2 * Copyright (c) 2014 The WebM project authors. All Rights Reserved.
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.
14 #include "./vpx_dsp_rtcd.h"
15 #include "vpx_dsp/inv_txfm.h"
17 static INLINE void idct4x4_1_add_kernel(uint8_t **dest, const int stride,
19 uint32x2_t *const d) {
22 *d = vld1_lane_u32((const uint32_t *)*dest, *d, 0);
23 *d = vld1_lane_u32((const uint32_t *)(*dest + stride), *d, 1);
24 a = vaddw_u8(vreinterpretq_u16_s16(res), vreinterpret_u8_u32(*d));
25 b = vqmovun_s16(vreinterpretq_s16_u16(a));
26 vst1_lane_u32((uint32_t *)*dest, vreinterpret_u32_u8(b), 0);
28 vst1_lane_u32((uint32_t *)*dest, vreinterpret_u32_u8(b), 1);
32 void vpx_idct4x4_1_add_neon(const tran_low_t *input, uint8_t *dest,
34 const int16_t out0 = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64));
35 const int16_t out1 = WRAPLOW(dct_const_round_shift(out0 * cospi_16_64));
36 const int16_t a1 = ROUND_POWER_OF_TWO(out1, 4);
37 const int16x8_t dc = vdupq_n_s16(a1);
38 uint32x2_t d = vdup_n_u32(0);
40 assert(!((intptr_t)dest % sizeof(uint32_t)));
41 assert(!(stride % sizeof(uint32_t)));
43 idct4x4_1_add_kernel(&dest, stride, dc, &d);
44 idct4x4_1_add_kernel(&dest, stride, dc, &d);