16(vld1_s16(x_filter_ptr + 2), 1);
src_ptr += 2;
#if AOM_ARCH_AARCH64
do {
uint8x8_t t0, t1, t2, t3;
load_u8_8x4(src_ptr, src_stride, &t0, &t1, &t2, &t3);
transpose_u8_8x4(&t0, &t1, &t2, &t3);
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s1 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s2 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t s3 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t3)));
int16x4_t s4 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s5 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t1)));
int16x4_t s6 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t2)));
int16x4_t d0 = convolve4_4_2d_h(s0, s1, s2, s3, x_filter, horiz_const);
int16x4_t d1 = convolve4_4_2d_h(s1, s2, s3, s4, x_filter, horiz_const);
int16x4_t d2 = convolve4_4_2d_h(s2, s3, s4, s5, x_filter, horiz_const);
int16x4_t d3 = convolve4_4_2d_h(s3, s4, s5, s6, x_filter, horiz_const);
transpose_s16_4x4d(&d0, &d1, &d2, &d3);
// Store 4 elements per row to avoid additional branches. This is safe if
// the actual block width is < 4 because the intermediate buffer is large
// enough to accommodate 128x128 blocks.
store_s16_4x4(dst_ptr, dst_stride, d0, d1, d2, d3);
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
height -= 4;
} while (height > 4);
#endif // AOM_ARCH_AARCH64
do {
uint8x8_t t0 = vld1_u8(src_ptr); // a0 a1 a2 a3 a4 a5 a6 a7
int16x4_t s0 = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s4 = vget_high_s16(vreinterpretq_s16_u16(vmovl_u8(t0)));
int16x4_t s1 = vext_s16(s0, s4, 1); // a1 a2 a3 a4
int16x4_t s2 = vext_s16(s0, s4, 2); // a2 a3 a4 a5
int16x4_t s3 = vext_s16(s0, s4, 3); // a3 a4 a5 a6
int16x4_t d0 = convolve4_4_2d_h(s0, s1, s2, s3, x_filter, horiz_const);
// Store 4 elements to avoid additional branches. (Safe if w == 2.)
vst1_s16(dst_ptr, d0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--height != 0);
} else {
// A shim of 1 << ((ROUND0_BITS - 1) - 1) enables us to use non-rounding
// shifts - which are generally faster than rounding shifts on modern CPUs.
// (The extra -1 is needed because we halved the filter values.)
const int16x8_t horiz_const = vdupq_n_s16((1 << (bd + FILTER_BITS - 2)) +
(1 << ((ROUND0_BITS - 1) - 1)));
// Filter values are even, so halve to reduce intermediate precision reqs.
const int16x8_t x_filter = vshrq_n_s16(vld1q_s16(x_filter_ptr), 1);
#if AOM_ARCH_AARCH64
while (height > 8) {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
uint8x8_t t0, t1, t2, t3, t4, t5, t6, t7;
load_u8_8x8(s, src_stride, &t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7);
transpose_u8_8x8(&t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7);
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
int16x8_t s1 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s2 = vreinterpretq_s16_u16(vmovl_u8(t2));
int16x8_t s3 = vreinterpretq_s16_u16(vmovl_u8(t3));
int16x8_t s4 = vreinterpretq_s16_u16(vmovl_u8(t4));
int16x8_t s5 = vreinterpretq_s16_u16(vmovl_u8(t5));
int16x8_t s6 = vreinterpretq_s16_u16(vmovl_u8(t6));
s += 7;
do {
load_u8_8x8(s, src_stride, &t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7);
transpose_u8_8x8(&t0, &t1, &t2, &t3, &t4, &t5, &t6, &t7);
int16x8_t s7 = vreinterpretq_s16_u16(vmovl_u8(t0));
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s9 = vreinterpretq_s16_u16(vmovl_u8(t2));
int16x8_t s10 = vreinterpretq_s16_u16(vmovl_u8(t3));
int16x8_t s11 = vreinterpretq_s16_u16(vmovl_u8(t4));
int16x8_t s12 = vreinterpretq_s16_u16(vmovl_u8(t5));
int16x8_t s13 = vreinterpretq_s16_u16(vmovl_u8(t6));
int16x8_t s14 = vreinterpretq_s16_u16(vmovl_u8(t7));
int16x8_t d0 = convolve8_8_2d_h(s0, s1, s2, s3, s4, s5, s6, s7,
x_filter, horiz_const);
int16x8_t d1 = convolve8_8_2d_h(s1, s2, s3, s4, s5, s6, s7, s8,
x_filter, horiz_const);
int16x8_t d2 = convolve8_8_2d_h(s2, s3, s4, s5, s6, s7, s8, s9,
x_filter, horiz_const);
int16x8_t d3 = convolve8_8_2d_h(s3, s4, s5, s6, s7, s8, s9, s10,
x_filter, horiz_const);
int16x8_t d4 = convolve8_8_2d_h(s4, s5, s6, s7, s8, s9, s10, s11,
x_filter, horiz_const);
int16x8_t d5 = convolve8_8_2d_h(s5, s6, s7, s8, s9, s10, s11, s12,
x_filter, horiz_const);
int16x8_t d6 = convolve8_8_2d_h(s6, s7, s8, s9, s10, s11, s12, s13,
x_filter, horiz_const);
int16x8_t d7 = convolve8_8_2d_h(s7, s8, s9, s10, s11, s12, s13, s14,
x_filter, horiz_const);
transpose_s16_8x8(&d0, &d1, &d2, &d3, &d4, &d5, &d6, &d7);
