summaryrefslogtreecommitdiffstats
path: root/third_party/aom/av1/common/mips
diff options
context:
space:
mode:
authortrav90 <travawine@palemoon.org>2018-10-15 21:45:30 -0500
committertrav90 <travawine@palemoon.org>2018-10-15 21:45:30 -0500
commit68569dee1416593955c1570d638b3d9250b33012 (patch)
treed960f017cd7eba3f125b7e8a813789ee2e076310 /third_party/aom/av1/common/mips
parent07c17b6b98ed32fcecff15c083ab0fd878de3cf0 (diff)
downloadUXP-68569dee1416593955c1570d638b3d9250b33012.tar
UXP-68569dee1416593955c1570d638b3d9250b33012.tar.gz
UXP-68569dee1416593955c1570d638b3d9250b33012.tar.lz
UXP-68569dee1416593955c1570d638b3d9250b33012.tar.xz
UXP-68569dee1416593955c1570d638b3d9250b33012.zip
Import aom library
This is the reference implementation for the Alliance for Open Media's av1 video code. The commit used was 4d668d7feb1f8abd809d1bca0418570a7f142a36.
Diffstat (limited to 'third_party/aom/av1/common/mips')
-rw-r--r--third_party/aom/av1/common/mips/dspr2/av1_itrans16_dspr2.c97
-rw-r--r--third_party/aom/av1/common/mips/dspr2/av1_itrans4_dspr2.c91
-rw-r--r--third_party/aom/av1/common/mips/dspr2/av1_itrans8_dspr2.c85
-rw-r--r--third_party/aom/av1/common/mips/msa/av1_idct16x16_msa.c80
-rw-r--r--third_party/aom/av1/common/mips/msa/av1_idct4x4_msa.c61
-rw-r--r--third_party/aom/av1/common/mips/msa/av1_idct8x8_msa.c79
6 files changed, 493 insertions, 0 deletions
diff --git a/third_party/aom/av1/common/mips/dspr2/av1_itrans16_dspr2.c b/third_party/aom/av1/common/mips/dspr2/av1_itrans16_dspr2.c
new file mode 100644
index 000000000..79f9338bd
--- /dev/null
+++ b/third_party/aom/av1/common/mips/dspr2/av1_itrans16_dspr2.c
@@ -0,0 +1,97 @@
+/*
+ * Copyright (c) 2016, 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 <stdio.h>
+
+#include "./aom_config.h"
+#include "./av1_rtcd.h"
+#include "av1/common/common.h"
+#include "av1/common/blockd.h"
+#include "av1/common/idct.h"
+#include "aom_dsp/mips/inv_txfm_dspr2.h"
+#include "aom_dsp/txfm_common.h"
+#include "aom_ports/mem.h"
+
+#if HAVE_DSPR2
+void av1_iht16x16_256_add_dspr2(const int16_t *input, uint8_t *dest, int pitch,
+ int tx_type) {
+ int i, j;
+ DECLARE_ALIGNED(32, int16_t, out[16 * 16]);
+ int16_t *outptr = out;
+ int16_t temp_out[16];
+ uint32_t pos = 45;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__("wrdsp %[pos], 1 \n\t" : : [pos] "r"(pos));
+
+ switch (tx_type) {
+ case DCT_DCT: // DCT in both horizontal and vertical
+ idct16_rows_dspr2(input, outptr, 16);
+ idct16_cols_add_blk_dspr2(out, dest, pitch);
+ break;
+ case ADST_DCT: // ADST in vertical, DCT in horizontal
+ idct16_rows_dspr2(input, outptr, 16);
+
+ outptr = out;
+
+ for (i = 0; i < 16; ++i) {
+ iadst16_dspr2(outptr, temp_out);
+
+ for (j = 0; j < 16; ++j)
+ dest[j * pitch + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6) +
+ dest[j * pitch + i]);
+ outptr += 16;
+ }
+ break;
+ case DCT_ADST: // DCT in vertical, ADST in horizontal
+ {
+ int16_t temp_in[16 * 16];
+
+ for (i = 0; i < 16; ++i) {
+ /* prefetch row */
