[Mplayer-cvslog] CVS: main/libvo fastmemcpy.h,1.5,1.6

[Zoltan Ponekker]

Update of /cvsroot/mplayer/main/libvo
In directory usw-pr-cvs1:/tmp/cvs-serv1003/libvo

Modified Files:
	fastmemcpy.h 
Log Message:
add 3dnow support

Index: fastmemcpy.h
===================================================================
RCS file: /cvsroot/mplayer/main/libvo/fastmemcpy.h,v
retrieving revision 1.5
retrieving revision 1.6
diff -C2 -r1.5 -r1.6
*** fastmemcpy.h	2001/04/14 17:56:44	1.5
--- fastmemcpy.h	2001/04/16 20:55:27	1.6
***************
*** 1,2 ****
--- 1,6 ----
+ 
+ #ifndef __MPLAYER_MEMCPY
+ #define __MPLAYER_MEMCPY
+ 
  /*
   This part of code was taken by from Linux-2.4.3 and slightly modified
***************
*** 6,10 ****
  If you have questions please contact with me: Nick Kurshev: nickols_k at mail.ru.
  */
! #ifdef HAVE_MMX2
  /* for small memory blocks (<256 bytes) this version is faster */
  #define small_memcpy(to,from,n)\
--- 10,19 ----
  If you have questions please contact with me: Nick Kurshev: nickols_k at mail.ru.
  */
! 
! // 3dnow memcpy support from kernel 2.4.2
! //  by Pontscho/fresh!mindworkz
! 
! #if defined( HAVE_MMX2 ) || defined( HAVE_3DNOW )
! 
  /* for small memory blocks (<256 bytes) this version is faster */
  #define small_memcpy(to,from,n)\
***************
*** 28,31 ****
--- 37,47 ----
  	  
  	__asm__ __volatile__ (
+ #if defined( HAVE_3DNOW ) && !defined( HAVE_MMX2 )
+ 	        "prefetch (%0)\n"
+ 	        "prefetch 64(%0)\n"
+ 	        "prefetch 128(%0)\n"
+         	"prefetch 192(%0)\n"
+         	"prefetch 256(%0)\n"
+ #else
  		"prefetchnta (%0)\n"
  		"prefetchnta 64(%0)\n"
***************
*** 33,36 ****
--- 49,53 ----
  		"prefetchnta 192(%0)\n"
  		"prefetchnta 256(%0)\n"
+ #endif
  		: : "r" (from) );
          /*
***************
*** 46,50 ****
--- 63,71 ----
  	{
  		__asm__ __volatile__ (
+ #if defined( HAVE_3DNOW ) && !defined( HAVE_MMX2 )
+         	"prefetch 320(%0)\n"
+ #else
  		"prefetchnta 320(%0)\n"
+ #endif
  #ifdef HAVE_SSE /* Only P3 (may be Cyrix3) */
  		"movups (%0), %%xmm0\n"
***************
*** 57,60 ****
--- 78,99 ----
  		"movntps %%xmm1, 48(%1)\n"
  #else /* Only K7 (may be other) */
+ #if defined( HAVE_3DNOW ) && !defined( HAVE_MMX2 )
+         	"movq (%0), %%mm0\n"
+         	"movq 8(%0), %%mm1\n"
+         	"movq 16(%0), %%mm2\n"
+         	"movq 24(%0), %%mm3\n"
+         	"movq %%mm0, (%1)\n"
+         	"movq %%mm1, 8(%1)\n"
+         	"movq %%mm2, 16(%1)\n"
+         	"movq %%mm3, 24(%1)\n"
+         	"movq 32(%0), %%mm0\n"
+         	"movq 40(%0), %%mm1\n"
+         	"movq 48(%0), %%mm2\n"
+         	"movq 56(%0), %%mm3\n"
+         	"movq %%mm0, 32(%1)\n"
+         	"movq %%mm1, 40(%1)\n"
+         	"movq %%mm2, 48(%1)\n"
+         	"movq %%mm3, 56(%1)\n"
+ #else
  		"movq (%0), %%mm0\n"
  		"movq 8(%0), %%mm1\n"
***************
*** 74,90 ****
--- 113,150 ----
  		"movntq %%mm3, 56(%1)\n"
  #endif
+ #endif
  		:: "r" (from), "r" (to) : "memory");
  		from+=64;
  		to+=64;
  	}
+ #if defined( HAVE_3DNOW ) && !defined( HAVE_MMX2 )
+ 		__asm__ __volatile__ ("femms":::"memory");
+ #else
  		__asm__ __volatile__ ("emms":::"memory");
+ #endif
  	}
  	/*
  	 *	Now do the tail of the block
  	 */
+ #if 0
  	small_memcpy(to, from, len);
+ #else
+         __asm__ __volatile__ (
+                 "shrl $1,%%ecx\n"
+                 "jnc 1f\n"
+                 "movsb\n"
+                 "1:\n"
+                 "shrl $1,%%ecx\n"
+                 "jnc 2f\n"
+                 "movsw\n"
+                 "2:\n"
+                 "rep ; movsl\n"
+         	::"D" (to), "S" (from),"c" (len)
+         	: "memory");
+ #endif
  	return p;
  }
  #define memcpy(a,b,c) fast_memcpy(a,b,c)
+ 
  #endif
  


_______________________________________________
Mplayer-cvslog mailing list
Mplayer-cvslog at lists.sourceforge.net
http://lists.sourceforge.net/lists/listinfo/mplayer-cvslog