Skip to content
GitLab
Select Git revision
  • v1.14.0-linphone default
  • v1.12.0-linphone
  • v1.9.0-linphone
  • v1.7.0-linphone
  • feature/update_to_v1.9.0-linphone
  • feature/uwp_nuget
  • m68-3440
  • m66-3359
  • sandbox/luoyi@google.com/convolve
  • mandarinduck
  • sandbox/wangch@google.com/vp9
  • v1.6.1_linphone
  • linphone-android
  • longtailedduck
  • m56-2924
  • nextgenv2
  • m54-2840
  • sandbox/yaowu@google.com/mergeaom
  • khakicampbell
  • m52-2743
  • v1.9.0
  • v1.9.0-rc1
  • v1.8.2
  • v1.8.1
  • v1.8.0
  • v1.7.0
  • v1.6.1
  • v1.6.0
  • v1.5.0
  • v1.4.0
  • v1.3.0
  • v1.2.0
  • v1.1.0
  • v1.0.0
  • v0.9.7-p1
  • v0.9.7
  • v0.9.6
  • v0.9.5
  • v0.9.2
  • v0.9.1
Find file Blame History Permalink
  • Jingning Han's avatar
    SSSE3 implementation of full inverse 8x8 2D-DCT · 52ae97b6
    Jingning Han authored 
    This commit enables SSSE3 version full inverse 8x8 2D-DCT and
reconstruction. It makes the runtime of vp9_idct8x8_64_add down
from 256 cycles (SSE2) to 246 cycles.

Change-Id: I0600feac894d6a443a3c9d18daf34156d4e225c3
    52ae97b6
vp9_rtcd_defs.pl 44.93 KiB
1 
sub vp9_common_forward_decls() {
2 
print <<EOF
3 
/*
4 
 * VP9
5 
 */
6
7 
#include "vpx/vpx_integer.h"
8 
#include "vp9/common/vp9_enums.h"
9
10 
struct macroblockd;
11
12 
/* Encoder forward decls */
13 
struct macroblock;
14 
struct vp9_variance_vtable;
15 
struct search_site_config;
16 
struct mv;
17 
union int_mv;
18 
struct yv12_buffer_config;
19 
EOF
20 
}
21 
forward_decls qw/vp9_common_forward_decls/;
22
23 
# x86inc.asm doesn't work if pic is enabled on 32 bit platforms so no assembly.
24 
if (vpx_config("CONFIG_USE_X86INC") eq "yes") {
25 
  $mmx_x86inc = 'mmx';
26 
  $sse_x86inc = 'sse';
27 
  $sse2_x86inc = 'sse2';
28 
  $ssse3_x86inc = 'ssse3';
29 
  $avx_x86inc = 'avx';
30 
  $avx2_x86inc = 'avx2';
31 
} else {
32 
  $mmx_x86inc = $sse_x86inc = $sse2_x86inc = $ssse3_x86inc =
33 
  $avx_x86inc = $avx2_x86inc = '';
