dsputil_alpha_asm.S 6.93 KB
Newer Older
1 2 3 4
/*
 * Alpha optimized DSP utils
 * Copyright (c) 2002 Falk Hueffner <falk@debian.org>
 *
5
 * This file is part of Libav.
6
 *
7
 * Libav is free software; you can redistribute it and/or
8 9
 * modify it under the terms of the GNU Lesser General Public
 * 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 15
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * 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 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38
 */

/*
 * These functions are scheduled for pca56. They should work
 * reasonably on ev6, though.
 */

#include "regdef.h"

/* Some nicer register names.  */
#define ta t10
#define tb t11
#define tc t12
#define td AT
/* Danger: these overlap with the argument list and the return value */
#define te a5
#define tf a4
#define tg a3
#define th v0
39

40 41 42 43 44
        .set noat
        .set noreorder
        .arch pca56
        .text

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
/************************************************************************
 * void put_pixels_axp_asm(uint8_t *block, const uint8_t *pixels,
 *                         int line_size, int h)
 */
        .align 6
        .globl put_pixels_axp_asm
        .ent put_pixels_axp_asm
put_pixels_axp_asm:
        .frame sp, 0, ra
        .prologue 0

        and     a1, 7, t0
        beq     t0, $aligned

        .align 4
$unaligned:
        ldq_u   t0, 0(a1)
        ldq_u   t1, 8(a1)
        addq    a1, a2, a1
        nop

        ldq_u   t2, 0(a1)
        ldq_u   t3, 8(a1)
        addq    a1, a2, a1
        nop

71
        ldq_u   t4, 0(a1)
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
        ldq_u   t5, 8(a1)
        addq    a1, a2, a1
        nop

        ldq_u   t6, 0(a1)
        ldq_u   t7, 8(a1)
        extql   t0, a1, t0
        addq    a1, a2, a1

        extqh   t1, a1, t1
        addq    a0, a2, t8
        extql   t2, a1, t2
        addq    t8, a2, t9

        extqh   t3, a1, t3
        addq    t9, a2, ta
        extql   t4, a1, t4
        or      t0, t1, t0

        extqh   t5, a1, t5
        or      t2, t3, t2
        extql   t6, a1, t6
        or      t4, t5, t4

        extqh   t7, a1, t7
        or      t6, t7, t6
        stq     t0, 0(a0)
        stq     t2, 0(t8)

        stq     t4, 0(t9)
        subq    a3, 4, a3
        stq     t6, 0(ta)
        addq    ta, a2, a0

        bne     a3, $unaligned
        ret

        .align 4
$aligned:
        ldq     t0, 0(a1)
        addq    a1, a2, a1
        ldq     t1, 0(a1)
        addq    a1, a2, a1

        ldq     t2, 0(a1)
        addq    a1, a2, a1
        ldq     t3, 0(a1)

120 121 122 123
        addq    a0, a2, t4
        addq    a1, a2, a1
        addq    t4, a2, t5
        subq    a3, 4, a3
124

125 126 127 128
        stq     t0, 0(a0)
        addq    t5, a2, t6
        stq     t1, 0(t4)
        addq    t6, a2, a0
129

130 131
        stq     t2, 0(t5)
        stq     t3, 0(t6)
132

133
        bne     a3, $aligned
134 135 136
        ret
        .end put_pixels_axp_asm

137
/************************************************************************
Diego Biurrun's avatar
Diego Biurrun committed
138
 * void put_pixels_clamped_mvi_asm(const int16_t *block, uint8_t *pixels,
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
 *                                 int line_size)
 */
        .align 6
        .globl put_pixels_clamped_mvi_asm
        .ent put_pixels_clamped_mvi_asm
put_pixels_clamped_mvi_asm:
        .frame sp, 0, ra
        .prologue 0

        lda     t8, -1
        lda     t9, 8           # loop counter
        zap     t8, 0xaa, t8    # 00ff00ff00ff00ff

        .align 4
1:      ldq     t0,  0(a0)
        ldq     t1,  8(a0)
        ldq     t2, 16(a0)
        ldq     t3, 24(a0)

        maxsw4  t0, zero, t0
        subq    t9, 2, t9
        maxsw4  t1, zero, t1
        lda     a0, 32(a0)

        maxsw4  t2, zero, t2
        addq    a1, a2, ta
        maxsw4  t3, zero, t3
        minsw4  t0, t8, t0
167