store_s16_8x8(d, dst_stride, d0, d1, d2, d3, d4, d5, d6, d7);
s0 = s8;
s1 = s9;
s2 = s10;
s3 = s11;
s4 = s12;
s5 = s13;
s6 = s14;
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += 8 * src_stride;
dst_ptr += 8 * dst_stride;
height -= 8;
}
#endif // AOM_ARCH_AARCH64
do {
const uint8_t *s = src_ptr;
int16_t *d = dst_ptr;
int width = w;
uint8x8_t t0 = vld1_u8(s); // a0 a1 a2 a3 a4 a5 a6 a7
int16x8_t s0 = vreinterpretq_s16_u16(vmovl_u8(t0));
do {
uint8x8_t t1 = vld1_u8(s + 8); // a8 a9 a10 a11 a12 a13 a14 a15
int16x8_t s8 = vreinterpretq_s16_u16(vmovl_u8(t1));
int16x8_t s1 = vextq_s16(s0, s8, 1); // a1 a2 a3 a4 a5 a6 a7 a8
int16x8_t s2 = vextq_s16(s0, s8, 2); // a2 a3 a4 a5 a6 a7 a8 a9
int16x8_t s3 = vextq_s16(s0, s8, 3); // a3 a4 a5 a6 a7 a8 a9 a10
int16x8_t s4 = vextq_s16(s0, s8, 4); // a4 a5 a6 a7 a8 a9 a10 a11
int16x8_t s5 = vextq_s16(s0, s8, 5); // a5 a6 a7 a8 a9 a10 a11 a12
int16x8_t s6 = vextq_s16(s0, s8, 6); // a6 a7 a8 a9 a10 a11 a12 a13
int16x8_t s7 = vextq_s16(s0, s8, 7); // a7 a8 a9 a10 a11 a12 a13 a14
int16x8_t d0 = convolve8_8_2d_h(s0, s1, s2, s3, s4, s5, s6, s7,
x_filter, horiz_const);
vst1q_s16(d, d0);
s0 = s8;
s += 8;
d += 8;
width -= 8;
} while (width > 0);
src_ptr += src_stride;
dst_ptr += dst_stride;
} while (--height != 0);
}
}
#endif // AOM_ARCH_AARCH64 && defined(__ARM_FEATURE_DOTPROD)
static INLINE int32x4_t
convolve12_4_2d_v(const int16x4_t s0, const int16x4_t s1, const int16x4_t s2,
const int16x4_t s3, const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7, const int16x4_t s8,
const int16x4_t s9, const int16x4_t s10, const int16x4_t s11,
const int16x8_t y_filter_0_7, const int16x4_t y_filter_8_11) {
const int16x4_t y_filter_0_3 = vget_low_s16(y_filter_0_7);
const int16x4_t y_filter_4_7 = vget_high_s16(y_filter_0_7);
int32x4_t sum = vmull_lane_s16(s0, y_filter_0_3, 0);
sum = vmlal_lane_s16(sum, s1, y_filter_0_3, 1);
sum = vmlal_lane_s16(sum, s2, y_filter_0_3, 2);
sum = vmlal_lane_s16(sum, s3, y_filter_0_3, 3);
sum = vmlal_lane_s16(sum, s4, y_filter_4_7, 0);
sum = vmlal_lane_s16(sum, s5, y_filter_4_7, 1);
sum = vmlal_lane_s16(sum, s6, y_filter_4_7, 2);
sum = vmlal_lane_s16(sum, s7, y_filter_4_7, 3);
sum = vmlal_lane_s16(sum, s8, y_filter_8_11, 0);
sum = vmlal_lane_s16(sum, s9, y_filter_8_11, 1);
sum = vmlal_lane_s16(sum, s10, y_filter_8_11, 2);
sum = vmlal_lane_s16(sum, s11, y_filter_8_11, 3);
return sum;
}
static INLINE uint8x8_t
convolve12_8_2d_v(const int16x8_t s0, const int16x8_t s1, const int16x8_t s2,
const int16x8_t s3, const int16x8_t s4, const int16x8_t s5,
const int16x8_t s6, const int16x8_t s7, const int16x8_t s8,
const int16x8_t s9, const int16x8_t s10, const int16x8_t s11,
const int16x8_t y_filter_0_7, const int16x4_t y_filter_8_11,
const int16x8_t sub_const) {
const int16x4_t y_filter_0_3 = vget_low_s16(y_filter_0_7);
const int16x4_t y_filter_4_7 = vget_high_s16(y_filter_0_7);
int32x4_t sum0 = vmull_lane_s16(vget_low_s16(s0), y_filter_0_3, 0);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s1), y_filter_0_3, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s2), y_filter_0_3, 2);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s3), y_filter_0_3, 3);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s4), y_filter_4_7, 0);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s5), y_filter_4_7, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s6), y_filter_4_7, 2);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s7), y_filter_4_7, 3);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s8), y_filter_8_11, 0);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s9), y_filter_8_11, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s10), y_filter_8_11, 2);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s11), y_filter_8_11, 3);
int32x4_t sum1 = vmull_lane_s16(vget_high_s16(s0), y_filter_0_3, 0);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s1), y_filter_0_3, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s2), y_filter_0_3, 2);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s3), y_filter_0_3, 3);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s4), y_filter_4_7, 0);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s5), y_filter_4_7, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s6), y_filter_4_7, 2);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s7), y_filter_4_7, 3);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s8), y_filter_8_11, 0);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s9), y_filter_8_11, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s10), y_filter_8_11, 2);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s11), y_filter_8_11, 3);
