--- interpolate8x8.c	2004/08/29 05:32:29	1.14
+++ interpolate8x8.c	2005/09/13 12:12:15	1.15
@@ -19,7 +19,7 @@
  *  along with this program ; if not, write to the Free Software
  *  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307 USA
  *
- * $Id: interpolate8x8.c,v 1.14 2004/08/29 05:32:29 syskin Exp $
+ * $Id: interpolate8x8.c,v 1.15 2005/09/13 12:12:15 suxen_drol Exp $
  *
  ****************************************************************************/
 
@@ -32,12 +32,15 @@
 INTERPOLATE8X8_PTR interpolate8x8_halfpel_v;
 INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv;
 
+INTERPOLATE8X8_PTR interpolate8x4_halfpel_h;
+INTERPOLATE8X8_PTR interpolate8x4_halfpel_v;
+INTERPOLATE8X8_PTR interpolate8x4_halfpel_hv;
+
 INTERPOLATE8X8_PTR interpolate8x8_halfpel_add;
 INTERPOLATE8X8_PTR interpolate8x8_halfpel_h_add;
 INTERPOLATE8X8_PTR interpolate8x8_halfpel_v_add;
 INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv_add;
 
-
 INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2;
 INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4;
 
@@ -139,6 +142,41 @@
 	}
 }
 
+/* dst = interpolate(src) */
+
+void
+interpolate8x4_halfpel_h_c(uint8_t * const dst,
+						   const uint8_t * const src,
+						   const uint32_t stride,
+						   const uint32_t rounding)
+{
+	uintptr_t j;
+
+	if (rounding) {
+		for (j = 0; j < 4*stride; j+=stride) {
+				dst[j + 0] = (uint8_t)((src[j + 0] + src[j + 1] )>>1);
+				dst[j + 1] = (uint8_t)((src[j + 1] + src[j + 2] )>>1);
+				dst[j + 2] = (uint8_t)((src[j + 2] + src[j + 3] )>>1);
+				dst[j + 3] = (uint8_t)((src[j + 3] + src[j + 4] )>>1);
+				dst[j + 4] = (uint8_t)((src[j + 4] + src[j + 5] )>>1);
+				dst[j + 5] = (uint8_t)((src[j + 5] + src[j + 6] )>>1);
+				dst[j + 6] = (uint8_t)((src[j + 6] + src[j + 7] )>>1);
+				dst[j + 7] = (uint8_t)((src[j + 7] + src[j + 8] )>>1);
+		}
+	} else {
+		for (j = 0; j < 4*stride; j+=stride) {
+				dst[j + 0] = (uint8_t)((src[j + 0] + src[j + 1] + 1)>>1);
+				dst[j + 1] = (uint8_t)((src[j + 1] + src[j + 2] + 1)>>1);
+				dst[j + 2] = (uint8_t)((src[j + 2] + src[j + 3] + 1)>>1);
+				dst[j + 3] = (uint8_t)((src[j + 3] + src[j + 4] + 1)>>1);
+				dst[j + 4] = (uint8_t)((src[j + 4] + src[j + 5] + 1)>>1);
+				dst[j + 5] = (uint8_t)((src[j + 5] + src[j + 6] + 1)>>1);
+				dst[j + 6] = (uint8_t)((src[j + 6] + src[j + 7] + 1)>>1);
+				dst[j + 7] = (uint8_t)((src[j + 7] + src[j + 8] + 1)>>1);
+		}
+	}
+}
+
 /* dst = (dst + interpolate(src)/2 */
 
