2 * Alpha optimized DSP utils
3 * Copyright (c) 2002 Falk Hueffner <falk@debian.org>
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2 of the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 #ifndef LIBAVCODEC_ALPHA_ASM_H
21 #define LIBAVCODEC_ALPHA_ASM_H
26 # define GNUC_PREREQ(maj, min) \
27 ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
29 # define GNUC_PREREQ(maj, min) 0
33 # define likely(x) __builtin_expect((x) != 0, 1)
34 # define unlikely(x) __builtin_expect((x) != 0, 0)
36 # define likely(x) (x)
37 # define unlikely(x) (x)
40 #define AMASK_BWX (1 << 0)
41 #define AMASK_FIX (1 << 1)
42 #define AMASK_CIX (1 << 2)
43 #define AMASK_MVI (1 << 8)
45 static inline uint64_t BYTE_VEC(uint64_t x)
52 static inline uint64_t WORD_VEC(uint64_t x)
59 #define sextw(x) ((int16_t) (x))
65 __typeof__(*(p)) __s[sizeof (uint64_t) / sizeof *(p)]; \
70 __typeof__(*(p)) __s[sizeof (int32_t) / sizeof *(p)]; \
76 __typeof__(*(p)) __s[sizeof (uint64_t) / sizeof *(p)]; \
77 } *) (p))->__l) = l; \
83 __typeof__(*(p)) __s[sizeof (int32_t) / sizeof *(p)]; \
84 } *) (p))->__l) = l; \
86 struct unaligned_long { uint64_t l; } __attribute__((packed));
87 #define ldq_u(p) (*(const uint64_t *) (((uint64_t) (p)) & ~7ul))
88 #define uldq(a) (((const struct unaligned_long *) (a))->l)
91 #define prefetch(p) __builtin_prefetch((p), 0, 1)
92 #define prefetch_en(p) __builtin_prefetch((p), 0, 0)
93 #define prefetch_m(p) __builtin_prefetch((p), 1, 1)
94 #define prefetch_men(p) __builtin_prefetch((p), 1, 0)
95 #define cmpbge __builtin_alpha_cmpbge
97 #define extql(a, b) __builtin_alpha_extql(a, (uint64_t) (b))
98 #define extwl(a, b) __builtin_alpha_extwl(a, (uint64_t) (b))
99 #define extqh(a, b) __builtin_alpha_extqh(a, (uint64_t) (b))
100 #define zap __builtin_alpha_zap
101 #define zapnot __builtin_alpha_zapnot
102 #define amask __builtin_alpha_amask
103 #define implver __builtin_alpha_implver
104 #define rpcc __builtin_alpha_rpcc
106 #define prefetch(p) asm volatile("ldl $31,%0" : : "m"(*(const char *) (p)) : "memory")
107 #define prefetch_en(p) asm volatile("ldq $31,%0" : : "m"(*(const char *) (p)) : "memory")
108 #define prefetch_m(p) asm volatile("lds $f31,%0" : : "m"(*(const char *) (p)) : "memory")
109 #define prefetch_men(p) asm volatile("ldt $f31,%0" : : "m"(*(const char *) (p)) : "memory")
110 #define cmpbge(a, b) ({ uint64_t __r; asm ("cmpbge %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; })
111 #define extql(a, b) ({ uint64_t __r; asm ("extql %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; })
112 #define extwl(a, b) ({ uint64_t __r; asm ("extwl %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; })
113 #define extqh(a, b) ({ uint64_t __r; asm ("extqh %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; })
114 #define zap(a, b) ({ uint64_t __r; asm ("zap %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; })
115 #define zapnot(a, b) ({ uint64_t __r; asm ("zapnot %r1,%2,%0" : "=r" (__r) : "rJ" (a), "rI" (b)); __r; })
116 #define amask(a) ({ uint64_t __r; asm ("amask %1,%0" : "=r" (__r) : "rI" (a)); __r; })
117 #define implver() ({ uint64_t __r; asm ("implver %0" : "=r" (__r)); __r; })
118 #define rpcc() ({ uint64_t __r; asm volatile ("rpcc %0" : "=r" (__r)); __r; })
120 #define wh64(p) asm volatile("wh64 (%0)" : : "r"(p) : "memory")
122 #if GNUC_PREREQ(3,3) && defined(__alpha_max__)
123 #define minub8 __builtin_alpha_minub8
124 #define minsb8 __builtin_alpha_minsb8
125 #define minuw4 __builtin_alpha_minuw4
126 #define minsw4 __builtin_alpha_minsw4
