1/*
2 * Alpha optimized DSP utils
3 * Copyright (c) 2002 Falk Hueffner <falk@debian.org>
4 *
5 * This file is part of Libav.
6 *
7 * Libav is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * Libav is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22/*
23 * These functions are scheduled for pca56. They should work
24 * reasonably on ev6, though.
25 */
26
27#include "regdef.h"
28
29/* Some nicer register names.  */
30#define ta t10
31#define tb t11
32#define tc t12
33#define td AT
34/* Danger: these overlap with the argument list and the return value */
35#define te a5
36#define tf a4
37#define tg a3
38#define th v0
39
40        .set noat
41        .set noreorder
42        .arch pca56
43        .text
44
45/************************************************************************
46 * void put_pixels_axp_asm(uint8_t *block, const uint8_t *pixels,
47 *                         int line_size, int h)
48 */
49        .align 6
50        .globl put_pixels_axp_asm
51        .ent put_pixels_axp_asm
52put_pixels_axp_asm:
53        .frame sp, 0, ra
54        .prologue 0
55
56        and     a1, 7, t0
57        beq     t0, $aligned
58
59        .align 4
60$unaligned:
61        ldq_u   t0, 0(a1)
62        ldq_u   t1, 8(a1)
63        addq    a1, a2, a1
64        nop
65
66        ldq_u   t2, 0(a1)
67        ldq_u   t3, 8(a1)
68        addq    a1, a2, a1
69        nop
70
71        ldq_u   t4, 0(a1)
72        ldq_u   t5, 8(a1)
73        addq    a1, a2, a1
74        nop
75
76        ldq_u   t6, 0(a1)
77        ldq_u   t7, 8(a1)
78        extql   t0, a1, t0
79        addq    a1, a2, a1
80
81        extqh   t1, a1, t1
82        addq    a0, a2, t8
83        extql   t2, a1, t2
84        addq    t8, a2, t9
85
86        extqh   t3, a1, t3
87        addq    t9, a2, ta
88        extql   t4, a1, t4
89        or      t0, t1, t0
90
91        extqh   t5, a1, t5
92        or      t2, t3, t2
93        extql   t6, a1, t6
94        or      t4, t5, t4
95
96        extqh   t7, a1, t7
97        or      t6, t7, t6
98        stq     t0, 0(a0)
99        stq     t2, 0(t8)
100
101        stq     t4, 0(t9)
102        subq    a3, 4, a3
103        stq     t6, 0(ta)
104        addq    ta, a2, a0
105
106        bne     a3, $unaligned
107        ret
108
109        .align 4
110$aligned:
111        ldq     t0, 0(a1)
112        addq    a1, a2, a1
113        ldq     t1, 0(a1)
114        addq    a1, a2, a1
115
116        ldq     t2, 0(a1)
117        addq    a1, a2, a1
118        ldq     t3, 0(a1)
119
120        addq    a0, a2, t4
121        addq    a1, a2, a1
122        addq    t4, a2, t5
123        subq    a3, 4, a3
124
125        stq     t0, 0(a0)
126        addq    t5, a2, t6
127        stq     t1, 0(t4)
128        addq    t6, a2, a0
129
130        stq     t2, 0(t5)
131        stq     t3, 0(t6)
132
133        bne     a3, $aligned
134        ret
135        .end put_pixels_axp_asm
136
137/************************************************************************
138 * void put_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels,
139 *                                 int line_size)
140 */
141        .align 6
142        .globl put_pixels_clamped_mvi_asm
143        .ent put_pixels_clamped_mvi_asm
144put_pixels_clamped_mvi_asm:
145        .frame sp, 0, ra
146        .prologue 0
147
148        lda     t8, -1
149        lda     t9, 8           # loop counter
150        zap     t8, 0xaa, t8    # 00ff00ff00ff00ff
151
152        .align 4
1531:      ldq     t0,  0(a0)
154        ldq     t1,  8(a0)
155        ldq     t2, 16(a0)
156        ldq     t3, 24(a0)
157