int16x8_t res =
vcombine_s16(vqrshrn_n_s32(sum0, 2 * FILTER_BITS - ROUND0_BITS),
vqrshrn_n_s32(sum1, 2 * FILTER_BITS - ROUND0_BITS));
res = vsubq_s16(res, sub_const);
return vqmovun_s16(res);
}
static INLINE void convolve_2d_sr_vert_12tap_neon(
int16_t *src_ptr, int src_stride, uint8_t *dst_ptr, int dst_stride, int w,
int h, const int16x8_t y_filter_0_7, const int16x4_t y_filter_8_11) {
const int bd = 8;
const int16x8_t sub_const = vdupq_n_s16(1 << (bd - 1));
if (w <= 4) {
int16x4_t s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10;
load_s16_4x11(src_ptr, src_stride, &s0, &s1, &s2, &s3, &s4, &s5, &s6, &s7,
&s8, &s9, &s10);
src_ptr += 11 * src_stride;
do {
int16x4_t s11, s12, s13, s14;
load_s16_4x4(src_ptr, src_stride, &s11, &s12, &s13, &s14);
int32x4_t d0 = convolve12_4_2d_v(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9,
s10, s11, y_filter_0_7, y_filter_8_11);
int32x4_t d1 = convolve12_4_2d_v(s1, s2, s3, s4, s5, s6, s7, s8, s9, s10,
s11, s12, y_filter_0_7, y_filter_8_11);
int32x4_t d2 = convolve12_4_2d_v(s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
s12, s13, y_filter_0_7, y_filter_8_11);
int32x4_t d3 =
convolve12_4_2d_v(s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13, s14,
y_filter_0_7, y_filter_8_11);
int16x8_t dd01 =
vcombine_s16(vqrshrn_n_s32(d0, 2 * FILTER_BITS - ROUND0_BITS),
vqrshrn_n_s32(d1, 2 * FILTER_BITS - ROUND0_BITS));
int16x8_t dd23 =
vcombine_s16(vqrshrn_n_s32(d2, 2 * FILTER_BITS - ROUND0_BITS),
vqrshrn_n_s32(d3, 2 * FILTER_BITS - ROUND0_BITS));
dd01 = vsubq_s16(dd01, sub_const);
dd23 = vsubq_s16(dd23, sub_const);
uint8x8_t d01 = vqmovun_s16(dd01);
uint8x8_t d23 = vqmovun_s16(dd23);
if (w == 2) {
store_u8_2x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_2x1(dst_ptr + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_2x1(dst_ptr + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_4x1(dst_ptr + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_4x1(dst_ptr + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s7 = s11;
s8 = s12;
s9 = s13;
s10 = s14;
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
} while (h > 0);
} else {
do {
int height = h;
int16_t *s = src_ptr;
uint8_t *d = dst_ptr;
int16x8_t s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10;
load_s16_8x11(s, src_stride, &s0, &s1, &s2, &s3, &s4, &s5, &s6, &s7, &s8,
&s9, &s10);
s += 11 * src_stride;
do {
int16x8_t s11, s12, s13, s14;
load_s16_8x4(s, src_stride, &s11, &s12, &s13, &s14);
uint8x8_t d0 =
convolve12_8_2d_v(s0, s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11,
y_filter_0_7, y_filter_8_11, sub_const);
uint8x8_t d1 =
convolve12_8_2d_v(s1, s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12,
y_filter_0_7, y_filter_8_11, sub_const);
uint8x8_t d2 =
convolve12_8_2d_v(s2, s3, s4, s5, s6, s7, s8, s9, s10, s11, s12,
s13, y_filter_0_7, y_filter_8_11, sub_const);
uint8x8_t d3 =
convolve12_8_2d_v(s3, s4, s5, s6, s7, s8, s9, s10, s11, s12, s13,
s14, y_filter_0_7, y_filter_8_11, sub_const);
if (h != 2) {
store_u8_8x4(d, dst_stride, d0, d1, d2, d3);
} else {
store_u8_8x2(d, dst_stride, d0, d1);
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s7 = s11;
s8 = s12;
s9 = s13;
s10 = s14;
s += 4 * src_stride;
d += 4 * dst_stride;
height -= 4;
} while (height > 0);
src_ptr += 8;
dst_ptr += 8;
w -= 8;
} while (w > 0);
}
}
static INLINE int16x4_t convolve8_4_2d_v(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x4_t s6, const int16x4_t s7,
const int16x8_t y_filter) {
const int16x4_t y_filter_lo = vget_low_s16(y_filter);
const int16x4_t y_filter_hi = vget_high_s16(y_filter);
int32x4_t sum = vmull_lane_s16(s0, y_filter_lo, 0);
sum = vmlal_lane_s16(sum, s1, y_filter_lo, 1);
sum = vmlal_lane_s16(sum, s2, y_filter_lo, 2);
sum = vmlal_lane_s16(sum, s3, y_filter_lo, 3);
sum = vmlal_lane_s16(sum, s4, y_filter_hi, 0);
sum = vmlal_lane_s16(sum, s5, y_filter_hi, 1);
sum = vmlal_lane_s16(sum, s6, y_filter_hi, 2);
sum = vmlal_lane_s16(sum, s7, y_filter_hi, 3);
return vqrshrn_n_s32(sum, 2 * FILTER_BITS - ROUND0_BITS);
}