168 169 170 171
        minsw4  t1, t8, t1
        minsw4  t2, t8, t2
        minsw4  t3, t8, t3
        pkwb    t0, t0
172

173 174 175 176
        pkwb    t1, t1
        pkwb    t2, t2
        pkwb    t3, t3
        stl     t0, 0(a1)
177

178 179 180 181 182 183 184 185 186 187
        stl     t1, 4(a1)
        addq    ta, a2, a1
        stl     t2, 0(ta)
        stl     t3, 4(ta)

        bne     t9, 1b
        ret
        .end put_pixels_clamped_mvi_asm

/************************************************************************
Diego Biurrun's avatar
Diego Biurrun committed
188
 * void add_pixels_clamped_mvi_asm(const int16_t *block, uint8_t *pixels,
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
 *                                 int line_size)
 */
        .align 6
        .globl add_pixels_clamped_mvi_asm
        .ent add_pixels_clamped_mvi_asm
add_pixels_clamped_mvi_asm:
        .frame sp, 0, ra
        .prologue 0

        lda     t1, -1
        lda     th, 8
        zap     t1, 0x33, tg
        nop

        srl     tg, 1, t0
        xor     tg, t0, tg      # 0x8000800080008000
        zap     t1, 0xaa, tf    # 0x00ff00ff00ff00ff

        .align 4
1:      ldl     t1, 0(a1)       # pix0 (try to hit cache line soon)
        ldl     t4, 4(a1)       # pix1
        addq    a1, a2, te      # pixels += line_size
        ldq     t0, 0(a0)       # shorts0

        ldl     t7, 0(te)       # pix2 (try to hit cache line soon)
        ldl     ta, 4(te)       # pix3
        ldq     t3, 8(a0)       # shorts1
        ldq     t6, 16(a0)      # shorts2

        ldq     t9, 24(a0)      # shorts3
        unpkbw  t1, t1          # 0 0 (quarter/op no.)
        and     t0, tg, t2      # 0 1
        unpkbw  t4, t4          # 1 0

        bic     t0, tg, t0      # 0 2
        unpkbw  t7, t7          # 2 0
        and     t3, tg, t5      # 1 1
226
        addq    t0, t1, t0      # 0 3
227 228 229 230 231

        xor     t0, t2, t0      # 0 4
        unpkbw  ta, ta          # 3 0
        and     t6, tg, t8      # 2 1
        maxsw4  t0, zero, t0    # 0 5
232

233 234 235 236
        bic     t3, tg, t3      # 1 2
        bic     t6, tg, t6      # 2 2
        minsw4  t0, tf, t0      # 0 6
        addq    t3, t4, t3      # 1 3
237

238 239 240 241 242 243 244 245 246 247 248 249
        pkwb    t0, t0          # 0 7
        xor     t3, t5, t3      # 1 4
        maxsw4  t3, zero, t3    # 1 5
        addq    t6, t7, t6      # 2 3

        xor     t6, t8, t6      # 2 4
        and     t9, tg, tb      # 3 1
        minsw4  t3, tf, t3      # 1 6
        bic     t9, tg, t9      # 3 2

        maxsw4  t6, zero, t6    # 2 5
        addq    t9, ta, t9      # 3 3
250
        stl     t0, 0(a1)       # 0 8
251 252 253 254 255 256
        minsw4  t6, tf, t6      # 2 6

        xor     t9, tb, t9      # 3 4
        maxsw4  t9, zero, t9    # 3 5
        lda     a0, 32(a0)      # block += 16;
        pkwb    t3, t3          # 1 7
257

258 259 260 261 262 263 264 265 266 267 268
        minsw4  t9, tf, t9      # 3 6
        subq    th, 2, th
        pkwb    t6, t6          # 2 7
        pkwb    t9, t9          # 3 7

        stl     t3, 4(a1)       # 1 8
        addq    te, a2, a1      # pixels += line_size
        stl     t6, 0(te)       # 2 8
        stl     t9, 4(te)       # 3 8

        bne     th, 1b
269
        ret
270
        .end add_pixels_clamped_mvi_asm