34 
}
35
36 
# this variable is for functions that are 64 bit only.
37 
if ($opts{arch} eq "x86_64") {
38 
  $mmx_x86_64 = 'mmx';
39 
  $sse2_x86_64 = 'sse2';
40 
  $ssse3_x86_64 = 'ssse3';
41 
  $avx_x86_64 = 'avx';
42 
  $avx2_x86_64 = 'avx2';
43 
} else {
44 
  $mmx_x86_64 = $sse2_x86_64 = $ssse3_x86_64 =
45 
  $avx_x86_64 = $avx2_x86_64 = '';
46 
}
47
48 
#
49 
# RECON
50 
#
51 
add_proto qw/void vp9_d207_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
52 
specialize qw/vp9_d207_predictor_4x4/, "$ssse3_x86inc";
53
54 
add_proto qw/void vp9_d45_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
55 
specialize qw/vp9_d45_predictor_4x4/, "$ssse3_x86inc";
56
57 
add_proto qw/void vp9_d63_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
58 
specialize qw/vp9_d63_predictor_4x4/, "$ssse3_x86inc";
59
60 
add_proto qw/void vp9_h_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
61 
specialize qw/vp9_h_predictor_4x4 neon dspr2/, "$ssse3_x86inc";
62
63 
add_proto qw/void vp9_d117_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
64 
specialize qw/vp9_d117_predictor_4x4/;
65
66 
add_proto qw/void vp9_d135_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
67 
specialize qw/vp9_d135_predictor_4x4/;
68
69 
add_proto qw/void vp9_d153_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left";
70 
specialize qw/vp9_d153_predictor_4x4/, "$ssse3_x86inc";
7172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140
add_proto qw/void vp9_v_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_v_predictor_4x4 neon/, "$sse_x86inc"; add_proto qw/void vp9_tm_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_tm_predictor_4x4 neon dspr2/, "$sse_x86inc"; add_proto qw/void vp9_dc_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_predictor_4x4 dspr2/, "$sse_x86inc"; add_proto qw/void vp9_dc_top_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_top_predictor_4x4/; add_proto qw/void vp9_dc_left_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_left_predictor_4x4/; add_proto qw/void vp9_dc_128_predictor_4x4/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_128_predictor_4x4/; add_proto qw/void vp9_d207_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d207_predictor_8x8/, "$ssse3_x86inc"; add_proto qw/void vp9_d45_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d45_predictor_8x8/, "$ssse3_x86inc"; add_proto qw/void vp9_d63_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d63_predictor_8x8/, "$ssse3_x86inc"; add_proto qw/void vp9_h_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_h_predictor_8x8 neon dspr2/, "$ssse3_x86inc"; add_proto qw/void vp9_d117_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d117_predictor_8x8/; add_proto qw/void vp9_d135_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d135_predictor_8x8/; add_proto qw/void vp9_d153_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d153_predictor_8x8/, "$ssse3_x86inc"; add_proto qw/void vp9_v_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_v_predictor_8x8 neon/, "$sse_x86inc"; add_proto qw/void vp9_tm_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_tm_predictor_8x8 neon dspr2/, "$sse2_x86inc"; add_proto qw/void vp9_dc_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_predictor_8x8 dspr2/, "$sse_x86inc"; add_proto qw/void vp9_dc_top_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_top_predictor_8x8/; add_proto qw/void vp9_dc_left_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_left_predictor_8x8/; add_proto qw/void vp9_dc_128_predictor_8x8/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_128_predictor_8x8/; add_proto qw/void vp9_d207_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d207_predictor_16x16/, "$ssse3_x86inc"; add_proto qw/void vp9_d45_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d45_predictor_16x16/, "$ssse3_x86inc"; add_proto qw/void vp9_d63_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d63_predictor_16x16/, "$ssse3_x86inc"; add_proto qw/void vp9_h_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_h_predictor_16x16 neon dspr2/, "$ssse3_x86inc";
141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210
add_proto qw/void vp9_d117_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d117_predictor_16x16/; add_proto qw/void vp9_d135_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d135_predictor_16x16/; add_proto qw/void vp9_d153_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d153_predictor_16x16/, "$ssse3_x86inc"; add_proto qw/void vp9_v_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_v_predictor_16x16 neon/, "$sse2_x86inc"; add_proto qw/void vp9_tm_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_tm_predictor_16x16 neon/, "$sse2_x86inc"; add_proto qw/void vp9_dc_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_predictor_16x16 dspr2/, "$sse2_x86inc"; add_proto qw/void