use multiply instead of divides for DC prediction on X86

Originally committed as revision 286 to svn://svn.ffmpeg.org/ffmpeg/trunk

diff --git a/libavcodec/msmpeg4.c b/libavcodec/msmpeg4.c
index b43db88..85436e9 100644 (file)
--- a/libavcodec/msmpeg4.c
+++ b/libavcodec/msmpeg4.c
@@ -50,6 +50,8 @@ static int msmpeg4_decode_dc(MpegEncContext * s, int n, int *dir_ptr);
 static int msmpeg4_decode_motion(MpegEncContext * s, 
                                  int *mx_ptr, int *my_ptr);
 
+extern UINT32 inverse[256];
+
 #ifdef DEBUG
 int intra_count = 0;
 int frame_count = 0;
@@ -438,26 +440,22 @@ static int msmpeg4_pred_dc(MpegEncContext * s, int n,
        fact they decided to store the quantized DC (which would lead
        to problems if Q could vary !) */
 #if defined ARCH_X86 && !defined PIC
-    /* using 16bit divisions as they are large enough and 2x as fast */
     asm volatile(
         "movl %3, %%eax                \n\t"
        "shrl $1, %%eax         \n\t"
        "addl %%eax, %2         \n\t"
        "addl %%eax, %1         \n\t"
        "addl %0, %%eax         \n\t"
-       "xorl %%edx, %%edx      \n\t"
-        "divw %w3              \n\t"
-       "movzwl %%ax, %0        \n\t"
+       "mull %4                \n\t"
+       "movl %%edx, %0         \n\t"
        "movl %1, %%eax         \n\t"
-       "xorl %%edx, %%edx      \n\t"
-        "divw %w3              \n\t"
-       "movzwl %%ax, %1        \n\t"
+       "mull %4                \n\t"
+       "movl %%edx, %1         \n\t"
        "movl %2, %%eax         \n\t"
-       "xorl %%edx, %%edx      \n\t"
-        "divw %w3              \n\t"
-       "movzwl %%ax, %2        \n\t"
-        : "+r" (a), "+r" (b), "+r" (c)
-       : "r" (scale)
+       "mull %4                \n\t"
+       "movl %%edx, %2         \n\t"
+       : "+r" (a), "+r" (b), "+r" (c)
+       : "g" (scale), "r" (inverse[scale])
        : "%eax", "%edx"
     );
 #else