 void
@@ -210,6 +248,42 @@
 	}
 }
 
+/* dst = interpolate(src) */
+
+void
+interpolate8x4_halfpel_v_c(uint8_t * const dst,
+						   const uint8_t * const src,
+						   const uint32_t stride,
+						   const uint32_t rounding)
+{
+	uintptr_t j;
+
+
+	if (rounding) {
+		for (j = 0; j < 4*stride; j+=stride) {
+				dst[j + 0] = (uint8_t)((src[j + 0] + src[j + stride + 0] )>>1);
+				dst[j + 1] = (uint8_t)((src[j + 1] + src[j + stride + 1] )>>1);
+				dst[j + 2] = (uint8_t)((src[j + 2] + src[j + stride + 2] )>>1);
+				dst[j + 3] = (uint8_t)((src[j + 3] + src[j + stride + 3] )>>1);
+				dst[j + 4] = (uint8_t)((src[j + 4] + src[j + stride + 4] )>>1);
+				dst[j + 5] = (uint8_t)((src[j + 5] + src[j + stride + 5] )>>1);
+				dst[j + 6] = (uint8_t)((src[j + 6] + src[j + stride + 6] )>>1);
+				dst[j + 7] = (uint8_t)((src[j + 7] + src[j + stride + 7] )>>1);
+		}
+	} else {
+		for (j = 0; j < 4*stride; j+=stride) {
+				dst[j + 0] = (uint8_t)((src[j + 0] + src[j + stride + 0] + 1)>>1);
+				dst[j + 1] = (uint8_t)((src[j + 1] + src[j + stride + 1] + 1)>>1);
+				dst[j + 2] = (uint8_t)((src[j + 2] + src[j + stride + 2] + 1)>>1);
+				dst[j + 3] = (uint8_t)((src[j + 3] + src[j + stride + 3] + 1)>>1);
+				dst[j + 4] = (uint8_t)((src[j + 4] + src[j + stride + 4] + 1)>>1);
+				dst[j + 5] = (uint8_t)((src[j + 5] + src[j + stride + 5] + 1)>>1);
+				dst[j + 6] = (uint8_t)((src[j + 6] + src[j + stride + 6] + 1)>>1);
+				dst[j + 7] = (uint8_t)((src[j + 7] + src[j + stride + 7] + 1)>>1);
+		}
+	}
+}
+
 /* dst = (dst + interpolate(src))/2 */
 
 void
@@ -281,6 +355,41 @@
 	}
 }
 
+/* dst = interpolate(src) */
+
+void
+interpolate8x4_halfpel_hv_c(uint8_t * const dst,
+							const uint8_t * const src,
+							const uint32_t stride,
+							const uint32_t rounding)
+{
+	uintptr_t j;
+
+	if (rounding) {
+		for (j = 0; j < 4*stride; j+=stride) {
+				dst[j + 0] = (uint8_t)((src[j+0] + src[j+1] + src[j+stride+0] + src[j+stride+1] +1)>>2);
+				dst[j + 1] = (uint8_t)((src[j+1] + src[j+2] + src[j+stride+1] + src[j+stride+2] +1)>>2);
+				dst[j + 2] = (uint8_t)((src[j+2] + src[j+3] + src[j+stride+2] + src[j+stride+3] +1)>>2);
+				dst[j + 3] = (uint8_t)((src[j+3] + src[j+4] + src[j+stride+3] + src[j+stride+4] +1)>>2);
+				dst[j + 4] = (uint8_t)((src[j+4] + src[j+5] + src[j+stride+4] + src[j+stride+5] +1)>>2);
+				dst[j + 5] = (uint8_t)((src[j+5] + src[j+6] + src[j+stride+5] + src[j+stride+6] +1)>>2);
+				dst[j + 6] = (uint8_t)((src[j+6] + src[j+7] + src[j+stride+6] + src[j+stride+7] +1)>>2);
+				dst[j + 7] = (uint8_t)((src[j+7] + src[j+8] + src[j+stride+7] + src[j+stride+8] +1)>>2);
+		}
+	} else {
+		for (j = 0; j < 4*stride; j+=stride) {
+				dst[j + 0] = (uint8_t)((src[j+0] + src[j+1] + src[j+stride+0] + src[j+stride+1] +2)>>2);
+				dst[j + 1] = (uint8_t)((src[j+1] + src[j+2] + src[j+stride+1] + src[j+stride+2] +2)>>2);
+				dst[j + 2] = (uint8_t)((src[j+2] + src[j+3] + src[j+stride+2] + src[j+stride+3] +2)>>2);
+				dst[j + 3] = (uint8_t)((src[j+3] + src[j+4] + src[j+stride+3] + src[j+stride+4] +2)>>2);
+				dst[j + 4] = (uint8_t)((src[j+4] + src[j+5] + src[j+stride+4] + src[j+stride+5] +2)>>2);
+				dst[j + 5] = (uint8_t)((src[j+5] + src[j+6] + src[j+stride+5] + src[j+stride+6] +2)>>2);
+				dst[j + 6] = (uint8_t)((src[j+6] + src[j+7] + src[j+stride+6] + src[j+stride+7] +2)>>2);
+				dst[j + 7] = (uint8_t)((src[j+7] + src[j+8] + src[j+stride+7] + src[j+stride+8] +2)>>2);
+		}
+	}
+}
+
 /* dst = (interpolate(src) + dst)/2 */
 
 void