vp9_dc_top_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_top_predictor_16x16/; add_proto qw/void vp9_dc_left_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_left_predictor_16x16/; add_proto qw/void vp9_dc_128_predictor_16x16/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_128_predictor_16x16/; add_proto qw/void vp9_d207_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d207_predictor_32x32/, "$ssse3_x86inc"; add_proto qw/void vp9_d45_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d45_predictor_32x32/, "$ssse3_x86inc"; add_proto qw/void vp9_d63_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d63_predictor_32x32/, "$ssse3_x86inc"; add_proto qw/void vp9_h_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_h_predictor_32x32 neon/, "$ssse3_x86inc"; add_proto qw/void vp9_d117_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d117_predictor_32x32/; add_proto qw/void vp9_d135_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d135_predictor_32x32/; add_proto qw/void vp9_d153_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_d153_predictor_32x32/; add_proto qw/void vp9_v_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_v_predictor_32x32 neon/, "$sse2_x86inc"; add_proto qw/void vp9_tm_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_tm_predictor_32x32 neon/, "$sse2_x86_64"; add_proto qw/void vp9_dc_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_predictor_32x32/, "$sse2_x86inc"; add_proto qw/void vp9_dc_top_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_top_predictor_32x32/; add_proto qw/void vp9_dc_left_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_left_predictor_32x32/; add_proto qw/void vp9_dc_128_predictor_32x32/, "uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left"; specialize qw/vp9_dc_128_predictor_32x32/; # # Loopfilter # add_proto qw/void vp9_lpf_vertical_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh";
211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280
specialize qw/vp9_lpf_vertical_16 sse2 neon dspr2/; add_proto qw/void vp9_lpf_vertical_16_dual/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh"; specialize qw/vp9_lpf_vertical_16_dual sse2 neon dspr2/; add_proto qw/void vp9_lpf_vertical_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count"; specialize qw/vp9_lpf_vertical_8 sse2 neon dspr2/; add_proto qw/void vp9_lpf_vertical_8_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1"; specialize qw/vp9_lpf_vertical_8_dual sse2 neon dspr2/; add_proto qw/void vp9_lpf_vertical_4/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count"; specialize qw/vp9_lpf_vertical_4 mmx neon dspr2/; add_proto qw/void vp9_lpf_vertical_4_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1"; specialize qw/vp9_lpf_vertical_4_dual sse2 neon dspr2/; add_proto qw/void vp9_lpf_horizontal_16/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count"; specialize qw/vp9_lpf_horizontal_16 sse2 avx2 neon dspr2/; add_proto qw/void vp9_lpf_horizontal_8/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count"; specialize qw/vp9_lpf_horizontal_8 sse2 neon dspr2/; add_proto qw/void vp9_lpf_horizontal_8_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1"; specialize qw/vp9_lpf_horizontal_8_dual sse2 neon dspr2/; add_proto qw/void vp9_lpf_horizontal_4/, "uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh, int count"; specialize qw/vp9_lpf_horizontal_4 mmx neon dspr2/; add_proto qw/void vp9_lpf_horizontal_4_dual/, "uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1"; specialize qw/vp9_lpf_horizontal_4_dual sse2 neon dspr2/; # # post proc # if (vpx_config("CONFIG_VP9_POSTPROC") eq "yes") { add_proto qw/void vp9_mbpost_proc_down/, "uint8_t *dst, int pitch, int rows, int cols, int flimit"; specialize qw/vp9_mbpost_proc_down mmx sse2/; $vp9_mbpost_proc_down_sse2=vp9_mbpost_proc_down_xmm; add_proto qw/void vp9_mbpost_proc_across_ip/, "uint8_t *src, int pitch, int rows, int cols, int flimit"; specialize qw/vp9_mbpost_proc_across_ip sse2/; $vp9_mbpost_proc_across_ip_sse2=vp9_mbpost_proc_across_ip_xmm; add_proto qw/void vp9_post_proc_down_and_across/, "const uint8_t *src_ptr, uint8_t *dst_ptr, int src_pixels_per_line, int dst_pixels_per_line, int rows, int cols, int flimit"; specialize qw/vp9_post_proc_down_and_across mmx sse2/; $vp9_post_proc_down_and_across_sse2=vp9_post_proc_down_and_across_xmm; add_proto qw/void vp9_plane_add_noise/, "uint8_t *Start, char *noise, char blackclamp[16], char whiteclamp[16], char bothclamp[16], unsigned int Width, unsigned int Height, int Pitch"; specialize qw/vp9_plane_add_noise mmx sse2/; $vp9_plane_add_noise_sse2=vp9_plane_add_noise_wmt; } add_proto qw/void vp9_blend_mb_inner/, "uint8_t *y, uint8_t *u, uint8_t *v, int y1, int u1, int v1, int alpha, int stride"; specialize qw/vp9_blend_mb_inner/; add_proto qw/void vp9_blend_mb_outer/, "uint8_t *y, uint8_t *u, uint8_t *v, int y1, int u1, int v1, int alpha, int stride"; specialize qw/vp9_blend_mb_outer/; add_proto qw/void vp9_blend_b/, "uint8_t *y, uint8_t *u, uint8_t *v, int y1, int u1, int v1, int alpha, int stride"; specialize qw/vp9_blend_b/; # # Sub Pixel Filters # add_proto qw/void vp9_convolve_copy/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve_copy neon dspr2/, "$sse2_x86inc"; add_proto qw/void vp9_convolve_avg/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve_avg neon dspr2/, "$sse2_x86inc";