+ prefetch_load((const uint8_t *)(input + 16));
+
+ iadst16_dspr2(input, outptr);
+ input += 16;
+ outptr += 16;
+ }
+
+ for (i = 0; i < 16; ++i)
+ for (j = 0; j < 16; ++j) temp_in[j * 16 + i] = out[i * 16 + j];
+
+ idct16_cols_add_blk_dspr2(temp_in, dest, pitch);
+ } break;
+ case ADST_ADST: // ADST in both directions
+ {
+ int16_t temp_in[16];
+
+ for (i = 0; i < 16; ++i) {
+ /* prefetch row */
+ prefetch_load((const uint8_t *)(input + 16));
+
+ iadst16_dspr2(input, outptr);
+ input += 16;
+ outptr += 16;
+ }
+
+ for (i = 0; i < 16; ++i) {
+ for (j = 0; j < 16; ++j) temp_in[j] = out[j * 16 + i];
+ iadst16_dspr2(temp_in, temp_out);
+ for (j = 0; j < 16; ++j)
+ dest[j * pitch + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6) +
+ dest[j * pitch + i]);
+ }
+ } break;
+ default: printf("av1_short_iht16x16_add_dspr2 : Invalid tx_type\n"); break;
+ }
+}
+#endif // #if HAVE_DSPR2
diff --git a/third_party/aom/av1/common/mips/dspr2/av1_itrans4_dspr2.c b/third_party/aom/av1/common/mips/dspr2/av1_itrans4_dspr2.c
new file mode 100644
index 000000000..0a9552376
--- /dev/null
+++ b/third_party/aom/av1/common/mips/dspr2/av1_itrans4_dspr2.c
@@ -0,0 +1,91 @@
+/*
+ * Copyright (c) 2016, 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 <stdio.h>
+
+#include "./aom_config.h"
+#include "./av1_rtcd.h"
+#include "av1/common/common.h"
+#include "av1/common/blockd.h"
+#include "av1/common/idct.h"
+#include "aom_dsp/mips/inv_txfm_dspr2.h"
+#include "aom_dsp/txfm_common.h"
+#include "aom_ports/mem.h"
+
+#if HAVE_DSPR2
+void av1_iht4x4_16_add_dspr2(const int16_t *input, uint8_t *dest,
+ int dest_stride, int tx_type) {
+ int i, j;
+ DECLARE_ALIGNED(32, int16_t, out[4 * 4]);
+ int16_t *outptr = out;
+ int16_t temp_in[4 * 4], temp_out[4];
+ uint32_t pos = 45;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__("wrdsp %[pos], 1 \n\t"
+ :
+ : [pos] "r"(pos));
+
+ switch (tx_type) {
+ case DCT_DCT: // DCT in both horizontal and vertical
+ aom_idct4_rows_dspr2(input, outptr);
+ aom_idct4_columns_add_blk_dspr2(&out[0], dest, dest_stride);
+ break;
+ case ADST_DCT: // ADST in vertical, DCT in horizontal
+ aom_idct4_rows_dspr2(input, outptr);
+
+ outptr = out;
+
+ for (i = 0; i < 4; ++i) {
+ iadst4_dspr2(outptr, temp_out);
+
+ for (j = 0; j < 4; ++j)
+ dest[j * dest_stride + i] = clip_pixel(
+ ROUND_POWER_OF_TWO(temp_out[j], 4) + dest[j * dest_stride + i]);
+
+ outptr += 4;
+ }
+ break;
+ case DCT_ADST: // DCT in vertical, ADST in horizontal
+ for (i = 0; i < 4; ++i) {
+ iadst4_dspr2(input, outptr);
+ input += 4;
+ outptr += 4;
+ }
+
+ for (i = 0; i < 4; ++i) {
+ for (j = 0; j < 4; ++j) {
+ temp_in[i * 4 + j] = out[j * 4 + i];
+ }
+ }
+ aom_idct4_columns_add_blk_dspr2(&temp_in[0], dest, dest_stride);
+ break;
+ case ADST_ADST: // ADST in both directions
+ for (i = 0; i < 4; ++i) {
+ iadst4_dspr2(input, outptr);
+ input += 4;
+ outptr += 4;
+ }
+
+ for (i = 0; i < 4; ++i) {
+ for (j = 0; j < 4; ++j) temp_in[j] = out[j * 4 + i];
+ iadst4_dspr2(temp_in, temp_out);
+
+ for (j = 0; j < 4; ++j)
