1 /*
2  *  Copyright (c) 2010 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 
12 #include "vpx_config.h"
13 #include "vp8_rtcd.h"
14 #include "vpx_ports/x86.h"
15 #include "vp8/encoder/block.h"
16 
17 void vp8_short_fdct4x4_mmx(short *input, short *output, int pitch);
vp8_short_fdct8x4_mmx(short * input,short * output,int pitch)18 void vp8_short_fdct8x4_mmx(short *input, short *output, int pitch)
19 {
20     vp8_short_fdct4x4_mmx(input,   output,    pitch);
21     vp8_short_fdct4x4_mmx(input + 4, output + 16, pitch);
22 }
23 
24 int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
25                                  short *qcoeff_ptr, short *dequant_ptr,
26                                  const short *scan_mask, short *round_ptr,
27                                  short *quant_ptr, short *dqcoeff_ptr);
vp8_fast_quantize_b_mmx(BLOCK * b,BLOCKD * d)28 void vp8_fast_quantize_b_mmx(BLOCK *b, BLOCKD *d)
29 {
30     const short *scan_mask   = vp8_default_zig_zag_mask;
31     short *coeff_ptr   = b->coeff;
32     short *zbin_ptr    = b->zbin;
33     short *round_ptr   = b->round;
34     short *quant_ptr   = b->quant_fast;
35     short *qcoeff_ptr  = d->qcoeff;
36     short *dqcoeff_ptr = d->dqcoeff;
37     short *dequant_ptr = d->dequant;
38 
39     *d->eob = (char)vp8_fast_quantize_b_impl_mmx(
40                                                  coeff_ptr,
41                                                  zbin_ptr,
42                                                  qcoeff_ptr,
43                                                  dequant_ptr,
44                                                  scan_mask,
45 
46                                                  round_ptr,
47                                                  quant_ptr,
48                                                  dqcoeff_ptr
49                                                  );
50 }
51 
52 int vp8_mbblock_error_mmx_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
vp8_mbblock_error_mmx(MACROBLOCK * mb,int dc)53 int vp8_mbblock_error_mmx(MACROBLOCK *mb, int dc)
54 {
55     short *coeff_ptr =  mb->block[0].coeff;
56     short *dcoef_ptr =  mb->e_mbd.block[0].dqcoeff;
57     return vp8_mbblock_error_mmx_impl(coeff_ptr, dcoef_ptr, dc);
58 }
59 
60 int vp8_mbuverror_mmx_impl(short *s_ptr, short *d_ptr);
vp8_mbuverror_mmx(MACROBLOCK * mb)61 int vp8_mbuverror_mmx(MACROBLOCK *mb)
62 {
63     short *s_ptr = &mb->coeff[256];
64     short *d_ptr = &mb->e_mbd.dqcoeff[256];
65     return vp8_mbuverror_mmx_impl(s_ptr, d_ptr);
66 }
67 
68 void vp8_subtract_b_mmx_impl(unsigned char *z,  int src_stride,
69                              short *diff, unsigned char *predictor,
70                              int pitch);
vp8_subtract_b_mmx(BLOCK * be,BLOCKD * bd,int pitch)71 void vp8_subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
72 {
73     unsigned char *z = *(be->base_src) + be->src;
74     unsigned int  src_stride = be->src_stride;
75     short *diff = &be->src_diff[0];
76     unsigned char *predictor = &bd->predictor[0];
77     vp8_subtract_b_mmx_impl(z, src_stride, diff, predictor, pitch);
78 }
79