summaryrefslogtreecommitdiffstats
path: root/media/libvpx/vp8/common/x86/idctllm_mmx.asm
blob: 96fa2c60d0ff468555241213da42a215f5ece679 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
;
;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
;
;  Use of this source code is governed by a BSD-style license
;  that can be found in the LICENSE file in the root of the source
;  tree. An additional intellectual property rights grant can be found
;  in the file PATENTS.  All contributing project authors may
;  be found in the AUTHORS file in the root of the source tree.
;


%include "vpx_ports/x86_abi_support.asm"

; /****************************************************************************
; * Notes:
; *
; * This implementation makes use of 16 bit fixed point version of two multiply
; * constants:
; *        1.   sqrt(2) * cos (pi/8)
; *        2.   sqrt(2) * sin (pi/8)
; * Because the first constant is bigger than 1, to maintain the same 16 bit
; * fixed point precision as the second one, we use a trick of
; *        x * a = x + x*(a-1)
; * so
; *        x * sqrt(2) * cos (pi/8) = x + x * (sqrt(2) *cos(pi/8)-1).
; *
; * For the second constant, because of the 16bit version is 35468, which
; * is bigger than 32768, in signed 16 bit multiply, it becomes a negative
; * number.
; *        (x * (unsigned)35468 >> 16) = x * (signed)35468 >> 16 + x
; *
; **************************************************************************/


;void vp8_short_idct4x4llm_mmx(short *input, unsigned char *pred,
;int pitch, unsigned char *dest,int stride)
global sym(vp8_short_idct4x4llm_mmx) PRIVATE
sym(vp8_short_idct4x4llm_mmx):
    push        rbp
    mov         rbp, rsp
    SHADOW_ARGS_TO_STACK 5
    GET_GOT     rbx
    push        rsi
    push        rdi
    ; end prolog

    mov         rax,    arg(0)              ;input
    mov         rsi,    arg(1)              ;pred

    movq        mm0,    [rax   ]
    movq        mm1,    [rax+ 8]
    movq        mm2,    [rax+16]
    movq        mm3,    [rax+24]

%if 0
    pxor        mm7,    mm7
    movq        [rax],   mm7
    movq        [rax+8], mm7
    movq        [rax+16],mm7
    movq        [rax+24],mm7
%endif
    movsxd      rax,    dword ptr arg(2)    ;pitch
    mov         rdx,    arg(3)              ;dest
    movsxd      rdi,    dword ptr arg(4)    ;stride


    psubw       mm0,            mm2             ; b1= 0-2
    paddw       mm2,            mm2             ;

    movq        mm5,            mm1
    paddw       mm2,            mm0             ; a1 =0+2

    pmulhw      mm5,            [GLOBAL(x_s1sqr2)];
    paddw       mm5,            mm1             ; ip1 * sin(pi/8) * sqrt(2)

    movq        mm7,            mm3             ;
    pmulhw      mm7,            [GLOBAL(x_c1sqr2less1)];

    paddw       mm7,            mm3             ; ip3 * cos(pi/8) * sqrt(2)
    psubw       mm7,            mm5             ; c1

    movq        mm5,            mm1
    movq        mm4,            mm3

    pmulhw      mm5,            [GLOBAL(x_c1sqr2less1)]
    paddw       mm5,            mm1

    pmulhw      mm3,            [GLOBAL(x_s1sqr2)]
    paddw       mm3,            mm4

    paddw       mm3,            mm5             ; d1
    movq        mm6,            mm2             ; a1

    movq        mm4,            mm0             ; b1
    paddw       mm2,            mm3             ;0

    paddw       mm4,            mm7             ;1
    psubw       mm0,            mm7             ;2

    psubw       mm6,            mm3             ;3

    movq        mm1,            mm2             ; 03 02 01 00
    movq        mm3,            mm4             ; 23 22 21 20

    punpcklwd   mm1,            mm0             ; 11 01 10 00
    punpckhwd   mm2,            mm0             ; 13 03 12 02

    punpcklwd   mm3,            mm6             ; 31 21 30 20
    punpckhwd   mm4,            mm6             ; 33 23 32 22

    movq        mm0,            mm1             ; 11 01 10 00
    movq        mm5,            mm2             ; 13 03 12 02

    punpckldq   mm0,            mm3             ; 30 20 10 00
    punpckhdq   mm1,            mm3             ; 31 21 11 01

    punpckldq   mm2,            mm4             ; 32 22 12 02
    punpckhdq   mm5,            mm4             ; 33 23 13 03

    movq        mm3,            mm5             ; 33 23 13 03

    psubw       mm0,            mm2             ; b1= 0-2
    paddw       mm2,            mm2             ;

    movq        mm5,            mm1
    paddw       mm2,            mm0             ; a1 =0+2

    pmulhw      mm5,            [GLOBAL(x_s1sqr2)];
    paddw       mm5,            mm1             ; ip1 * sin(pi/8) * sqrt(2)

    movq        mm7,            mm3             ;
    pmulhw      mm7,            [GLOBAL(x_c1sqr2less1)];

    paddw       mm7,            mm3             ; ip3 * cos(pi/8) * sqrt(2)
    psubw       mm7,            mm5             ; c1

