diff alpha/asm.h @ 214:73df666cacc7 libavcodec

Alpha optimizations by Falk Hueffner <falk.hueffner@student.uni-tuebingen.de>
author nickols_k
date Sun, 20 Jan 2002 14:48:02 +0000
parents
children 718a22dc121f
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/alpha/asm.h	Sun Jan 20 14:48:02 2002 +0000
@@ -0,0 +1,141 @@
+/*
+ * Alpha optimized DSP utils
+ * Copyright (c) 2002 Falk Hueffner <falk@debian.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+#ifndef LIBAVCODEC_ALPHA_ASM_H
+#define LIBAVCODEC_ALPHA_ASM_H
+
+#include <stdint.h>
+
+#define AMASK_BWX (1 << 0)
+#define AMASK_FIX (1 << 1)
+#define AMASK_MVI (1 << 8)
+
+static inline uint64_t BYTE_VEC(uint64_t x)
+{
+    x |= x <<  8;
+    x |= x << 16;
+    x |= x << 32;
+    return x;
+}
+static inline uint64_t WORD_VEC(uint64_t x)
+{
+    x |= x << 16;
+    x |= x << 32;
+    return x;
+}
+
+static inline int32_t ldl(const void* p)
+{
+    return *(const int32_t*) p;
+}
+static inline uint64_t ldq(const void* p)
+{
+    return *(const uint64_t*) p;
+}
+/* FIXME ccc doesn't seem to get it? Use inline asm?  */
+static inline uint64_t ldq_u(const void* p)
+{
+    return *(const uint64_t*) ((uintptr_t) p & ~7ul);
+}
+static inline void stl(uint32_t l, void* p)
+{
+    *(uint32_t*) p = l;
+}
+static inline void stq(uint64_t l, void* p)
+{
+    *(uint64_t*) p = l;
+}
+
+#ifdef __GNUC__
+#define OPCODE1(name)						\
+static inline uint64_t name(uint64_t l)				\
+{								\
+    uint64_t r;							\
+    asm (#name " %1, %0" : "=r" (r) : "r" (l));			\
+    return r;							\
+}
+
+#define OPCODE2(name)						\
+static inline uint64_t name(uint64_t l1, uint64_t l2)		\
+{								\
+    uint64_t r;							\
+    asm (#name " %1, %2, %0" : "=r" (r) : "r" (l1), "rI" (l2));	\
+    return r;							\
+}
+
+/* We don't want gcc to move this around or combine it with another
+   rpcc, so mark it volatile.  */
+static inline uint64_t rpcc(void)
+{
+    uint64_t r;
+    asm volatile ("rpcc %0" : "=r" (r));
+    return r;
+}
+
+static inline uint64_t uldq(const void* v)
+{
+    struct foo {
+	unsigned long l;
+    } __attribute__((packed));
+
+    return ((const struct foo*) v)->l;
+}
+
+#elif defined(__DECC)		/* Compaq "ccc" compiler */
+
+#include <c_asm.h>
+#define OPCODE1(name)							\
+static inline uint64_t name(uint64_t l)					\
+{									\
+    return asm (#name " %a0, %v0", l);					\
+}
+
+#define OPCODE2(name)							\
+static inline uint64_t name(uint64_t l1, uint64_t l2)			\
+{									\
+    return asm (#name " %a0, %a1, %v0", l1, l2);			\
+}
+
+static inline uint64_t rpcc(void)
+{
+    return asm  ("rpcc %v0");
+}
+
+static inline uint64_t uldq(const void* v)
+{
+    return *(const __unaligned uint64_t *) v;
+}
+
+#endif
+
+OPCODE1(amask);
+OPCODE1(unpkbw);
+OPCODE1(pkwb);
+OPCODE2(extql);
+OPCODE2(extqh);
+OPCODE2(zap);
+OPCODE2(cmpbge);
+OPCODE2(minsw4);
+OPCODE2(minuw4);
+OPCODE2(minub8);
+OPCODE2(maxsw4);
+OPCODE2(maxuw4);
+OPCODE2(perr);
+
+#endif /* LIBAVCODEC_ALPHA_ASM_H */