static INLINE uint8x8_t convolve8_8_2d_v(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t s6, const int16x8_t s7,
const int16x8_t y_filter,
const int16x8_t sub_const) {
const int16x4_t y_filter_lo = vget_low_s16(y_filter);
const int16x4_t y_filter_hi = vget_high_s16(y_filter);
int32x4_t sum0 = vmull_lane_s16(vget_low_s16(s0), y_filter_lo, 0);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s1), y_filter_lo, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s2), y_filter_lo, 2);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s3), y_filter_lo, 3);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s4), y_filter_hi, 0);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s5), y_filter_hi, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s6), y_filter_hi, 2);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s7), y_filter_hi, 3);
int32x4_t sum1 = vmull_lane_s16(vget_high_s16(s0), y_filter_lo, 0);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s1), y_filter_lo, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s2), y_filter_lo, 2);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s3), y_filter_lo, 3);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s4), y_filter_hi, 0);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s5), y_filter_hi, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s6), y_filter_hi, 2);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s7), y_filter_hi, 3);
int16x8_t res =
vcombine_s16(vqrshrn_n_s32(sum0, 2 * FILTER_BITS - ROUND0_BITS),
vqrshrn_n_s32(sum1, 2 * FILTER_BITS - ROUND0_BITS));
res = vsubq_s16(res, sub_const);
return vqmovun_s16(res);
}
static INLINE void convolve_2d_sr_vert_8tap_neon(int16_t *src_ptr,
int src_stride,
uint8_t *dst_ptr,
int dst_stride, int w, int h,
const int16x8_t y_filter) {
const int bd = 8;
const int16x8_t sub_const = vdupq_n_s16(1 << (bd - 1));
if (w <= 4) {
int16x4_t s0, s1, s2, s3, s4, s5, s6;
load_s16_4x7(src_ptr, src_stride, &s0, &s1, &s2, &s3, &s4, &s5, &s6);
src_ptr += 7 * src_stride;
do {
#if AOM_ARCH_AARCH64
int16x4_t s7, s8, s9, s10;
load_s16_4x4(src_ptr, src_stride, &s7, &s8, &s9, &s10);
int16x4_t d0 = convolve8_4_2d_v(s0, s1, s2, s3, s4, s5, s6, s7, y_filter);
int16x4_t d1 = convolve8_4_2d_v(s1, s2, s3, s4, s5, s6, s7, s8, y_filter);
int16x4_t d2 = convolve8_4_2d_v(s2, s3, s4, s5, s6, s7, s8, s9, y_filter);
int16x4_t d3 =
convolve8_4_2d_v(s3, s4, s5, s6, s7, s8, s9, s10, y_filter);
uint8x8_t d01 = vqmovun_s16(vsubq_s16(vcombine_s16(d0, d1), sub_const));
uint8x8_t d23 = vqmovun_s16(vsubq_s16(vcombine_s16(d2, d3), sub_const));
if (w == 2) {
store_u8_2x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_2x1(dst_ptr + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_2x1(dst_ptr + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_4x1(dst_ptr + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_4x1(dst_ptr + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
#else // !AOM_ARCH_AARCH64
int16x4_t s7 = vld1_s16(src_ptr);
int16x4_t d0 = convolve8_4_2d_v(s0, s1, s2, s3, s4, s5, s6, s7, y_filter);
uint8x8_t d01 =
vqmovun_s16(vsubq_s16(vcombine_s16(d0, vdup_n_s16(0)), sub_const));
if (w == 2) {
store_u8_2x1(dst_ptr, d01, 0);
} else {
store_u8_4x1(dst_ptr, d01, 0);
}
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
s5 = s6;
s6 = s7;
src_ptr += src_stride;
dst_ptr += dst_stride;
h--;
#endif // AOM_ARCH_AARCH64
} while (h > 0);
} else {
// Width is a multiple of 8 and height is a multiple of 4.
do {
int height = h;
int16_t *s = src_ptr;
uint8_t *d = dst_ptr;
int16x8_t s0, s1, s2, s3, s4, s5, s6;
load_s16_8x7(s, src_stride, &s0, &s1, &s2, &s3, &s4, &s5, &s6);
s += 7 * src_stride;
do {
#if AOM_ARCH_AARCH64
int16x8_t s7, s8, s9, s10;
load_s16_8x4(s, src_stride, &s7, &s8, &s9, &s10);
uint8x8_t d0 = convolve8_8_2d_v(s0, s1, s2, s3, s4, s5, s6, s7,
y_filter, sub_const);
uint8x8_t d1 = convolve8_8_2d_v(s1, s2, s3, s4, s5, s6, s7, s8,
y_filter, sub_const);
uint8x8_t d2 = convolve8_8_2d_v(s2, s3, s4, s5, s6, s7, s8, s9,
y_filter, sub_const);
uint8x8_t d3 = convolve8_8_2d_v(s3, s4, s5, s6, s7, s8, s9, s10,
y_filter, sub_const);
if (h != 2) {
store_u8_8x4(d, dst_stride, d0, d1, d2, d3);
} else {
store_u8_8x2(d, dst_stride, d0, d1);
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s5 = s9;
s6 = s10;
s += 4 * src_stride;
d += 4 * dst_stride;
height -= 4;
#else // !AOM_ARCH_AARCH64