281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350
add_proto qw/void vp9_convolve8/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve8 sse2 ssse3 avx2 neon dspr2/; add_proto qw/void vp9_convolve8_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve8_horiz sse2 ssse3 avx2 neon dspr2/; add_proto qw/void vp9_convolve8_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve8_vert sse2 ssse3 avx2 neon dspr2/; add_proto qw/void vp9_convolve8_avg/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve8_avg sse2 ssse3 neon dspr2/; add_proto qw/void vp9_convolve8_avg_horiz/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve8_avg_horiz sse2 ssse3 neon dspr2/; add_proto qw/void vp9_convolve8_avg_vert/, "const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h"; specialize qw/vp9_convolve8_avg_vert sse2 ssse3 neon dspr2/; # # dct # add_proto qw/void vp9_idct4x4_1_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct4x4_1_add sse2 neon dspr2/; add_proto qw/void vp9_idct4x4_16_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct4x4_16_add sse2 neon dspr2/; add_proto qw/void vp9_idct8x8_1_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct8x8_1_add sse2 neon dspr2/; add_proto qw/void vp9_idct8x8_64_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct8x8_64_add sse2 neon dspr2/, "$ssse3_x86_64"; add_proto qw/void vp9_idct8x8_10_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct8x8_10_add sse2 neon dspr2/; add_proto qw/void vp9_idct16x16_1_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct16x16_1_add sse2 neon dspr2/; add_proto qw/void vp9_idct16x16_256_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct16x16_256_add sse2 neon dspr2/; add_proto qw/void vp9_idct16x16_10_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct16x16_10_add sse2 neon dspr2/; add_proto qw/void vp9_idct32x32_1024_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct32x32_1024_add sse2 neon dspr2/; add_proto qw/void vp9_idct32x32_34_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct32x32_34_add sse2 neon dspr2/; $vp9_idct32x32_34_add_neon=vp9_idct32x32_1024_add_neon; add_proto qw/void vp9_idct32x32_1_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_idct32x32_1_add sse2 neon dspr2/; add_proto qw/void vp9_iht4x4_16_add/, "const int16_t *input, uint8_t *dest, int dest_stride, int tx_type"; specialize qw/vp9_iht4x4_16_add sse2 neon dspr2/; add_proto qw/void vp9_iht8x8_64_add/, "const int16_t *input, uint8_t *dest, int dest_stride, int tx_type"; specialize qw/vp9_iht8x8_64_add sse2 neon dspr2/; add_proto qw/void vp9_iht16x16_256_add/, "const int16_t *input, uint8_t *output, int pitch, int tx_type"; specialize qw/vp9_iht16x16_256_add sse2 dspr2/; # dct and add add_proto qw/void vp9_iwht4x4_1_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_iwht4x4_1_add/;
351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420
add_proto qw/void vp9_iwht4x4_16_add/, "const int16_t *input, uint8_t *dest, int dest_stride"; specialize qw/vp9_iwht4x4_16_add/; # # Encoder functions below this point. # if (vpx_config("CONFIG_VP9_ENCODER") eq "yes") { # variance add_proto qw/unsigned int vp9_variance32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance32x16/, "$sse2_x86inc", "$avx2_x86inc"; add_proto qw/unsigned int vp9_variance16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance16x32/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance64x32/, "$sse2_x86inc", "$avx2_x86inc"; add_proto qw/unsigned int vp9_variance32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance32x64/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance32x32/, "$sse2_x86inc", "$avx2_x86inc"; add_proto qw/unsigned int vp9_variance64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance64x64/, "$sse2_x86inc", "$avx2_x86inc"; add_proto qw/unsigned int vp9_variance16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance16x16 mmx/, "$sse2_x86inc", "$avx2_x86inc"; add_proto qw/void vp9_get_sse_sum_16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum"; specialize qw/vp9_get_sse_sum_16x16 sse2/; $vp9_get_sse_sum_16x16_sse2=vp9_get16x16var_sse2; add_proto qw/unsigned int vp9_variance16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance16x8 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance8x16 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance8x8 mmx/, "$sse2_x86inc"; add_proto qw/void vp9_get_sse_sum_8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, int *sum"; specialize qw/vp9_get_sse_sum_8x8 sse2/; $vp9_get_sse_sum_8x8_sse2=vp9_get8x8var_sse2; add_proto qw/unsigned int vp9_variance8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance8x4/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance4x8/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_variance4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_variance4x4 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance64x64 avx2/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance64x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance64x64 avx2/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance32x64/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x64/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance32x64/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse";
421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490
specialize qw/vp9_sub_pixel_variance64x32/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance64x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance64x32/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance32x16/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance32x16/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance16x32/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance16x32/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance32x32 avx2/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance32x32/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance32x32 avx2/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance16x16/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance16x16/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance8x16/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance8x16/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance8x16/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance16x8/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance16x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance16x8/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance8x8/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance8x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance8x8/, "$sse2_x86inc", "$ssse3_x86inc"; # TODO(jingning): need to convert 8x4/4x8 functions into mmx/sse form add_proto qw/unsigned int vp9_sub_pixel_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance8x4/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance8x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance8x4/, "$sse2_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance4x8/, "$sse_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_avg_variance4x8/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance4x8/, "$sse_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sub_pixel_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse"; specialize qw/vp9_sub_pixel_variance4x4/, "$sse_x86inc", "$ssse3_x86inc"; #vp9_sub_pixel_variance4x4_sse2=vp9_sub_pixel_variance4x4_wmt add_proto qw/unsigned int vp9_sub_pixel_avg_variance4x4/, "const uint8_t *src_ptr, int source_stride, int xoffset, int yoffset, const uint8_t *ref_ptr, int ref_stride, unsigned int *sse, const uint8_t *second_pred"; specialize qw/vp9_sub_pixel_avg_variance4x4/, "$sse_x86inc", "$ssse3_x86inc"; add_proto qw/unsigned int vp9_sad64x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad64x64/, "$sse2_x86inc";
491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560
add_proto qw/unsigned int vp9_sad32x64/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad32x64/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad64x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad64x32/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad32x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad32x16/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad16x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad16x32/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad32x32/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad32x32/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad16x16 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad16x8 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad8x16 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad8x8 mmx/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad8x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad8x4/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad4x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad4x8/, "$sse_x86inc"; add_proto qw/unsigned int vp9_sad4x4/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int max_sad"; specialize qw/vp9_sad4x4 mmx/, "$sse_x86inc"; add_proto qw/unsigned int