127 #define maxub8 __builtin_alpha_maxub8
128 #define maxsb8 __builtin_alpha_maxsb8
129 #define maxuw4 __builtin_alpha_maxuw4
130 #define maxsw4 __builtin_alpha_maxsw4
131 #define perr __builtin_alpha_perr
132 #define pklb __builtin_alpha_pklb
133 #define pkwb __builtin_alpha_pkwb
134 #define unpkbl __builtin_alpha_unpkbl
135 #define unpkbw __builtin_alpha_unpkbw
137 #define minub8(a, b) ({ uint64_t __r; asm (".arch ev6; minub8 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
138 #define minsb8(a, b) ({ uint64_t __r; asm (".arch ev6; minsb8 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
139 #define minuw4(a, b) ({ uint64_t __r; asm (".arch ev6; minuw4 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
140 #define minsw4(a, b) ({ uint64_t __r; asm (".arch ev6; minsw4 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
141 #define maxub8(a, b) ({ uint64_t __r; asm (".arch ev6; maxub8 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
142 #define maxsb8(a, b) ({ uint64_t __r; asm (".arch ev6; maxsb8 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
143 #define maxuw4(a, b) ({ uint64_t __r; asm (".arch ev6; maxuw4 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
144 #define maxsw4(a, b) ({ uint64_t __r; asm (".arch ev6; maxsw4 %r1,%2,%0" : "=r" (__r) : "%rJ" (a), "rI" (b)); __r; })
145 #define perr(a, b) ({ uint64_t __r; asm (".arch ev6; perr %r1,%r2,%0" : "=r" (__r) : "%rJ" (a), "rJ" (b)); __r; })
146 #define pklb(a) ({ uint64_t __r; asm (".arch ev6; pklb %r1,%0" : "=r" (__r) : "rJ" (a)); __r; })
147 #define pkwb(a) ({ uint64_t __r; asm (".arch ev6; pkwb %r1,%0" : "=r" (__r) : "rJ" (a)); __r; })
148 #define unpkbl(a) ({ uint64_t __r; asm (".arch ev6; unpkbl %r1,%0" : "=r" (__r) : "rJ" (a)); __r; })
149 #define unpkbw(a) ({ uint64_t __r; asm (".arch ev6; unpkbw %r1,%0" : "=r" (__r) : "rJ" (a)); __r; })
152 #elif defined(__DECC) /* Digital/Compaq/hp "ccc" compiler */
155 #define ldq(p) (*(const uint64_t *) (p))
156 #define ldl(p) (*(const int32_t *) (p))
157 #define stq(l, p) do { *(uint64_t *) (p) = (l); } while (0)
158 #define stl(l, p) do { *(int32_t *) (p) = (l); } while (0)
159 #define ldq_u(a) asm ("ldq_u %v0,0(%a0)", a)
160 #define uldq(a) (*(const __unaligned uint64_t *) (a))
161 #define cmpbge(a, b) asm ("cmpbge %a0,%a1,%v0", a, b)
162 #define extql(a, b) asm ("extql %a0,%a1,%v0", a, b)
163 #define extwl(a, b) asm ("extwl %a0,%a1,%v0", a, b)
164 #define extqh(a, b) asm ("extqh %a0,%a1,%v0", a, b)
165 #define zap(a, b) asm ("zap %a0,%a1,%v0", a, b)
166 #define zapnot(a, b) asm ("zapnot %a0,%a1,%v0", a, b)
167 #define amask(a) asm ("amask %a0,%v0", a)
168 #define implver() asm ("implver %v0")
169 #define rpcc() asm ("rpcc %v0")
170 #define minub8(a, b) asm ("minub8 %a0,%a1,%v0", a, b)
171 #define minsb8(a, b) asm ("minsb8 %a0,%a1,%v0", a, b)
172 #define minuw4(a, b) asm ("minuw4 %a0,%a1,%v0", a, b)
173 #define minsw4(a, b) asm ("minsw4 %a0,%a1,%v0", a, b)
174 #define maxub8(a, b) asm ("maxub8 %a0,%a1,%v0", a, b)
175 #define maxsb8(a, b) asm ("maxsb8 %a0,%a1,%v0", a, b)
176 #define maxuw4(a, b) asm ("maxuw4 %a0,%a1,%v0", a, b)
177 #define maxsw4(a, b) asm ("maxsw4 %a0,%a1,%v0", a, b)
178 #define perr(a, b) asm ("perr %a0,%a1,%v0", a, b)
179 #define pklb(a) asm ("pklb %a0,%v0", a)
180 #define pkwb(a) asm ("pkwb %a0,%v0", a)
181 #define unpkbl(a) asm ("unpkbl %a0,%v0", a)
182 #define unpkbw(a) asm ("unpkbw %a0,%v0", a)
183 #define wh64(a) asm ("wh64 %a0", a)
186 #error "Unknown compiler!"
189 #endif /* LIBAVCODEC_ALPHA_ASM_H */