int16x8_t s7 = vld1q_s16(s);
uint8x8_t d0 = convolve8_8_2d_v(s0, s1, s2, s3, s4, s5, s6, s7,
y_filter, sub_const);
vst1_u8(d, d0);
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
s5 = s6;
s6 = s7;
s += src_stride;
d += dst_stride;
height--;
#endif // AOM_ARCH_AARCH64
} while (height > 0);
src_ptr += 8;
dst_ptr += 8;
w -= 8;
} while (w > 0);
}
}
static INLINE int16x4_t convolve6_4_2d_v(const int16x4_t s0, const int16x4_t s1,
const int16x4_t s2, const int16x4_t s3,
const int16x4_t s4, const int16x4_t s5,
const int16x8_t y_filter) {
const int16x4_t y_filter_lo = vget_low_s16(y_filter);
const int16x4_t y_filter_hi = vget_high_s16(y_filter);
int32x4_t sum = vmull_lane_s16(s0, y_filter_lo, 1);
sum = vmlal_lane_s16(sum, s1, y_filter_lo, 2);
sum = vmlal_lane_s16(sum, s2, y_filter_lo, 3);
sum = vmlal_lane_s16(sum, s3, y_filter_hi, 0);
sum = vmlal_lane_s16(sum, s4, y_filter_hi, 1);
sum = vmlal_lane_s16(sum, s5, y_filter_hi, 2);
return vqrshrn_n_s32(sum, 2 * FILTER_BITS - ROUND0_BITS);
}
static INLINE uint8x8_t convolve6_8_2d_v(const int16x8_t s0, const int16x8_t s1,
const int16x8_t s2, const int16x8_t s3,
const int16x8_t s4, const int16x8_t s5,
const int16x8_t y_filter,
const int16x8_t sub_const) {
const int16x4_t y_filter_lo = vget_low_s16(y_filter);
const int16x4_t y_filter_hi = vget_high_s16(y_filter);
int32x4_t sum0 = vmull_lane_s16(vget_low_s16(s0), y_filter_lo, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s1), y_filter_lo, 2);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s2), y_filter_lo, 3);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s3), y_filter_hi, 0);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s4), y_filter_hi, 1);
sum0 = vmlal_lane_s16(sum0, vget_low_s16(s5), y_filter_hi, 2);
int32x4_t sum1 = vmull_lane_s16(vget_high_s16(s0), y_filter_lo, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s1), y_filter_lo, 2);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s2), y_filter_lo, 3);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s3), y_filter_hi, 0);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s4), y_filter_hi, 1);
sum1 = vmlal_lane_s16(sum1, vget_high_s16(s5), y_filter_hi, 2);
int16x8_t res =
vcombine_s16(vqrshrn_n_s32(sum0, 2 * FILTER_BITS - ROUND0_BITS),
vqrshrn_n_s32(sum1, 2 * FILTER_BITS - ROUND0_BITS));
res = vsubq_s16(res, sub_const);
return vqmovun_s16(res);
}
static INLINE void convolve_2d_sr_vert_6tap_neon(int16_t *src_ptr,
int src_stride,
uint8_t *dst_ptr,
int dst_stride, int w, int h,
const int16x8_t y_filter) {
const int bd = 8;
const int16x8_t sub_const = vdupq_n_s16(1 << (bd - 1));
if (w <= 4) {
int16x4_t s0, s1, s2, s3, s4;
load_s16_4x5(src_ptr, src_stride, &s0, &s1, &s2, &s3, &s4);
src_ptr += 5 * src_stride;
do {
#if AOM_ARCH_AARCH64
int16x4_t s5, s6, s7, s8;
load_s16_4x4(src_ptr, src_stride, &s5, &s6, &s7, &s8);
int16x4_t d0 = convolve6_4_2d_v(s0, s1, s2, s3, s4, s5, y_filter);
int16x4_t d1 = convolve6_4_2d_v(s1, s2, s3, s4, s5, s6, y_filter);
int16x4_t d2 = convolve6_4_2d_v(s2, s3, s4, s5, s6, s7, y_filter);
int16x4_t d3 = convolve6_4_2d_v(s3, s4, s5, s6, s7, s8, y_filter);
uint8x8_t d01 = vqmovun_s16(vsubq_s16(vcombine_s16(d0, d1), sub_const));
uint8x8_t d23 = vqmovun_s16(vsubq_s16(vcombine_s16(d2, d3), sub_const));
if (w == 2) {
store_u8_2x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_2x1(dst_ptr + 1 * dst_stride, d01, 2);
if (h != 2) {
store_u8_2x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_2x1(dst_ptr + 3 * dst_stride, d23, 2);
}
} else {
store_u8_4x1(dst_ptr + 0 * dst_stride, d01, 0);
store_u8_4x1(dst_ptr + 1 * dst_stride, d01, 1);
if (h != 2) {
store_u8_4x1(dst_ptr + 2 * dst_stride, d23, 0);
store_u8_4x1(dst_ptr + 3 * dst_stride, d23, 1);
}
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
src_ptr += 4 * src_stride;
dst_ptr += 4 * dst_stride;
h -= 4;
#else // !AOM_ARCH_AARCH64
int16x4_t s5 = vld1_s16(src_ptr);
int16x4_t d0 = convolve6_4_2d_v(s0, s1, s2, s3, s4, s5, y_filter);
uint8x8_t d01 =
vqmovun_s16(vsubq_s16(vcombine_s16(d0, vdup_n_s16(0)), sub_const));
if (w == 2) {
store_u8_2x1(dst_ptr, d01, 0);
} else {
store_u8_4x1(dst_ptr, d01, 0);
}
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
src_ptr += src_stride;
dst_ptr += dst_stride;
h--;
#endif // AOM_ARCH_AARCH64
} while (h > 0);
} else {
// Width is a multiple of 8 and height is a multiple of 4.