vp9_sad64x64_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad64x64_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad32x64_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad32x64_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad64x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad64x32_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad32x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad32x16_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad16x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad16x32_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad32x32_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad32x32_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad16x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad16x16_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad16x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad16x8_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad8x16_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad8x16_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad8x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad8x8_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad8x4_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad8x4_avg/, "$sse2_x86inc"; add_proto qw/unsigned int vp9_sad4x8_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad";
561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630
specialize qw/vp9_sad4x8_avg/, "$sse_x86inc"; add_proto qw/unsigned int vp9_sad4x4_avg/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, const uint8_t *second_pred, unsigned int max_sad"; specialize qw/vp9_sad4x4_avg/, "$sse_x86inc"; add_proto qw/void vp9_sad64x64x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad64x64x3/; add_proto qw/void vp9_sad32x32x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad32x32x3/; add_proto qw/void vp9_sad16x16x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad16x16x3 sse3 ssse3/; add_proto qw/void vp9_sad16x8x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad16x8x3 sse3 ssse3/; add_proto qw/void vp9_sad8x16x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad8x16x3 sse3/; add_proto qw/void vp9_sad8x8x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad8x8x3 sse3/; add_proto qw/void vp9_sad4x4x3/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad4x4x3 sse3/; add_proto qw/void vp9_sad64x64x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad64x64x8/; add_proto qw/void vp9_sad32x32x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad32x32x8/; add_proto qw/void vp9_sad16x16x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad16x16x8 sse4/; add_proto qw/void vp9_sad16x8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad16x8x8 sse4/; add_proto qw/void vp9_sad8x16x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad8x16x8 sse4/; add_proto qw/void vp9_sad8x8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad8x8x8 sse4/; add_proto qw/void vp9_sad8x4x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad8x4x8/; add_proto qw/void vp9_sad4x8x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad4x8x8/; add_proto qw/void vp9_sad4x4x8/, "const uint8_t *src_ptr, int src_stride, const uint8_t *ref_ptr, int ref_stride, uint32_t *sad_array"; specialize qw/vp9_sad4x4x8 sse4/; add_proto qw/void vp9_sad64x64x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad64x64x4d sse2 avx2/; add_proto qw/void vp9_sad32x64x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad32x64x4d sse2/; add_proto qw/void vp9_sad64x32x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad64x32x4d sse2/; add_proto qw/void vp9_sad32x16x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad32x16x4d sse2/; add_proto qw/void vp9_sad16x32x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad16x32x4d sse2/; add_proto qw/void vp9_sad32x32x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad32x32x4d sse2 avx2/;
631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700
add_proto qw/void vp9_sad16x16x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad16x16x4d sse2/; add_proto qw/void vp9_sad16x8x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad16x8x4d sse2/; add_proto qw/void vp9_sad8x16x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad8x16x4d sse2/; add_proto qw/void vp9_sad8x8x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad8x8x4d sse2/; # TODO(jingning): need to convert these 4x8/8x4 functions into sse2 form add_proto qw/void vp9_sad8x4x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad8x4x4d sse2/; add_proto qw/void vp9_sad4x8x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad4x8x4d sse/; add_proto qw/void vp9_sad4x4x4d/, "const uint8_t *src_ptr, int src_stride, const uint8_t* const ref_ptr[], int ref_stride, unsigned int *sad_array"; specialize qw/vp9_sad4x4x4d sse/; add_proto qw/unsigned int vp9_mse16x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse"; specialize qw/vp9_mse16x16 mmx/, "$sse2_x86inc", "$avx2_x86inc"; add_proto qw/unsigned int