Removed the asm version of translate_bytes(). When both the C

version and the asm version are inlined, and everything is cached,
the asm version is 1.75 times slower than the C version on P5's.
On K6's, it is only 1.25 times slower.

1 files changed, 0 insertions, 29 deletions

diff --git a/sys/i386/isa/sound/audio.c b/sys/i386/isa/sound/audio.c
index 8c15d79..d92dd49 100644
--- a/sys/i386/isa/sound/audio.c
+++ b/sys/i386/isa/sound/audio.c
@@ -57,15 +57,6 @@ static char    *wr_dma_buf[MAX_AUDIO_DEV];
 static int      audio_format[MAX_AUDIO_DEV];
 static int      local_conversion[MAX_AUDIO_DEV];
 
-#if defined(NO_INLINE_ASM) || !defined(__i386__)
-static void
-translate_bytes(const u_char *table, u_char *buff, int n);
-
-#else
-extern __inline void
-translate_bytes(const void *table, void *buff, int n);
-#endif
-
 static int
 set_format(int dev, int fmt)
 {
@@ -148,7 +139,6 @@ audio_release(int dev, struct fileinfo * file)
 
 }
 
-#if defined(NO_INLINE_ASM) || !defined(__i386__)
 static void
 translate_bytes(const u_char *table, u_char *buff, int n)
 {
@@ -161,25 +151,6 @@ translate_bytes(const u_char *table, u_char *buff, int n)
 	buff[i] = table[buff[i]];
 }
 
-#else
-extern __inline void
-translate_bytes(const void *table, void *buff, int n)
-{
-    if (n > 0) {
-	__asm __volatile(
-		"cld\n"
-		"1:\tlodsb\n\t"
-		"xlatb\n\t"
-		"stosb\n\t"
-		"loop 1b\n\t"
-		: "=b" (table), "=c" (n), "=D" (buff), "=S" (buff)
-		:  "0" (table),  "1" (n),  "2" (buff),  "3" (buff)
-		: "ax", "memory");
-    }
-}
-
-#endif
-
 int
 audio_write(int dev, struct fileinfo * file, snd_rw_buf * buf, int count)
 {