do {
int height = h;
int16_t *s = src_ptr;
uint8_t *d = dst_ptr;
int16x8_t s0, s1, s2, s3, s4;
load_s16_8x5(s, src_stride, &s0, &s1, &s2, &s3, &s4);
s += 5 * src_stride;
do {
#if AOM_ARCH_AARCH64
int16x8_t s5, s6, s7, s8;
load_s16_8x4(s, src_stride, &s5, &s6, &s7, &s8);
uint8x8_t d0 =
convolve6_8_2d_v(s0, s1, s2, s3, s4, s5, y_filter, sub_const);
uint8x8_t d1 =
convolve6_8_2d_v(s1, s2, s3, s4, s5, s6, y_filter, sub_const);
uint8x8_t d2 =
convolve6_8_2d_v(s2, s3, s4, s5, s6, s7, y_filter, sub_const);
uint8x8_t d3 =
convolve6_8_2d_v(s3, s4, s5, s6, s7, s8, y_filter, sub_const);
if (h != 2) {
store_u8_8x4(d, dst_stride, d0, d1, d2, d3);
} else {
store_u8_8x2(d, dst_stride, d0, d1);
}
s0 = s4;
s1 = s5;
s2 = s6;
s3 = s7;
s4 = s8;
s += 4 * src_stride;
d += 4 * dst_stride;
height -= 4;
#else // !AOM_ARCH_AARCH64
int16x8_t s5 = vld1q_s16(s);
uint8x8_t d0 =
convolve6_8_2d_v(s0, s1, s2, s3, s4, s5, y_filter, sub_const);
vst1_u8(d, d0);
s0 = s1;
s1 = s2;
s2 = s3;
s3 = s4;
s4 = s5;
s += src_stride;
d += dst_stride;
height--;
#endif // AOM_ARCH_AARCH64
} while (height > 0);
src_ptr += 8;
dst_ptr += 8;
w -= 8;
} while (w > 0);
}
}
void av1_convolve_2d_sr_neon(const uint8_t *src, int src_stride, uint8_t *dst,
int dst_stride, int w, int h,
const InterpFilterParams *filter_params_x,
const InterpFilterParams *filter_params_y,
const int subpel_x_qn, const int subpel_y_qn,
ConvolveParams *conv_params) {
(void)conv_params;
const int y_filter_taps = get_filter_tap(filter_params_y, subpel_y_qn);
const int clamped_y_taps = y_filter_taps < 6 ? 6 : y_filter_taps;
const int im_h = h + clamped_y_taps - 1;
const int im_stride = MAX_SB_SIZE;
const int vert_offset = clamped_y_taps / 2 - 1;
const int horiz_offset = filter_params_x->taps / 2 - 1;
const uint8_t *src_ptr = src - vert_offset * src_stride - horiz_offset;
const int16_t *x_filter_ptr = av1_get_interp_filter_subpel_kernel(
filter_params_x, subpel_x_qn & SUBPEL_MASK);
const int16_t *y_filter_ptr = av1_get_interp_filter_subpel_kernel(
filter_params_y, subpel_y_qn & SUBPEL_MASK);
if (filter_params_x->taps > 8) {
DECLARE_ALIGNED(16, int16_t,
im_block[(MAX_SB_SIZE + MAX_FILTER_TAP - 1) * MAX_SB_SIZE]);
const int16x8_t x_filter_0_7 = vld1q_s16(x_filter_ptr);
const int16x4_t x_filter_8_11 = vld1_s16(x_filter_ptr + 8);
const int16x8_t y_filter_0_7 = vld1q_s16(y_filter_ptr);
const int16x4_t y_filter_8_11 = vld1_s16(y_filter_ptr + 8);
convolve_2d_sr_horiz_12tap_neon(src_ptr, src_stride, im_block, im_stride, w,
im_h, x_filter_0_7, x_filter_8_11);
convolve_2d_sr_vert_12tap_neon(im_block, im_stride, dst, dst_stride, w, h,
y_filter_0_7, y_filter_8_11);
} else {
DECLARE_ALIGNED(16, int16_t,
im_block[(MAX_SB_SIZE + HORIZ_EXTRA_ROWS) * MAX_SB_SIZE]);
convolve_2d_sr_horiz_neon(src_ptr, src_stride, im_block, im_stride, w, im_h,
x_filter_ptr);
const int16x8_t y_filter = vld1q_s16(y_filter_ptr);
if (clamped_y_taps <= 6) {
convolve_2d_sr_vert_6tap_neon(im_block, im_stride, dst, dst_stride, w, h,
y_filter);
} else {
convolve_2d_sr_vert_8tap_neon(im_block, im_stride, dst, dst_stride, w, h,
y_filter);
}
}
}
static INLINE void scaledconvolve_horiz_w4(
const uint8_t *src, const ptrdiff_t src_stride, uint8_t *dst,
const ptrdiff_t dst_stride, const InterpKernel *const x_filters,
const int x0_q4, const int x_step_q4, const int w, const int h) {
DECLARE_ALIGNED(16, uint8_t, temp[4 * 4]);
int x, y, z;
src -= SUBPEL_TAPS / 2 - 1;
y = h;
do {
int x_q4 = x0_q4;
x = 0;
do {
// process 4 src_x steps
for (z = 0; z < 4; ++z) {
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
if (x_q4 & SUBPEL_MASK) {
const int16x8_t filters = vld1q_s16(x_filters[x_q4 & SUBPEL_MASK]);
uint8x8_t s[8], d;
int16x8_t ss[4];
int16x4_t t[8], tt;
load_u8_8x4(src_x, src_stride, &s[0], &s[1], &s[2], &s[3]);
transpose_u8_8x4(&s[0], &s[1], &s[2], &s[3]);
ss[0] = vreinterpretq_s16_u16(vmovl_u8(s[0]));
ss[1] = vreinterpretq_s16_u16(vmovl_u8(s[1]));
ss[2] = vreinterpretq_s16_u16(vmovl_u8(s[2]));
ss[3] = vreinterpretq_s16_u16(vmovl_u8(s[3]));