+ dest[j * dest_stride + i] = clip_pixel(
+ ROUND_POWER_OF_TWO(temp_out[j], 4) + dest[j * dest_stride + i]);
+ }
+ break;
+ default: printf("av1_short_iht4x4_add_dspr2 : Invalid tx_type\n"); break;
+ }
+}
+#endif // #if HAVE_DSPR2
diff --git a/third_party/aom/av1/common/mips/dspr2/av1_itrans8_dspr2.c b/third_party/aom/av1/common/mips/dspr2/av1_itrans8_dspr2.c
new file mode 100644
index 000000000..8bf5b4f0e
--- /dev/null
+++ b/third_party/aom/av1/common/mips/dspr2/av1_itrans8_dspr2.c
@@ -0,0 +1,85 @@
+/*
+ * Copyright (c) 2016, 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 <stdio.h>
+
+#include "./aom_config.h"
+#include "./av1_rtcd.h"
+#include "av1/common/common.h"
+#include "av1/common/blockd.h"
+#include "aom_dsp/mips/inv_txfm_dspr2.h"
+#include "aom_dsp/txfm_common.h"
+#include "aom_ports/mem.h"
+
+#if HAVE_DSPR2
+void av1_iht8x8_64_add_dspr2(const int16_t *input, uint8_t *dest,
+ int dest_stride, int tx_type) {
+ int i, j;
+ DECLARE_ALIGNED(32, int16_t, out[8 * 8]);
+ int16_t *outptr = out;
+ int16_t temp_in[8 * 8], temp_out[8];
+ uint32_t pos = 45;
+
+ /* bit positon for extract from acc */
+ __asm__ __volatile__("wrdsp %[pos], 1 \n\t" : : [pos] "r"(pos));
+
+ switch (tx_type) {
+ case DCT_DCT: // DCT in both horizontal and vertical
+ idct8_rows_dspr2(input, outptr, 8);
+ idct8_columns_add_blk_dspr2(&out[0], dest, dest_stride);
+ break;
+ case ADST_DCT: // ADST in vertical, DCT in horizontal
+ idct8_rows_dspr2(input, outptr, 8);
+
+ for (i = 0; i < 8; ++i) {
+ iadst8_dspr2(&out[i * 8], temp_out);
+
+ for (j = 0; j < 8; ++j)
+ dest[j * dest_stride + i] = clip_pixel(
+ ROUND_POWER_OF_TWO(temp_out[j], 5) + dest[j * dest_stride + i]);
+ }
+ break;
+ case DCT_ADST: // DCT in vertical, ADST in horizontal
+ for (i = 0; i < 8; ++i) {
+ iadst8_dspr2(input, outptr);
+ input += 8;
+ outptr += 8;
+ }
+
+ for (i = 0; i < 8; ++i) {
+ for (j = 0; j < 8; ++j) {
+ temp_in[i * 8 + j] = out[j * 8 + i];
+ }
+ }
+ idct8_columns_add_blk_dspr2(&temp_in[0], dest, dest_stride);
+ break;
+ case ADST_ADST: // ADST in both directions
+ for (i = 0; i < 8; ++i) {
+ iadst8_dspr2(input, outptr);
+ input += 8;
+ outptr += 8;
+ }
+
+ for (i = 0; i < 8; ++i) {
+ for (j = 0; j < 8; ++j) temp_in[j] = out[j * 8 + i];
+
+ iadst8_dspr2(temp_in, temp_out);
+
+ for (j = 0; j < 8; ++j)
+ dest[j * dest_stride + i] = clip_pixel(
+ ROUND_POWER_OF_TWO(temp_out[j], 5) + dest[j * dest_stride + i]);
+ }
+ break;
+ default: printf("av1_short_iht8x8_add_dspr2 : Invalid tx_type\n"); break;
+ }
+}
+#endif // #if HAVE_DSPR2
diff --git a/third_party/aom/av1/common/mips/msa/av1_idct16x16_msa.c b/third_party/aom/av1/common/mips/msa/av1_idct16x16_msa.c
new file mode 100644
index 000000000..4bd0a1635
--- /dev/null
+++ b/third_party/aom/av1/common/mips/msa/av1_idct16x16_msa.c
@@ -0,0 +1,80 @@
+/*
+ * Copyright (c) 2016, 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 "av1/common/enums.h"
+#include "aom_dsp/mips/inv_txfm_msa.h"
+
+void av1_iht16x16_256_add_msa(const int16_t *input, uint8_t *dst,