158        maxsw4  t0, zero, t0
159        subq    t9, 2, t9
160        maxsw4  t1, zero, t1
161        lda     a0, 32(a0)
162
163        maxsw4  t2, zero, t2
164        addq    a1, a2, ta
165        maxsw4  t3, zero, t3
166        minsw4  t0, t8, t0
167
168        minsw4  t1, t8, t1
169        minsw4  t2, t8, t2
170        minsw4  t3, t8, t3
171        pkwb    t0, t0
172
173        pkwb    t1, t1
174        pkwb    t2, t2
175        pkwb    t3, t3
176        stl     t0, 0(a1)
177
178        stl     t1, 4(a1)
179        addq    ta, a2, a1
180        stl     t2, 0(ta)
181        stl     t3, 4(ta)
182
183        bne     t9, 1b
184        ret
185        .end put_pixels_clamped_mvi_asm
186
187/************************************************************************
188 * void add_pixels_clamped_mvi_asm(const DCTELEM *block, uint8_t *pixels,
189 *                                 int line_size)
190 */
191        .align 6
192        .globl add_pixels_clamped_mvi_asm
193        .ent add_pixels_clamped_mvi_asm
194add_pixels_clamped_mvi_asm:
195        .frame sp, 0, ra
196        .prologue 0
197
198        lda     t1, -1
199        lda     th, 8
200        zap     t1, 0x33, tg
201        nop
202
203        srl     tg, 1, t0
204        xor     tg, t0, tg      # 0x8000800080008000
205        zap     t1, 0xaa, tf    # 0x00ff00ff00ff00ff
206
207        .align 4
2081:      ldl     t1, 0(a1)       # pix0 (try to hit cache line soon)
209        ldl     t4, 4(a1)       # pix1
210        addq    a1, a2, te      # pixels += line_size
211        ldq     t0, 0(a0)       # shorts0
212
213        ldl     t7, 0(te)       # pix2 (try to hit cache line soon)
214        ldl     ta, 4(te)       # pix3
215        ldq     t3, 8(a0)       # shorts1
216        ldq     t6, 16(a0)      # shorts2
217
218        ldq     t9, 24(a0)      # shorts3
219        unpkbw  t1, t1          # 0 0 (quarter/op no.)
220        and     t0, tg, t2      # 0 1
221        unpkbw  t4, t4          # 1 0
222
223        bic     t0, tg, t0      # 0 2
224        unpkbw  t7, t7          # 2 0
225        and     t3, tg, t5      # 1 1
226        addq    t0, t1, t0      # 0 3
227
228        xor     t0, t2, t0      # 0 4
229        unpkbw  ta, ta          # 3 0
230        and     t6, tg, t8      # 2 1
231        maxsw4  t0, zero, t0    # 0 5
232
233        bic     t3, tg, t3      # 1 2
234        bic     t6, tg, t6      # 2 2
235        minsw4  t0, tf, t0      # 0 6
236        addq    t3, t4, t3      # 1 3
237
238        pkwb    t0, t0          # 0 7
239        xor     t3, t5, t3      # 1 4
240        maxsw4  t3, zero, t3    # 1 5
241        addq    t6, t7, t6      # 2 3
242
243        xor     t6, t8, t6      # 2 4
244        and     t9, tg, tb      # 3 1
245        minsw4  t3, tf, t3      # 1 6
246        bic     t9, tg, t9      # 3 2
247
248        maxsw4  t6, zero, t6    # 2 5
249        addq    t9, ta, t9      # 3 3
250        stl     t0, 0(a1)       # 0 8
251        minsw4  t6, tf, t6      # 2 6
252
253        xor     t9, tb, t9      # 3 4
254        maxsw4  t9, zero, t9    # 3 5
255        lda     a0, 32(a0)      # block += 16;
256        pkwb    t3, t3          # 1 7
257
258        minsw4  t9, tf, t9      # 3 6
259        subq    th, 2, th
260        pkwb    t6, t6          # 2 7
261        pkwb    t9, t9          # 3 7
262
263        stl     t3, 4(a1)       # 1 8
264        addq    te, a2, a1      # pixels += line_size
265        stl     t6, 0(te)       # 2 8
266        stl     t9, 4(te)       # 3 8
267
268        bne     th, 1b
269        ret
270        .end add_pixels_clamped_mvi_asm
271