1 /*
2 * Copyright (c) 2017 The WebM project authors. All Rights Reserved.
3 *
4 * Use of this source code is governed by a BSD-style license
5 * that can be found in the LICENSE file in the root of the source
6 * tree. An additional intellectual property rights grant can be found
7 * in the file PATENTS. All contributing project authors may
8 * be found in the AUTHORS file in the root of the source tree.
9 */
10
11 #include <smmintrin.h> // SSE4.1
12
13 #include "./vpx_dsp_rtcd.h"
14 #include "vpx_dsp/x86/highbd_inv_txfm_sse2.h"
15 #include "vpx_dsp/x86/highbd_inv_txfm_sse4.h"
16 #include "vpx_dsp/x86/inv_txfm_sse2.h"
17 #include "vpx_dsp/x86/transpose_sse2.h"
18
vpx_highbd_idct4x4_16_add_sse4_1(const tran_low_t * input,uint16_t * dest,int stride,int bd)19 void vpx_highbd_idct4x4_16_add_sse4_1(const tran_low_t *input, uint16_t *dest,
20 int stride, int bd) {
21 __m128i io[4];
22
23 io[0] = _mm_load_si128((const __m128i *)(input + 0));
24 io[1] = _mm_load_si128((const __m128i *)(input + 4));
25 io[2] = _mm_load_si128((const __m128i *)(input + 8));
26 io[3] = _mm_load_si128((const __m128i *)(input + 12));
27
28 if (bd == 8) {
29 __m128i io_short[2];
30
31 io_short[0] = _mm_packs_epi32(io[0], io[1]);
32 io_short[1] = _mm_packs_epi32(io[2], io[3]);
33 idct4_sse2(io_short);
34 idct4_sse2(io_short);
35 io_short[0] = _mm_add_epi16(io_short[0], _mm_set1_epi16(8));
36 io_short[1] = _mm_add_epi16(io_short[1], _mm_set1_epi16(8));
37 io[0] = _mm_srai_epi16(io_short[0], 4);
38 io[1] = _mm_srai_epi16(io_short[1], 4);
39 } else {
40 highbd_idct4_sse4_1(io);
41 highbd_idct4_sse4_1(io);
42 io[0] = wraplow_16bit_shift4(io[0], io[1], _mm_set1_epi32(8));
43 io[1] = wraplow_16bit_shift4(io[2], io[3], _mm_set1_epi32(8));
44 }
45
46 recon_and_store_4x4(io, dest, stride, bd);
47 }
48