t[0] = vget_low_s16(ss[0]);
t[1] = vget_low_s16(ss[1]);
t[2] = vget_low_s16(ss[2]);
t[3] = vget_low_s16(ss[3]);
t[4] = vget_high_s16(ss[0]);
t[5] = vget_high_s16(ss[1]);
t[6] = vget_high_s16(ss[2]);
t[7] = vget_high_s16(ss[3]);
tt = convolve8_4(t[0], t[1], t[2], t[3], t[4], t[5], t[6], t[7],
filters);
d = vqrshrun_n_s16(vcombine_s16(tt, tt), 7);
store_u8_4x1(&temp[4 * z], d, 0);
} else {
int i;
for (i = 0; i < 4; ++i) {
temp[z * 4 + i] = src_x[i * src_stride + 3];
}
}
x_q4 += x_step_q4;
}
// transpose the 4x4 filters values back to dst
{
const uint8x8x4_t d4 = vld4_u8(temp);
store_u8_4x1(&dst[x + 0 * dst_stride], d4.val[0], 0);
store_u8_4x1(&dst[x + 1 * dst_stride], d4.val[1], 0);
store_u8_4x1(&dst[x + 2 * dst_stride], d4.val[2], 0);
store_u8_4x1(&dst[x + 3 * dst_stride], d4.val[3], 0);
}
x += 4;
} while (x < w);
src += src_stride * 4;
dst += dst_stride * 4;
y -= 4;
} while (y > 0);
}
static INLINE void scaledconvolve_horiz_w8(
const uint8_t *src, const ptrdiff_t src_stride, uint8_t *dst,
const ptrdiff_t dst_stride, const InterpKernel *const x_filters,
const int x0_q4, const int x_step_q4, const int w, const int h) {
DECLARE_ALIGNED(16, uint8_t, temp[8 * 8]);
int x, y, z;
src -= SUBPEL_TAPS / 2 - 1;
// This function processes 8x8 areas. The intermediate height is not always
// a multiple of 8, so force it to be a multiple of 8 here.
y = (h + 7) & ~7;
do {
int x_q4 = x0_q4;
x = 0;
do {
uint8x8_t d[8];
// process 8 src_x steps
for (z = 0; z < 8; ++z) {
const uint8_t *const src_x = &src[x_q4 >> SUBPEL_BITS];
if (x_q4 & SUBPEL_MASK) {
const int16x8_t filters = vld1q_s16(x_filters[x_q4 & SUBPEL_MASK]);
uint8x8_t s[8];
load_u8_8x8(src_x, src_stride, &s[0], &s[1], &s[2], &s[3], &s[4],
&s[5], &s[6], &s[7]);
transpose_u8_8x8(&s[0], &s[1], &s[2], &s[3], &s[4], &s[5], &s[6],
&s[7]);
d[0] = scale_filter_8(s, filters);
vst1_u8(&temp[8 * z], d[0]);
} else {
int i;
for (i = 0; i < 8; ++i) {
temp[z * 8 + i] = src_x[i * src_stride + 3];
}
}
x_q4 += x_step_q4;
}
// transpose the 8x8 filters values back to dst
load_u8_8x8(temp, 8, &d[0], &d[1], &d[2], &d[3], &d[4], &d[5], &d[6],
&d[7]);
transpose_u8_8x8(&d[0], &d[1], &d[2], &d[3], &d[4], &d[5], &d[6], &d[7]);
store_u8_8x8(dst + x, dst_stride, d[0], d[1], d[2], d[3], d[4], d[5],
d[6], d[7]);
x += 8;
} while (x < w);
src += src_stride * 8;
dst += dst_stride * 8;
} while (y -= 8);
}
static INLINE void scaledconvolve_vert_w4(
const uint8_t *src, const ptrdiff_t src_stride, uint8_t *dst,
const ptrdiff_t dst_stride, const InterpKernel *const y_filters,
const int y0_q4, const int y_step_q4, const int w, const int h) {
int y;
int y_q4 = y0_q4;
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
y = h;
do {
const unsigned char *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
if (y_q4 & SUBPEL_MASK) {
const int16x8_t filters = vld1q_s16(y_filters[y_q4 & SUBPEL_MASK]);
uint8x8_t s[8], d;
int16x4_t t[8], tt;
load_u8_8x8(src_y, src_stride, &s[0], &s[1], &s[2], &s[3], &s[4], &s[5],
&s[6], &s[7]);
t[0] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[0])));
t[1] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[1])));
t[2] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[2])));
t[3] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[3])));
t[4] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[4])));
t[5] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[5])));
t[6] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[6])));
t[7] = vget_low_s16(vreinterpretq_s16_u16(vmovl_u8(s[7])));
tt = convolve8_4(t[0], t[1], t[2], t[3], t[4], t[5], t[6], t[7], filters);
d = vqrshrun_n_s16(vcombine_s16(tt, tt), 7);
store_u8_4x1(dst, d, 0);
} else {
memcpy(dst, &src_y[3 * src_stride], w);
}
dst += dst_stride;
y_q4 += y_step_q4;
} while (--y);
}
static INLINE void scaledconvolve_vert_w8(
const uint8_t *src, const ptrdiff_t src_stride, uint8_t *dst,
const ptrdiff_t dst_stride, const InterpKernel *const y_filters,
const int y0_q4, const int y_step_q4, const int w, const int h) {