vp9_mse8x16/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse"; specialize qw/vp9_mse8x16/; add_proto qw/unsigned int vp9_mse16x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse"; specialize qw/vp9_mse16x8/; add_proto qw/unsigned int vp9_mse8x8/, "const uint8_t *src_ptr, int source_stride, const uint8_t *ref_ptr, int recon_stride, unsigned int *sse"; specialize qw/vp9_mse8x8/; add_proto qw/unsigned int vp9_get_mb_ss/, "const int16_t *"; specialize qw/vp9_get_mb_ss mmx sse2/; # ENCODEMB INVOKE add_proto qw/int64_t vp9_block_error/, "const int16_t *coeff, const int16_t *dqcoeff, intptr_t block_size, int64_t *ssz"; specialize qw/vp9_block_error/, "$sse2_x86inc"; add_proto qw/void vp9_subtract_block/, "int rows, int cols, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr, ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride"; specialize qw/vp9_subtract_block/, "$sse2_x86inc"; add_proto qw/void vp9_quantize_b/, "const int16_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, int16_t *qcoeff_ptr, int16_t *dqcoeff_ptr, const int16_t *dequant_ptr, int zbin_oq_value, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan"; specialize qw/vp9_quantize_b/, "$ssse3_x86_64"; add_proto qw/void vp9_quantize_b_32x32/, "const int16_t *coeff_ptr, intptr_t n_coeffs, int skip_block, const int16_t *zbin_ptr, const int16_t *round_ptr, const int16_t *quant_ptr, const int16_t *quant_shift_ptr, int16_t *qcoeff_ptr, int16_t *dqcoeff_ptr, const int16_t *dequant_ptr, int zbin_oq_value, uint16_t *eob_ptr, const int16_t *scan, const int16_t *iscan"; specialize qw/vp9_quantize_b_32x32/, "$ssse3_x86_64"; # # Structured Similarity (SSIM) # if (vpx_config("CONFIG_INTERNAL_STATS") eq "yes") { add_proto qw/void vp9_ssim_parms_8x8/, "uint8_t *s, int sp, uint8_t *r, int rp, unsigned long *sum_s, unsigned long *sum_r, unsigned long *sum_sq_s, unsigned long *sum_sq_r, unsigned long *sum_sxr"; specialize qw/vp9_ssim_parms_8x8/, "$sse2_x86_64"; add_proto qw/void vp9_ssim_parms_16x16/, "uint8_t *s, int sp, uint8_t *r, int rp, unsigned long *sum_s, unsigned long *sum_r, unsigned long *sum_sq_s, unsigned long *sum_sq_r, unsigned long *sum_sxr"; specialize qw/vp9_ssim_parms_16x16/, "$sse2_x86_64"; } # fdct functions add_proto qw/void vp9_fht4x4/, "const int16_t *input, int16_t *output, int stride, int tx_type"; specialize qw/vp9_fht4x4 sse2 avx2/; add_proto qw/void vp9_fht8x8/, "const int16_t *input, int16_t *output, int stride, int tx_type"; specialize qw/vp9_fht8x8 sse2 avx2/; add_proto qw/void vp9_fht16x16/, "const int16_t *input, int16_t *output, int stride, int tx_type";
701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746
specialize qw/vp9_fht16x16 sse2 avx2/; add_proto qw/void vp9_fwht4x4/, "const int16_t *input, int16_t *output, int stride"; specialize qw/vp9_fwht4x4/; add_proto qw/void vp9_fdct4x4/, "const int16_t *input, int16_t *output, int stride"; specialize qw/vp9_fdct4x4 sse2 avx2/; add_proto qw/void vp9_fdct8x8/, "const int16_t *input, int16_t *output, int stride"; specialize qw/vp9_fdct8x8 sse2 avx2/, "$ssse3_x86_64"; add_proto qw/void vp9_fdct16x16/, "const int16_t *input, int16_t *output, int stride"; specialize qw/vp9_fdct16x16 sse2 avx2/; add_proto qw/void vp9_fdct32x32/, "const int16_t *input, int16_t *output, int stride"; specialize qw/vp9_fdct32x32 sse2 avx2/; add_proto qw/void vp9_fdct32x32_rd/, "const int16_t *input, int16_t *output, int stride"; specialize qw/vp9_fdct32x32_rd sse2 avx2/; # # Motion search # add_proto qw/int vp9_full_search_sad/, "const struct macroblock *x, const struct mv *ref_mv, int sad_per_bit, int distance, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv, struct mv *best_mv"; specialize qw/vp9_full_search_sad sse3 sse4_1/; $vp9_full_search_sad_sse3=vp9_full_search_sadx3; $vp9_full_search_sad_sse4_1=vp9_full_search_sadx8; add_proto qw/int vp9_refining_search_sad/, "const struct macroblock *x, struct mv *ref_mv, int sad_per_bit, int distance, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv"; specialize qw/vp9_refining_search_sad sse3/; $vp9_refining_search_sad_sse3=vp9_refining_search_sadx4; add_proto qw/int vp9_diamond_search_sad/, "const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv"; specialize qw/vp9_diamond_search_sad sse3/; $vp9_diamond_search_sad_sse3=vp9_diamond_search_sadx4; add_proto qw/int vp9_full_range_search/, "const struct macroblock *x, const struct search_site_config *cfg, struct mv *ref_mv, struct mv *best_mv, int search_param, int sad_per_bit, int *num00, const struct vp9_variance_vtable *fn_ptr, const struct mv *center_mv"; specialize qw/vp9_full_range_search/; add_proto qw/void vp9_temporal_filter_apply/, "uint8_t *frame1, unsigned int stride, uint8_t *frame2, unsigned int block_size, int strength, int filter_weight, unsigned int *accumulator, uint16_t *count"; specialize qw/vp9_temporal_filter_apply sse2/; } # end encoder functions 1;