[SCM] libav/experimental: add FASTDIV macro for ARM. Reported speed-up on MPEG-4 decoding: 1.8% Patch by Siarhei Siamashka %siarhei P siamashka A gmail P com% Original thread: Date: Sep 28, 2006 2:26 AM Subject: [Ffmpeg-devel] [PATCH] ARM implementation of FASTDIV

[siretart at users.alioth.debian.org]

The following commit has been merged in the experimental branch:
commit eeebe6ad1fe61fc0be2e63658abb6441ab38c916
Author: Siarhei Siamashka <siarhei.siamashka at gmail.com>
Date:   Thu Sep 28 08:04:16 2006 +0000

    add FASTDIV macro for ARM. Reported speed-up on MPEG-4 decoding: 1.8%
    Patch by Siarhei Siamashka %siarhei P siamashka A gmail P com%
    Original thread:
    Date: Sep 28, 2006 2:26 AM
    Subject: [Ffmpeg-devel] [PATCH] ARM implementation of FASTDIV
    
    Originally committed as revision 6366 to svn://svn.ffmpeg.org/ffmpeg/trunk

diff --git a/libavutil/internal.h b/libavutil/internal.h
index 83f776c..96a6d29 100644
--- a/libavutil/internal.h
+++ b/libavutil/internal.h
@@ -114,6 +114,17 @@ extern const uint32_t inverse[256];
             );\
         ret;\
     })
+#elif defined(ARCH_ARMV4L)
+#    define FASTDIV(a,b) \
+    ({\
+        int ret,dmy;\
+        asm volatile(\
+            "umull %1, %0, %2, %3"\
+            :"=&r"(ret),"=&r"(dmy)\
+            :"r"(a),"r"(inverse[b])\
+            );\
+        ret;\
+    })
 #elif defined(CONFIG_FASTDIV)
 #    define FASTDIV(a,b)   ((uint32_t)((((uint64_t)a)*inverse[b])>>32))
 #else

-- 
Libav/FFmpeg packaging