+ int32_t dst_stride, int32_t tx_type) {
+ int32_t i;
+ DECLARE_ALIGNED(32, int16_t, out[16 * 16]);
+ int16_t *out_ptr = &out[0];
+
+ switch (tx_type) {
+ case DCT_DCT:
+ /* transform rows */
+ for (i = 0; i < 2; ++i) {
+ /* process 16 * 8 block */
+ aom_idct16_1d_rows_msa((input + (i << 7)), (out_ptr + (i << 7)));
+ }
+
+ /* transform columns */
+ for (i = 0; i < 2; ++i) {
+ /* process 8 * 16 block */
+ aom_idct16_1d_columns_addblk_msa((out_ptr + (i << 3)), (dst + (i << 3)),
+ dst_stride);
+ }
+ break;
+ case ADST_DCT:
+ /* transform rows */
+ for (i = 0; i < 2; ++i) {
+ /* process 16 * 8 block */
+ aom_idct16_1d_rows_msa((input + (i << 7)), (out_ptr + (i << 7)));
+ }
+
+ /* transform columns */
+ for (i = 0; i < 2; ++i) {
+ aom_iadst16_1d_columns_addblk_msa((out_ptr + (i << 3)),
+ (dst + (i << 3)), dst_stride);
+ }
+ break;
+ case DCT_ADST:
+ /* transform rows */
+ for (i = 0; i < 2; ++i) {
+ /* process 16 * 8 block */
+ aom_iadst16_1d_rows_msa((input + (i << 7)), (out_ptr + (i << 7)));
+ }
+
+ /* transform columns */
+ for (i = 0; i < 2; ++i) {
+ /* process 8 * 16 block */
+ aom_idct16_1d_columns_addblk_msa((out_ptr + (i << 3)), (dst + (i << 3)),
+ dst_stride);
+ }
+ break;
+ case ADST_ADST:
+ /* transform rows */
+ for (i = 0; i < 2; ++i) {
+ /* process 16 * 8 block */
+ aom_iadst16_1d_rows_msa((input + (i << 7)), (out_ptr + (i << 7)));
+ }
+
+ /* transform columns */
+ for (i = 0; i < 2; ++i) {
+ aom_iadst16_1d_columns_addblk_msa((out_ptr + (i << 3)),
+ (dst + (i << 3)), dst_stride);
+ }
+ break;
+ default: assert(0); break;
+ }
+}
diff --git a/third_party/aom/av1/common/mips/msa/av1_idct4x4_msa.c b/third_party/aom/av1/common/mips/msa/av1_idct4x4_msa.c
new file mode 100644
index 000000000..8364f8dc4
--- /dev/null
+++ b/third_party/aom/av1/common/mips/msa/av1_idct4x4_msa.c
@@ -0,0 +1,61 @@
+/*
+ * Copyright (c) 2016, 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 "av1/common/enums.h"
+#include "aom_dsp/mips/inv_txfm_msa.h"
+
+void av1_iht4x4_16_add_msa(const int16_t *input, uint8_t *dst,
+ int32_t dst_stride, int32_t tx_type) {
+ v8i16 in0, in1, in2, in3;
+
+ /* load vector elements of 4x4 block */
+ LD4x4_SH(input, in0, in1, in2, in3);
+ TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
+
+ switch (tx_type) {
+ case DCT_DCT:
+ /* DCT in horizontal */
+ AOM_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ /* DCT in vertical */
+ TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
+ AOM_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ break;
+ case ADST_DCT:
+ /* DCT in horizontal */
+ AOM_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ /* ADST in vertical */
+ TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
+ AOM_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ break;
+ case DCT_ADST:
+ /* ADST in horizontal */
+ AOM_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ /* DCT in vertical */
+ TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
+ AOM_IDCT4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ break;