int y;
int y_q4 = y0_q4;
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
y = h;
do {
const unsigned char *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
if (y_q4 & SUBPEL_MASK) {
const int16x8_t filters = vld1q_s16(y_filters[y_q4 & SUBPEL_MASK]);
uint8x8_t s[8], d;
load_u8_8x8(src_y, src_stride, &s[0], &s[1], &s[2], &s[3], &s[4], &s[5],
&s[6], &s[7]);
d = scale_filter_8(s, filters);
vst1_u8(dst, d);
} else {
memcpy(dst, &src_y[3 * src_stride], w);
}
dst += dst_stride;
y_q4 += y_step_q4;
} while (--y);
}
static INLINE void scaledconvolve_vert_w16(
const uint8_t *src, const ptrdiff_t src_stride, uint8_t *dst,
const ptrdiff_t dst_stride, const InterpKernel *const y_filters,
const int y0_q4, const int y_step_q4, const int w, const int h) {
int x, y;
int y_q4 = y0_q4;
src -= src_stride * (SUBPEL_TAPS / 2 - 1);
y = h;
do {
const unsigned char *src_y = &src[(y_q4 >> SUBPEL_BITS) * src_stride];
if (y_q4 & SUBPEL_MASK) {
x = 0;
do {
const int16x8_t filters = vld1q_s16(y_filters[y_q4 & SUBPEL_MASK]);
uint8x16_t ss[8];
uint8x8_t s[8], d[2];
load_u8_16x8(src_y, src_stride, &ss[0], &ss[1], &ss[2], &ss[3], &ss[4],
&ss[5], &ss[6], &ss[7]);
s[0] = vget_low_u8(ss[0]);
s[1] = vget_low_u8(ss[1]);
s[2] = vget_low_u8(ss[2]);
s[3] = vget_low_u8(ss[3]);
s[4] = vget_low_u8(ss[4]);
s[5] = vget_low_u8(ss[5]);
s[6] = vget_low_u8(ss[6]);
s[7] = vget_low_u8(ss[7]);
d[0] = scale_filter_8(s, filters);
s[0] = vget_high_u8(ss[0]);
s[1] = vget_high_u8(ss[1]);
s[2] = vget_high_u8(ss[2]);
s[3] = vget_high_u8(ss[3]);
s[4] = vget_high_u8(ss[4]);
s[5] = vget_high_u8(ss[5]);
s[6] = vget_high_u8(ss[6]);
s[7] = vget_high_u8(ss[7]);
d[1] = scale_filter_8(s, filters);
vst1q_u8(&dst[x], vcombine_u8(d[0], d[1]));
src_y += 16;
x += 16;
} while (x < w);
} else {
memcpy(dst, &src_y[3 * src_stride], w);
}
dst += dst_stride;
y_q4 += y_step_q4;
} while (--y);
}
void aom_scaled_2d_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst,
ptrdiff_t dst_stride, const InterpKernel *filter,
int x0_q4, int x_step_q4, int y0_q4, int y_step_q4,
int w, int h) {
// Note: Fixed size intermediate buffer, temp, places limits on parameters.
// 2d filtering proceeds in 2 steps:
// (1) Interpolate horizontally into an intermediate buffer, temp.
// (2) Interpolate temp vertically to derive the sub-pixel result.
// Deriving the maximum number of rows in the temp buffer (135):
// --Smallest scaling factor is x1/2 ==> y_step_q4 = 32 (Normative).
// --Largest block size is 64x64 pixels.
// --64 rows in the downscaled frame span a distance of (64 - 1) * 32 in the
// original frame (in 1/16th pixel units).
// --Must round-up because block may be located at sub-pixel position.
// --Require an additional SUBPEL_TAPS rows for the 8-tap filter tails.
// --((64 - 1) * 32 + 15) >> 4 + 8 = 135.
// --Require an additional 8 rows for the horiz_w8 transpose tail.
// When calling in frame scaling function, the smallest scaling factor is x1/4
// ==> y_step_q4 = 64. Since w and h are at most 16, the temp buffer is still
// big enough.
DECLARE_ALIGNED(16, uint8_t, temp[(135 + 8) * 64]);
const int intermediate_height =
(((h - 1) * y_step_q4 + y0_q4) >> SUBPEL_BITS) + SUBPEL_TAPS;
assert(w <= 64);
assert(h <= 64);
assert(y_step_q4 <= 32 || (y_step_q4 <= 64 && h <= 32));
assert(x_step_q4 <= 64);
if (w >= 8) {
scaledconvolve_horiz_w8(src - src_stride * (SUBPEL_TAPS / 2 - 1),
src_stride, temp, 64, filter, x0_q4, x_step_q4, w,
intermediate_height);
} else {
scaledconvolve_horiz_w4(src - src_stride * (SUBPEL_TAPS / 2 - 1),
src_stride, temp, 64, filter, x0_q4, x_step_q4, w,
intermediate_height);
}
if (w >= 16) {
scaledconvolve_vert_w16(temp + 64 * (SUBPEL_TAPS / 2 - 1), 64, dst,
dst_stride, filter, y0_q4, y_step_q4, w, h);
} else if (w == 8) {
scaledconvolve_vert_w8(temp + 64 * (SUBPEL_TAPS / 2 - 1), 64, dst,
dst_stride, filter, y0_q4, y_step_q4, w, h);
} else {
scaledconvolve_vert_w4(temp + 64 * (SUBPEL_TAPS / 2 - 1), 64, dst,
dst_stride, filter, y0_q4, y_step_q4, w, h);
}
}