    movq        mm5,            mm1
    movq        mm4,            mm3

    pmulhw      mm5,            [GLOBAL(x_c1sqr2less1)]
    paddw       mm5,            mm1

    pmulhw      mm3,            [GLOBAL(x_s1sqr2)]
    paddw       mm3,            mm4

    paddw       mm3,            mm5             ; d1
    paddw       mm0,            [GLOBAL(fours)]

    paddw       mm2,            [GLOBAL(fours)]
    movq        mm6,            mm2             ; a1

    movq        mm4,            mm0             ; b1
    paddw       mm2,            mm3             ;0

    paddw       mm4,            mm7             ;1
    psubw       mm0,            mm7             ;2

    psubw       mm6,            mm3             ;3
    psraw       mm2,            3

    psraw       mm0,            3
    psraw       mm4,            3

    psraw       mm6,            3

    movq        mm1,            mm2             ; 03 02 01 00
    movq        mm3,            mm4             ; 23 22 21 20

    punpcklwd   mm1,            mm0             ; 11 01 10 00
    punpckhwd   mm2,            mm0             ; 13 03 12 02

    punpcklwd   mm3,            mm6             ; 31 21 30 20
    punpckhwd   mm4,            mm6             ; 33 23 32 22

    movq        mm0,            mm1             ; 11 01 10 00
    movq        mm5,            mm2             ; 13 03 12 02

    punpckldq   mm0,            mm3             ; 30 20 10 00
    punpckhdq   mm1,            mm3             ; 31 21 11 01

    punpckldq   mm2,            mm4             ; 32 22 12 02
    punpckhdq   mm5,            mm4             ; 33 23 13 03

    pxor        mm7,            mm7

    movd        mm4,            [rsi]
    punpcklbw   mm4,            mm7
    paddsw      mm0,            mm4
    packuswb    mm0,            mm7
    movd        [rdx],          mm0

    movd        mm4,            [rsi+rax]
    punpcklbw   mm4,            mm7
    paddsw      mm1,            mm4
    packuswb    mm1,            mm7
    movd        [rdx+rdi],      mm1

    movd        mm4,            [rsi+2*rax]
    punpcklbw   mm4,            mm7
    paddsw      mm2,            mm4
    packuswb    mm2,            mm7
    movd        [rdx+rdi*2],    mm2

    add         rdx,            rdi
    add         rsi,            rax

    movd        mm4,            [rsi+2*rax]
    punpcklbw   mm4,            mm7
    paddsw      mm5,            mm4
    packuswb    mm5,            mm7
    movd        [rdx+rdi*2],    mm5

    ; begin epilog
    pop rdi
    pop rsi
    RESTORE_GOT
    UNSHADOW_ARGS
    pop         rbp
    ret

;void vp8_dc_only_idct_add_mmx(
;short input_dc,
;unsigned char *pred_ptr,
;int pred_stride,
;unsigned char *dst_ptr,
;int stride)
global sym(vp8_dc_only_idct_add_mmx) PRIVATE
sym(vp8_dc_only_idct_add_mmx):
    push        rbp
    mov         rbp, rsp
    SHADOW_ARGS_TO_STACK 5
    GET_GOT     rbx
    ; end prolog

        movd        mm5,            arg(0) ;input_dc
        mov         rax,            arg(1) ;pred_ptr
        movsxd      rdx,            dword ptr arg(2) ;pred_stride

        pxor        mm0,            mm0

        paddw       mm5,            [GLOBAL(fours)]
        lea         rcx,            [rdx + rdx*2]

        psraw       mm5,            3

        punpcklwd   mm5,            mm5

        punpckldq   mm5,            mm5

        movd        mm1,            [rax]
        movd        mm2,            [rax+rdx]
        movd        mm3,            [rax+2*rdx]
        movd        mm4,            [rax+rcx]

        mov         rax,            arg(3) ;d -- destination
        movsxd      rdx,            dword ptr arg(4) ;dst_stride

        punpcklbw   mm1,            mm0
        paddsw      mm1,            mm5
        packuswb    mm1,            mm0              ; pack and unpack to saturate
        lea         rcx,            [rdx + rdx*2]

        punpcklbw   mm2,            mm0
        paddsw      mm2,            mm5
        packuswb    mm2,            mm0              ; pack and unpack to saturate

        punpcklbw   mm3,            mm0
        paddsw      mm3,            mm5
        packuswb    mm3,            mm0              ; pack and unpack to saturate

        punpcklbw   mm4,            mm0
        paddsw      mm4,            mm5
        packuswb    mm4,            mm0              ; pack and unpack to saturate

        movd        [rax],          mm1
        movd        [rax+rdx],      mm2
        movd        [rax+2*rdx],    mm3
        movd        [rax+rcx],      mm4

    ; begin epilog
    RESTORE_GOT
    UNSHADOW_ARGS
    pop         rbp
    ret

SECTION_RODATA
align 16
x_s1sqr2:
    times 4 dw 0x8A8C
align 16
x_c1sqr2less1:
    times 4 dw 0x4E7B
align 16
fours:
    times 4 dw 0x0004