xref: /aosp_15_r20/external/libvpx/vpx_dsp/arm/idct4x4_1_add_neon.c (revision fb1b10ab9aebc7c7068eedab379b749d7e3900be)
1 /*
2  *  Copyright (c) 2014 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 <arm_neon.h>
12 #include <assert.h>
13 
14 #include "./vpx_dsp_rtcd.h"
15 #include "vpx_dsp/arm/mem_neon.h"
16 #include "vpx_dsp/inv_txfm.h"
17 
idct4x4_1_add_kernel(uint8_t ** dest,const int stride,const int16x8_t res,uint32x2_t * const d)18 static INLINE void idct4x4_1_add_kernel(uint8_t **dest, const int stride,
19                                         const int16x8_t res,
20                                         uint32x2_t *const d) {
21   uint16x8_t a;
22   uint8x8_t b;
23   *d = vld1_lane_u32((const uint32_t *)*dest, *d, 0);
24   *d = vld1_lane_u32((const uint32_t *)(*dest + stride), *d, 1);
25   a = vaddw_u8(vreinterpretq_u16_s16(res), vreinterpret_u8_u32(*d));
26   b = vqmovun_s16(vreinterpretq_s16_u16(a));
27   vst1_lane_u32((uint32_t *)*dest, vreinterpret_u32_u8(b), 0);
28   *dest += stride;
29   vst1_lane_u32((uint32_t *)*dest, vreinterpret_u32_u8(b), 1);
30   *dest += stride;
31 }
32 
vpx_idct4x4_1_add_neon(const tran_low_t * input,uint8_t * dest,int stride)33 void vpx_idct4x4_1_add_neon(const tran_low_t *input, uint8_t *dest,
34                             int stride) {
35   const int16_t out0 =
36       WRAPLOW(dct_const_round_shift((int16_t)input[0] * cospi_16_64));
37   const int16_t out1 = WRAPLOW(dct_const_round_shift(out0 * cospi_16_64));
38   const int16_t a1 = ROUND_POWER_OF_TWO(out1, 4);
39   const int16x8_t dc = vdupq_n_s16(a1);
40   uint32x2_t d = vdup_n_u32(0);
41 
42   assert(!((intptr_t)dest % sizeof(uint32_t)));
43   assert(!(stride % sizeof(uint32_t)));
44 
45   idct4x4_1_add_kernel(&dest, stride, dc, &d);
46   idct4x4_1_add_kernel(&dest, stride, dc, &d);
47 }
48