+ case ADST_ADST:
+ /* ADST in horizontal */
+ AOM_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ /* ADST in vertical */
+ TRANSPOSE4x4_SH_SH(in0, in1, in2, in3, in0, in1, in2, in3);
+ AOM_IADST4x4(in0, in1, in2, in3, in0, in1, in2, in3);
+ break;
+ default: assert(0); break;
+ }
+
+ /* final rounding (add 2^3, divide by 2^4) and shift */
+ SRARI_H4_SH(in0, in1, in2, in3, 4);
+ /* add block and store 4x4 */
+ ADDBLK_ST4x4_UB(in0, in1, in2, in3, dst, dst_stride);
+}
diff --git a/third_party/aom/av1/common/mips/msa/av1_idct8x8_msa.c b/third_party/aom/av1/common/mips/msa/av1_idct8x8_msa.c
new file mode 100644
index 000000000..71117051b
--- /dev/null
+++ b/third_party/aom/av1/common/mips/msa/av1_idct8x8_msa.c
@@ -0,0 +1,79 @@
+/*
+ * Copyright (c) 2016, 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 "av1/common/enums.h"
+#include "aom_dsp/mips/inv_txfm_msa.h"
+
+void av1_iht8x8_64_add_msa(const int16_t *input, uint8_t *dst,
+ int32_t dst_stride, int32_t tx_type) {
+ v8i16 in0, in1, in2, in3, in4, in5, in6, in7;
+
+ /* load vector elements of 8x8 block */
+ LD_SH8(input, 8, in0, in1, in2, in3, in4, in5, in6, in7);
+
+ TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3,
+ in4, in5, in6, in7);
+
+ switch (tx_type) {
+ case DCT_DCT:
+ /* DCT in horizontal */
+ AOM_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3,
+ in4, in5, in6, in7);
+ /* DCT in vertical */
+ TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2,
+ in3, in4, in5, in6, in7);
+ AOM_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3,
+ in4, in5, in6, in7);
+ break;
+ case ADST_DCT:
+ /* DCT in horizontal */
+ AOM_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3,
+ in4, in5, in6, in7);
+ /* ADST in vertical */
+ TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2,
+ in3, in4, in5, in6, in7);
+ AOM_ADST8(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3, in4,
+ in5, in6, in7);
+ break;
+ case DCT_ADST:
+ /* ADST in horizontal */
+ AOM_ADST8(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3, in4,
+ in5, in6, in7);
+ /* DCT in vertical */
+ TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2,
+ in3, in4, in5, in6, in7);
+ AOM_IDCT8x8_1D(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3,
+ in4, in5, in6, in7);
+ break;
+ case ADST_ADST:
+ /* ADST in horizontal */
+ AOM_ADST8(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3, in4,
+ in5, in6, in7);
+ /* ADST in vertical */
+ TRANSPOSE8x8_SH_SH(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2,
+ in3, in4, in5, in6, in7);
+ AOM_ADST8(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3, in4,
+ in5, in6, in7);
+ break;
+ default: assert(0); break;
+ }
+
+ /* final rounding (add 2^4, divide by 2^5) and shift */
+ SRARI_H4_SH(in0, in1, in2, in3, 5);
+ SRARI_H4_SH(in4, in5, in6, in7, 5);
+
+ /* add block and store 8x8 */
+ AOM_ADDBLK_ST8x4_UB(dst, dst_stride, in0, in1, in2, in3);
+ dst += (4 * dst_stride);
+ AOM_ADDBLK_ST8x4_UB(dst, dst_stride, in4, in5, in6, in7);
+}